{
  "best_global_step": 600,
  "best_metric": 0.2460305392742157,
  "best_model_checkpoint": "saves/lora/gemma-3-1b-it/train_cb_1745950310/checkpoint-600",
  "epoch": 701.7610619469026,
  "eval_steps": 200,
  "global_step": 40000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08849557522123894,
      "grad_norm": 33.863826751708984,
      "learning_rate": 4.999999876629946e-05,
      "loss": 2.6928,
      "num_input_tokens_seen": 3008,
      "step": 5
    },
    {
      "epoch": 0.17699115044247787,
      "grad_norm": 17.438322067260742,
      "learning_rate": 4.999999375439123e-05,
      "loss": 1.1,
      "num_input_tokens_seen": 6128,
      "step": 10
    },
    {
      "epoch": 0.26548672566371684,
      "grad_norm": 15.865092277526855,
      "learning_rate": 4.9999984887169785e-05,
      "loss": 0.5093,
      "num_input_tokens_seen": 9056,
      "step": 15
    },
    {
      "epoch": 0.35398230088495575,
      "grad_norm": 11.007576942443848,
      "learning_rate": 4.9999972164636506e-05,
      "loss": 0.1975,
      "num_input_tokens_seen": 12656,
      "step": 20
    },
    {
      "epoch": 0.4424778761061947,
      "grad_norm": 16.244543075561523,
      "learning_rate": 4.999995558679334e-05,
      "loss": 0.5269,
      "num_input_tokens_seen": 15808,
      "step": 25
    },
    {
      "epoch": 0.5309734513274337,
      "grad_norm": 6.314994812011719,
      "learning_rate": 4.999993515364287e-05,
      "loss": 0.3717,
      "num_input_tokens_seen": 18304,
      "step": 30
    },
    {
      "epoch": 0.6194690265486725,
      "grad_norm": 14.20385456085205,
      "learning_rate": 4.999991086518822e-05,
      "loss": 0.1708,
      "num_input_tokens_seen": 21712,
      "step": 35
    },
    {
      "epoch": 0.7079646017699115,
      "grad_norm": 5.261157035827637,
      "learning_rate": 4.999988272143315e-05,
      "loss": 0.1173,
      "num_input_tokens_seen": 24576,
      "step": 40
    },
    {
      "epoch": 0.7964601769911505,
      "grad_norm": 9.52596664428711,
      "learning_rate": 4.999985072238199e-05,
      "loss": 0.0917,
      "num_input_tokens_seen": 27088,
      "step": 45
    },
    {
      "epoch": 0.8849557522123894,
      "grad_norm": 7.049123287200928,
      "learning_rate": 4.999981486803969e-05,
      "loss": 0.2997,
      "num_input_tokens_seen": 29648,
      "step": 50
    },
    {
      "epoch": 0.9734513274336283,
      "grad_norm": 0.6120202541351318,
      "learning_rate": 4.999977515841176e-05,
      "loss": 0.1974,
      "num_input_tokens_seen": 32032,
      "step": 55
    },
    {
      "epoch": 1.0530973451327434,
      "grad_norm": 12.33362865447998,
      "learning_rate": 4.9999731593504344e-05,
      "loss": 0.0836,
      "num_input_tokens_seen": 34624,
      "step": 60
    },
    {
      "epoch": 1.1415929203539823,
      "grad_norm": 9.521320343017578,
      "learning_rate": 4.999968417332415e-05,
      "loss": 0.0943,
      "num_input_tokens_seen": 37600,
      "step": 65
    },
    {
      "epoch": 1.2300884955752212,
      "grad_norm": 6.166086673736572,
      "learning_rate": 4.999963289787848e-05,
      "loss": 0.2222,
      "num_input_tokens_seen": 40256,
      "step": 70
    },
    {
      "epoch": 1.3185840707964602,
      "grad_norm": 10.627964973449707,
      "learning_rate": 4.999957776717526e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 43440,
      "step": 75
    },
    {
      "epoch": 1.407079646017699,
      "grad_norm": 16.413455963134766,
      "learning_rate": 4.9999518781222984e-05,
      "loss": 0.1155,
      "num_input_tokens_seen": 46192,
      "step": 80
    },
    {
      "epoch": 1.495575221238938,
      "grad_norm": 10.394840240478516,
      "learning_rate": 4.9999455940030746e-05,
      "loss": 0.1563,
      "num_input_tokens_seen": 49360,
      "step": 85
    },
    {
      "epoch": 1.584070796460177,
      "grad_norm": 12.060955047607422,
      "learning_rate": 4.999938924360824e-05,
      "loss": 0.4079,
      "num_input_tokens_seen": 51840,
      "step": 90
    },
    {
      "epoch": 1.672566371681416,
      "grad_norm": 9.565608024597168,
      "learning_rate": 4.999931869196575e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 54960,
      "step": 95
    },
    {
      "epoch": 1.7610619469026547,
      "grad_norm": 0.12458274513483047,
      "learning_rate": 4.999924428511416e-05,
      "loss": 0.1414,
      "num_input_tokens_seen": 57776,
      "step": 100
    },
    {
      "epoch": 1.8495575221238938,
      "grad_norm": 10.379055976867676,
      "learning_rate": 4.999916602306494e-05,
      "loss": 0.1734,
      "num_input_tokens_seen": 60624,
      "step": 105
    },
    {
      "epoch": 1.9380530973451329,
      "grad_norm": 0.38981056213378906,
      "learning_rate": 4.999908390583016e-05,
      "loss": 0.0351,
      "num_input_tokens_seen": 63424,
      "step": 110
    },
    {
      "epoch": 2.017699115044248,
      "grad_norm": 8.939726829528809,
      "learning_rate": 4.999899793342247e-05,
      "loss": 0.1362,
      "num_input_tokens_seen": 65544,
      "step": 115
    },
    {
      "epoch": 2.106194690265487,
      "grad_norm": 1.1709493398666382,
      "learning_rate": 4.999890810585516e-05,
      "loss": 0.0021,
      "num_input_tokens_seen": 68552,
      "step": 120
    },
    {
      "epoch": 2.1946902654867255,
      "grad_norm": 0.22275066375732422,
      "learning_rate": 4.999881442314206e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 71256,
      "step": 125
    },
    {
      "epoch": 2.2831858407079646,
      "grad_norm": 4.021029472351074,
      "learning_rate": 4.9998716885297617e-05,
      "loss": 0.067,
      "num_input_tokens_seen": 74200,
      "step": 130
    },
    {
      "epoch": 2.3716814159292037,
      "grad_norm": 7.759142875671387,
      "learning_rate": 4.999861549233688e-05,
      "loss": 0.2343,
      "num_input_tokens_seen": 76696,
      "step": 135
    },
    {
      "epoch": 2.4601769911504423,
      "grad_norm": 5.954124450683594,
      "learning_rate": 4.999851024427548e-05,
      "loss": 0.0737,
      "num_input_tokens_seen": 79640,
      "step": 140
    },
    {
      "epoch": 2.5486725663716814,
      "grad_norm": 15.326788902282715,
      "learning_rate": 4.999840114112965e-05,
      "loss": 0.1341,
      "num_input_tokens_seen": 82632,
      "step": 145
    },
    {
      "epoch": 2.6371681415929205,
      "grad_norm": 19.98508071899414,
      "learning_rate": 4.999828818291621e-05,
      "loss": 0.0808,
      "num_input_tokens_seen": 85496,
      "step": 150
    },
    {
      "epoch": 2.725663716814159,
      "grad_norm": 0.07969048619270325,
      "learning_rate": 4.999817136965259e-05,
      "loss": 0.0805,
      "num_input_tokens_seen": 89128,
      "step": 155
    },
    {
      "epoch": 2.814159292035398,
      "grad_norm": 0.0036373238544911146,
      "learning_rate": 4.9998050701356794e-05,
      "loss": 0.0858,
      "num_input_tokens_seen": 92152,
      "step": 160
    },
    {
      "epoch": 2.9026548672566372,
      "grad_norm": 13.564420700073242,
      "learning_rate": 4.999792617804744e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 94632,
      "step": 165
    },
    {
      "epoch": 2.991150442477876,
      "grad_norm": 0.0744665190577507,
      "learning_rate": 4.9997797799743724e-05,
      "loss": 0.1869,
      "num_input_tokens_seen": 97544,
      "step": 170
    },
    {
      "epoch": 3.0707964601769913,
      "grad_norm": 0.007517859805375338,
      "learning_rate": 4.999766556646545e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 99992,
      "step": 175
    },
    {
      "epoch": 3.15929203539823,
      "grad_norm": 2.781057357788086,
      "learning_rate": 4.9997529478232996e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 102824,
      "step": 180
    },
    {
      "epoch": 3.247787610619469,
      "grad_norm": 0.03420477733016014,
      "learning_rate": 4.9997389535067365e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 105608,
      "step": 185
    },
    {
      "epoch": 3.336283185840708,
      "grad_norm": 0.021007733419537544,
      "learning_rate": 4.999724573699012e-05,
      "loss": 0.0016,
      "num_input_tokens_seen": 108712,
      "step": 190
    },
    {
      "epoch": 3.4247787610619467,
      "grad_norm": 0.03134279325604439,
      "learning_rate": 4.9997098084023457e-05,
      "loss": 0.067,
      "num_input_tokens_seen": 111608,
      "step": 195
    },
    {
      "epoch": 3.5132743362831858,
      "grad_norm": 14.289039611816406,
      "learning_rate": 4.999694657619013e-05,
      "loss": 0.1231,
      "num_input_tokens_seen": 114504,
      "step": 200
    },
    {
      "epoch": 3.5132743362831858,
      "eval_loss": 0.29828834533691406,
      "eval_runtime": 1.0601,
      "eval_samples_per_second": 23.582,
      "eval_steps_per_second": 12.263,
      "num_input_tokens_seen": 114504,
      "step": 200
    },
    {
      "epoch": 3.601769911504425,
      "grad_norm": 22.218660354614258,
      "learning_rate": 4.999679121351352e-05,
      "loss": 0.1848,
      "num_input_tokens_seen": 117592,
      "step": 205
    },
    {
      "epoch": 3.6902654867256635,
      "grad_norm": 0.20400172472000122,
      "learning_rate": 4.9996631996017565e-05,
      "loss": 0.0779,
      "num_input_tokens_seen": 120216,
      "step": 210
    },
    {
      "epoch": 3.7787610619469025,
      "grad_norm": 0.14002181589603424,
      "learning_rate": 4.9996468923726835e-05,
      "loss": 0.0765,
      "num_input_tokens_seen": 122904,
      "step": 215
    },
    {
      "epoch": 3.8672566371681416,
      "grad_norm": 3.200087308883667,
      "learning_rate": 4.999630199666647e-05,
      "loss": 0.1641,
      "num_input_tokens_seen": 125832,
      "step": 220
    },
    {
      "epoch": 3.9557522123893807,
      "grad_norm": 9.47286605834961,
      "learning_rate": 4.999613121486222e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 129512,
      "step": 225
    },
    {
      "epoch": 4.035398230088496,
      "grad_norm": 18.087106704711914,
      "learning_rate": 4.999595657834041e-05,
      "loss": 0.0775,
      "num_input_tokens_seen": 131872,
      "step": 230
    },
    {
      "epoch": 4.123893805309734,
      "grad_norm": 0.1326597034931183,
      "learning_rate": 4.999577808712798e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 134624,
      "step": 235
    },
    {
      "epoch": 4.212389380530974,
      "grad_norm": 0.019225556403398514,
      "learning_rate": 4.999559574125244e-05,
      "loss": 0.0015,
      "num_input_tokens_seen": 137168,
      "step": 240
    },
    {
      "epoch": 4.300884955752212,
      "grad_norm": 0.025584626942873,
      "learning_rate": 4.9995409540741934e-05,
      "loss": 0.1199,
      "num_input_tokens_seen": 139904,
      "step": 245
    },
    {
      "epoch": 4.389380530973451,
      "grad_norm": 11.48475170135498,
      "learning_rate": 4.999521948562516e-05,
      "loss": 0.0813,
      "num_input_tokens_seen": 142496,
      "step": 250
    },
    {
      "epoch": 4.477876106194691,
      "grad_norm": 0.12274622917175293,
      "learning_rate": 4.999502557593143e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 145632,
      "step": 255
    },
    {
      "epoch": 4.566371681415929,
      "grad_norm": 0.26201310753822327,
      "learning_rate": 4.999482781169066e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 148224,
      "step": 260
    },
    {
      "epoch": 4.654867256637168,
      "grad_norm": 0.010263658128678799,
      "learning_rate": 4.9994626192933324e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 150928,
      "step": 265
    },
    {
      "epoch": 4.743362831858407,
      "grad_norm": 0.00390905374661088,
      "learning_rate": 4.999442071969054e-05,
      "loss": 0.0179,
      "num_input_tokens_seen": 153808,
      "step": 270
    },
    {
      "epoch": 4.831858407079646,
      "grad_norm": 0.05844023451209068,
      "learning_rate": 4.999421139199397e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 157248,
      "step": 275
    },
    {
      "epoch": 4.920353982300885,
      "grad_norm": 8.711661338806152,
      "learning_rate": 4.999399820987592e-05,
      "loss": 0.0748,
      "num_input_tokens_seen": 160352,
      "step": 280
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.0009341348777525127,
      "learning_rate": 4.999378117336924e-05,
      "loss": 0.0004,
      "num_input_tokens_seen": 162728,
      "step": 285
    },
    {
      "epoch": 5.088495575221239,
      "grad_norm": 0.08718397468328476,
      "learning_rate": 4.9993560282507415e-05,
      "loss": 0.0003,
      "num_input_tokens_seen": 165368,
      "step": 290
    },
    {
      "epoch": 5.176991150442478,
      "grad_norm": 6.004726409912109,
      "learning_rate": 4.9993335537324495e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 168456,
      "step": 295
    },
    {
      "epoch": 5.265486725663717,
      "grad_norm": 0.007734427694231272,
      "learning_rate": 4.999310693785516e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 171512,
      "step": 300
    },
    {
      "epoch": 5.353982300884955,
      "grad_norm": 0.03447817638516426,
      "learning_rate": 4.9992874484134653e-05,
      "loss": 0.0004,
      "num_input_tokens_seen": 174920,
      "step": 305
    },
    {
      "epoch": 5.442477876106195,
      "grad_norm": 0.06811797618865967,
      "learning_rate": 4.999263817619882e-05,
      "loss": 0.001,
      "num_input_tokens_seen": 177496,
      "step": 310
    },
    {
      "epoch": 5.530973451327434,
      "grad_norm": 0.00865409430116415,
      "learning_rate": 4.9992398014084105e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 180328,
      "step": 315
    },
    {
      "epoch": 5.619469026548672,
      "grad_norm": 10.527838706970215,
      "learning_rate": 4.999215399782754e-05,
      "loss": 0.1955,
      "num_input_tokens_seen": 183000,
      "step": 320
    },
    {
      "epoch": 5.707964601769912,
      "grad_norm": 0.1341608464717865,
      "learning_rate": 4.999190612746675e-05,
      "loss": 0.0906,
      "num_input_tokens_seen": 185512,
      "step": 325
    },
    {
      "epoch": 5.79646017699115,
      "grad_norm": 0.8191097378730774,
      "learning_rate": 4.999165440303998e-05,
      "loss": 0.0029,
      "num_input_tokens_seen": 188776,
      "step": 330
    },
    {
      "epoch": 5.88495575221239,
      "grad_norm": 0.7769455909729004,
      "learning_rate": 4.999139882458603e-05,
      "loss": 0.0029,
      "num_input_tokens_seen": 191816,
      "step": 335
    },
    {
      "epoch": 5.9734513274336285,
      "grad_norm": 0.0037951546255499125,
      "learning_rate": 4.9991139392144314e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 194344,
      "step": 340
    },
    {
      "epoch": 6.053097345132743,
      "grad_norm": 0.0024611952248960733,
      "learning_rate": 4.999087610575485e-05,
      "loss": 0.0009,
      "num_input_tokens_seen": 197000,
      "step": 345
    },
    {
      "epoch": 6.1415929203539825,
      "grad_norm": 20.97620391845703,
      "learning_rate": 4.999060896545824e-05,
      "loss": 0.1099,
      "num_input_tokens_seen": 199992,
      "step": 350
    },
    {
      "epoch": 6.230088495575221,
      "grad_norm": 0.004613489378243685,
      "learning_rate": 4.999033797129568e-05,
      "loss": 0.1047,
      "num_input_tokens_seen": 203208,
      "step": 355
    },
    {
      "epoch": 6.31858407079646,
      "grad_norm": 0.07178334146738052,
      "learning_rate": 4.999006312330894e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 205976,
      "step": 360
    },
    {
      "epoch": 6.407079646017699,
      "grad_norm": 0.8935253620147705,
      "learning_rate": 4.998978442154043e-05,
      "loss": 0.0014,
      "num_input_tokens_seen": 209176,
      "step": 365
    },
    {
      "epoch": 6.495575221238938,
      "grad_norm": 0.00312030129134655,
      "learning_rate": 4.9989501866033125e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 211896,
      "step": 370
    },
    {
      "epoch": 6.584070796460177,
      "grad_norm": 0.0038556319195777178,
      "learning_rate": 4.998921545683059e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 214824,
      "step": 375
    },
    {
      "epoch": 6.672566371681416,
      "grad_norm": 0.0007887660758569837,
      "learning_rate": 4.9988925193976996e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 217192,
      "step": 380
    },
    {
      "epoch": 6.761061946902655,
      "grad_norm": 0.00054190430091694,
      "learning_rate": 4.998863107751711e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 219816,
      "step": 385
    },
    {
      "epoch": 6.849557522123893,
      "grad_norm": 0.0012305851560086012,
      "learning_rate": 4.998833310749629e-05,
      "loss": 0.0029,
      "num_input_tokens_seen": 222360,
      "step": 390
    },
    {
      "epoch": 6.938053097345133,
      "grad_norm": 0.014657985419034958,
      "learning_rate": 4.998803128396047e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 225864,
      "step": 395
    },
    {
      "epoch": 7.017699115044247,
      "grad_norm": 0.0017939229728654027,
      "learning_rate": 4.9987725606956215e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 228504,
      "step": 400
    },
    {
      "epoch": 7.017699115044247,
      "eval_loss": 0.27719274163246155,
      "eval_runtime": 1.0555,
      "eval_samples_per_second": 23.686,
      "eval_steps_per_second": 12.317,
      "num_input_tokens_seen": 228504,
      "step": 400
    },
    {
      "epoch": 7.106194690265487,
      "grad_norm": 0.021979769691824913,
      "learning_rate": 4.998741607653066e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 231416,
      "step": 405
    },
    {
      "epoch": 7.1946902654867255,
      "grad_norm": 0.004623004235327244,
      "learning_rate": 4.9987102692731523e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 233832,
      "step": 410
    },
    {
      "epoch": 7.283185840707965,
      "grad_norm": 0.13863764703273773,
      "learning_rate": 4.9986785455607157e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 236776,
      "step": 415
    },
    {
      "epoch": 7.371681415929204,
      "grad_norm": 0.008844073861837387,
      "learning_rate": 4.9986464365206456e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 239720,
      "step": 420
    },
    {
      "epoch": 7.460176991150442,
      "grad_norm": 0.0020716292783617973,
      "learning_rate": 4.9986139421578956e-05,
      "loss": 0.0013,
      "num_input_tokens_seen": 242344,
      "step": 425
    },
    {
      "epoch": 7.548672566371682,
      "grad_norm": 0.00820455327630043,
      "learning_rate": 4.998581062477477e-05,
      "loss": 0.065,
      "num_input_tokens_seen": 245208,
      "step": 430
    },
    {
      "epoch": 7.6371681415929205,
      "grad_norm": 0.001240293262526393,
      "learning_rate": 4.998547797484458e-05,
      "loss": 0.0262,
      "num_input_tokens_seen": 248232,
      "step": 435
    },
    {
      "epoch": 7.725663716814159,
      "grad_norm": 0.25911208987236023,
      "learning_rate": 4.9985141471839706e-05,
      "loss": 0.0012,
      "num_input_tokens_seen": 251080,
      "step": 440
    },
    {
      "epoch": 7.814159292035399,
      "grad_norm": 0.004413544666022062,
      "learning_rate": 4.998480111581203e-05,
      "loss": 0.0297,
      "num_input_tokens_seen": 253576,
      "step": 445
    },
    {
      "epoch": 7.902654867256637,
      "grad_norm": 0.00036551171797327697,
      "learning_rate": 4.998445690681405e-05,
      "loss": 0.0618,
      "num_input_tokens_seen": 256616,
      "step": 450
    },
    {
      "epoch": 7.991150442477876,
      "grad_norm": 0.00046993137220852077,
      "learning_rate": 4.9984108844898834e-05,
      "loss": 0.0004,
      "num_input_tokens_seen": 259544,
      "step": 455
    },
    {
      "epoch": 8.070796460176991,
      "grad_norm": 0.005696712527424097,
      "learning_rate": 4.9983756930120076e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 262232,
      "step": 460
    },
    {
      "epoch": 8.15929203539823,
      "grad_norm": 0.00035371974809095263,
      "learning_rate": 4.9983401162532025e-05,
      "loss": 0.1894,
      "num_input_tokens_seen": 265048,
      "step": 465
    },
    {
      "epoch": 8.247787610619469,
      "grad_norm": 0.002434206660836935,
      "learning_rate": 4.998304154218955e-05,
      "loss": 0.0006,
      "num_input_tokens_seen": 268008,
      "step": 470
    },
    {
      "epoch": 8.336283185840708,
      "grad_norm": 0.019680555909872055,
      "learning_rate": 4.998267806914812e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 270568,
      "step": 475
    },
    {
      "epoch": 8.424778761061948,
      "grad_norm": 0.02505149506032467,
      "learning_rate": 4.998231074346378e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 273224,
      "step": 480
    },
    {
      "epoch": 8.513274336283185,
      "grad_norm": 4.953235626220703,
      "learning_rate": 4.998193956519317e-05,
      "loss": 0.003,
      "num_input_tokens_seen": 276472,
      "step": 485
    },
    {
      "epoch": 8.601769911504425,
      "grad_norm": 0.004216191358864307,
      "learning_rate": 4.9981564534393545e-05,
      "loss": 0.0011,
      "num_input_tokens_seen": 279432,
      "step": 490
    },
    {
      "epoch": 8.690265486725664,
      "grad_norm": 0.021341145038604736,
      "learning_rate": 4.998118565112272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 282232,
      "step": 495
    },
    {
      "epoch": 8.778761061946902,
      "grad_norm": 0.0006672691670246422,
      "learning_rate": 4.998080291543914e-05,
      "loss": 0.0002,
      "num_input_tokens_seen": 284776,
      "step": 500
    },
    {
      "epoch": 8.867256637168142,
      "grad_norm": 0.0006633815355598927,
      "learning_rate": 4.9980416327401826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 287864,
      "step": 505
    },
    {
      "epoch": 8.955752212389381,
      "grad_norm": 0.0005416591884568334,
      "learning_rate": 4.998002588707038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 290712,
      "step": 510
    },
    {
      "epoch": 9.035398230088495,
      "grad_norm": 0.0009420346468687057,
      "learning_rate": 4.997963159450503e-05,
      "loss": 0.119,
      "num_input_tokens_seen": 293160,
      "step": 515
    },
    {
      "epoch": 9.123893805309734,
      "grad_norm": 0.0011862157844007015,
      "learning_rate": 4.9979233449766575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 296472,
      "step": 520
    },
    {
      "epoch": 9.212389380530974,
      "grad_norm": 0.003778920043259859,
      "learning_rate": 4.997883145291641e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 299112,
      "step": 525
    },
    {
      "epoch": 9.300884955752213,
      "grad_norm": 0.006915211211889982,
      "learning_rate": 4.9978425604016536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 302024,
      "step": 530
    },
    {
      "epoch": 9.389380530973451,
      "grad_norm": 0.01128897350281477,
      "learning_rate": 4.9978015903129536e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 304872,
      "step": 535
    },
    {
      "epoch": 9.47787610619469,
      "grad_norm": 0.0019245955627411604,
      "learning_rate": 4.997760235031859e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 307560,
      "step": 540
    },
    {
      "epoch": 9.56637168141593,
      "grad_norm": 0.0014792667934671044,
      "learning_rate": 4.9977184945647473e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 310840,
      "step": 545
    },
    {
      "epoch": 9.654867256637168,
      "grad_norm": 0.0035617174580693245,
      "learning_rate": 4.997676368918055e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 313800,
      "step": 550
    },
    {
      "epoch": 9.743362831858407,
      "grad_norm": 0.2023031860589981,
      "learning_rate": 4.9976338580982794e-05,
      "loss": 0.0018,
      "num_input_tokens_seen": 316664,
      "step": 555
    },
    {
      "epoch": 9.831858407079647,
      "grad_norm": 0.004004614427685738,
      "learning_rate": 4.9975909621119755e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 319656,
      "step": 560
    },
    {
      "epoch": 9.920353982300885,
      "grad_norm": 0.05386053025722504,
      "learning_rate": 4.997547680965758e-05,
      "loss": 0.0005,
      "num_input_tokens_seen": 322536,
      "step": 565
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.000544583483133465,
      "learning_rate": 4.997504014666302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 324624,
      "step": 570
    },
    {
      "epoch": 10.08849557522124,
      "grad_norm": 0.0020309293176978827,
      "learning_rate": 4.997459963220342e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 327488,
      "step": 575
    },
    {
      "epoch": 10.176991150442477,
      "grad_norm": 8.602375984191895,
      "learning_rate": 4.997415526634671e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 330240,
      "step": 580
    },
    {
      "epoch": 10.265486725663717,
      "grad_norm": 0.040568217635154724,
      "learning_rate": 4.99737070491614e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 332976,
      "step": 585
    },
    {
      "epoch": 10.353982300884956,
      "grad_norm": 0.19300879538059235,
      "learning_rate": 4.997325498071663e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 335504,
      "step": 590
    },
    {
      "epoch": 10.442477876106194,
      "grad_norm": 0.017961055040359497,
      "learning_rate": 4.997279906108211e-05,
      "loss": 0.0027,
      "num_input_tokens_seen": 338512,
      "step": 595
    },
    {
      "epoch": 10.530973451327434,
      "grad_norm": 0.0011345158563926816,
      "learning_rate": 4.9972339290328155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 341136,
      "step": 600
    },
    {
      "epoch": 10.530973451327434,
      "eval_loss": 0.2460305392742157,
      "eval_runtime": 1.0608,
      "eval_samples_per_second": 23.568,
      "eval_steps_per_second": 12.255,
      "num_input_tokens_seen": 341136,
      "step": 600
    },
    {
      "epoch": 10.619469026548673,
      "grad_norm": 0.0019419572781771421,
      "learning_rate": 4.9971875668525646e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 343664,
      "step": 605
    },
    {
      "epoch": 10.70796460176991,
      "grad_norm": 0.0012515434063971043,
      "learning_rate": 4.997140819574609e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 346656,
      "step": 610
    },
    {
      "epoch": 10.79646017699115,
      "grad_norm": 0.0019906782545149326,
      "learning_rate": 4.997093687206159e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 349712,
      "step": 615
    },
    {
      "epoch": 10.88495575221239,
      "grad_norm": 0.007730717305094004,
      "learning_rate": 4.997046169754482e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 352896,
      "step": 620
    },
    {
      "epoch": 10.973451327433628,
      "grad_norm": 0.0005708897951990366,
      "learning_rate": 4.996998267226905e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 356016,
      "step": 625
    },
    {
      "epoch": 11.053097345132743,
      "grad_norm": 0.002106466330587864,
      "learning_rate": 4.996949979630817e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 358432,
      "step": 630
    },
    {
      "epoch": 11.141592920353983,
      "grad_norm": 0.004296265542507172,
      "learning_rate": 4.996901306973663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 361008,
      "step": 635
    },
    {
      "epoch": 11.230088495575222,
      "grad_norm": 0.00020183685410302132,
      "learning_rate": 4.996852249262949e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 363728,
      "step": 640
    },
    {
      "epoch": 11.31858407079646,
      "grad_norm": 0.0011755888117477298,
      "learning_rate": 4.996802806506241e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 367232,
      "step": 645
    },
    {
      "epoch": 11.4070796460177,
      "grad_norm": 0.0012576512526720762,
      "learning_rate": 4.996752978711164e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 370448,
      "step": 650
    },
    {
      "epoch": 11.495575221238939,
      "grad_norm": 0.01160577405244112,
      "learning_rate": 4.996702765885401e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 373504,
      "step": 655
    },
    {
      "epoch": 11.584070796460177,
      "grad_norm": 0.0036855109501630068,
      "learning_rate": 4.9966521680366964e-05,
      "loss": 0.0015,
      "num_input_tokens_seen": 376464,
      "step": 660
    },
    {
      "epoch": 11.672566371681416,
      "grad_norm": 0.0013581538805738091,
      "learning_rate": 4.9966011851728524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 379216,
      "step": 665
    },
    {
      "epoch": 11.761061946902656,
      "grad_norm": 0.0003862738376483321,
      "learning_rate": 4.996549817301731e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 381664,
      "step": 670
    },
    {
      "epoch": 11.849557522123893,
      "grad_norm": 0.0024767389986664057,
      "learning_rate": 4.9964980644312544e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 384432,
      "step": 675
    },
    {
      "epoch": 11.938053097345133,
      "grad_norm": 0.0010958511848002672,
      "learning_rate": 4.996445926569403e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 387296,
      "step": 680
    },
    {
      "epoch": 12.017699115044248,
      "grad_norm": 0.0021056572441011667,
      "learning_rate": 4.996393403724218e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 390160,
      "step": 685
    },
    {
      "epoch": 12.106194690265486,
      "grad_norm": 0.0032860245555639267,
      "learning_rate": 4.9963404959037985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 392944,
      "step": 690
    },
    {
      "epoch": 12.194690265486726,
      "grad_norm": 0.012496787123382092,
      "learning_rate": 4.996287203116303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 395984,
      "step": 695
    },
    {
      "epoch": 12.283185840707965,
      "grad_norm": 0.005093643441796303,
      "learning_rate": 4.996233525369951e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 398848,
      "step": 700
    },
    {
      "epoch": 12.371681415929203,
      "grad_norm": 0.007033719681203365,
      "learning_rate": 4.99617946267302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 401536,
      "step": 705
    },
    {
      "epoch": 12.460176991150442,
      "grad_norm": 0.0038142423145473003,
      "learning_rate": 4.996125015033846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 404864,
      "step": 710
    },
    {
      "epoch": 12.548672566371682,
      "grad_norm": 0.0026015089824795723,
      "learning_rate": 4.996070182460827e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 407744,
      "step": 715
    },
    {
      "epoch": 12.63716814159292,
      "grad_norm": 0.0004416556912474334,
      "learning_rate": 4.996014964962418e-05,
      "loss": 0.0002,
      "num_input_tokens_seen": 410416,
      "step": 720
    },
    {
      "epoch": 12.725663716814159,
      "grad_norm": 0.002048628870397806,
      "learning_rate": 4.9959593625471344e-05,
      "loss": 0.0004,
      "num_input_tokens_seen": 413488,
      "step": 725
    },
    {
      "epoch": 12.814159292035399,
      "grad_norm": 0.0003395231906324625,
      "learning_rate": 4.995903375223552e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 416048,
      "step": 730
    },
    {
      "epoch": 12.902654867256636,
      "grad_norm": 0.00046195462346076965,
      "learning_rate": 4.995847003000302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 418944,
      "step": 735
    },
    {
      "epoch": 12.991150442477876,
      "grad_norm": 0.2302684485912323,
      "learning_rate": 4.9957902458860804e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 421632,
      "step": 740
    },
    {
      "epoch": 13.070796460176991,
      "grad_norm": 0.004933107644319534,
      "learning_rate": 4.995733103889639e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 424608,
      "step": 745
    },
    {
      "epoch": 13.15929203539823,
      "grad_norm": 0.000595191668253392,
      "learning_rate": 4.99567557701979e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 427088,
      "step": 750
    },
    {
      "epoch": 13.247787610619469,
      "grad_norm": 0.0007170175667852163,
      "learning_rate": 4.995617665285403e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 429744,
      "step": 755
    },
    {
      "epoch": 13.336283185840708,
      "grad_norm": 0.0013576113851740956,
      "learning_rate": 4.99555936869541e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 432752,
      "step": 760
    },
    {
      "epoch": 13.424778761061948,
      "grad_norm": 0.0006917788414284587,
      "learning_rate": 4.995500687258803e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 435456,
      "step": 765
    },
    {
      "epoch": 13.513274336283185,
      "grad_norm": 0.0020785953383892775,
      "learning_rate": 4.995441620984628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 438608,
      "step": 770
    },
    {
      "epoch": 13.601769911504425,
      "grad_norm": 0.0012403321452438831,
      "learning_rate": 4.995382169881996e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 441248,
      "step": 775
    },
    {
      "epoch": 13.690265486725664,
      "grad_norm": 0.07358422130346298,
      "learning_rate": 4.9953223339600755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 444160,
      "step": 780
    },
    {
      "epoch": 13.778761061946902,
      "grad_norm": 0.0001986919087357819,
      "learning_rate": 4.995262113228091e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 447200,
      "step": 785
    },
    {
      "epoch": 13.867256637168142,
      "grad_norm": 0.000986117753200233,
      "learning_rate": 4.995201507695332e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 450320,
      "step": 790
    },
    {
      "epoch": 13.955752212389381,
      "grad_norm": 0.040793124586343765,
      "learning_rate": 4.995140517371144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 453328,
      "step": 795
    },
    {
      "epoch": 14.035398230088495,
      "grad_norm": 0.00034844339825212955,
      "learning_rate": 4.995079142264932e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 455488,
      "step": 800
    },
    {
      "epoch": 14.035398230088495,
      "eval_loss": 0.32453352212905884,
      "eval_runtime": 1.0603,
      "eval_samples_per_second": 23.578,
      "eval_steps_per_second": 12.261,
      "num_input_tokens_seen": 455488,
      "step": 800
    },
    {
      "epoch": 14.123893805309734,
      "grad_norm": 0.0008354753954336047,
      "learning_rate": 4.995017382386162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 458400,
      "step": 805
    },
    {
      "epoch": 14.212389380530974,
      "grad_norm": 0.004144582431763411,
      "learning_rate": 4.994955237744356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 461168,
      "step": 810
    },
    {
      "epoch": 14.300884955752213,
      "grad_norm": 0.0005700404872186482,
      "learning_rate": 4.994892708349101e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 463840,
      "step": 815
    },
    {
      "epoch": 14.389380530973451,
      "grad_norm": 0.0011508880415931344,
      "learning_rate": 4.994829794210035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 467008,
      "step": 820
    },
    {
      "epoch": 14.47787610619469,
      "grad_norm": 0.017156166955828667,
      "learning_rate": 4.994766495336864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 469856,
      "step": 825
    },
    {
      "epoch": 14.56637168141593,
      "grad_norm": 0.00038427498657256365,
      "learning_rate": 4.994702811739348e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 472672,
      "step": 830
    },
    {
      "epoch": 14.654867256637168,
      "grad_norm": 0.0006264999974519014,
      "learning_rate": 4.994638743427308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 475648,
      "step": 835
    },
    {
      "epoch": 14.743362831858407,
      "grad_norm": 0.0004542154201772064,
      "learning_rate": 4.994574290410624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 479216,
      "step": 840
    },
    {
      "epoch": 14.831858407079647,
      "grad_norm": 0.0008495101355947554,
      "learning_rate": 4.9945094526992364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 481744,
      "step": 845
    },
    {
      "epoch": 14.920353982300885,
      "grad_norm": 0.00041137964581139386,
      "learning_rate": 4.994444230303142e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 484640,
      "step": 850
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.00029967798036523163,
      "learning_rate": 4.994378623232402e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 486672,
      "step": 855
    },
    {
      "epoch": 15.08849557522124,
      "grad_norm": 0.0002696386072784662,
      "learning_rate": 4.99431263149713e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 489680,
      "step": 860
    },
    {
      "epoch": 15.176991150442477,
      "grad_norm": 0.0006223974633030593,
      "learning_rate": 4.9942462551075056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 492864,
      "step": 865
    },
    {
      "epoch": 15.265486725663717,
      "grad_norm": 0.0005648406222462654,
      "learning_rate": 4.994179494073764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 495568,
      "step": 870
    },
    {
      "epoch": 15.353982300884956,
      "grad_norm": 0.0008057455997914076,
      "learning_rate": 4.9941123484062e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 498544,
      "step": 875
    },
    {
      "epoch": 15.442477876106194,
      "grad_norm": 0.0024189669638872147,
      "learning_rate": 4.99404481811517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 501472,
      "step": 880
    },
    {
      "epoch": 15.530973451327434,
      "grad_norm": 0.021692069247364998,
      "learning_rate": 4.9939769032110864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 504064,
      "step": 885
    },
    {
      "epoch": 15.619469026548673,
      "grad_norm": 0.0006123051862232387,
      "learning_rate": 4.993908603704423e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 506816,
      "step": 890
    },
    {
      "epoch": 15.70796460176991,
      "grad_norm": 0.00013217881496530026,
      "learning_rate": 4.9938399196057126e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 509552,
      "step": 895
    },
    {
      "epoch": 15.79646017699115,
      "grad_norm": 0.00033828429877758026,
      "learning_rate": 4.993770850925547e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 512576,
      "step": 900
    },
    {
      "epoch": 15.88495575221239,
      "grad_norm": 0.0014007972786203027,
      "learning_rate": 4.993701397674577e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 515664,
      "step": 905
    },
    {
      "epoch": 15.973451327433628,
      "grad_norm": 0.00011489438475109637,
      "learning_rate": 4.993631559863515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 518464,
      "step": 910
    },
    {
      "epoch": 16.053097345132745,
      "grad_norm": 0.0018798514502122998,
      "learning_rate": 4.9935613375031283e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 520584,
      "step": 915
    },
    {
      "epoch": 16.141592920353983,
      "grad_norm": 0.0003538216988090426,
      "learning_rate": 4.993490730604248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 524104,
      "step": 920
    },
    {
      "epoch": 16.23008849557522,
      "grad_norm": 0.0010786502389237285,
      "learning_rate": 4.993419739177761e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 527352,
      "step": 925
    },
    {
      "epoch": 16.31858407079646,
      "grad_norm": 0.0008712239796295762,
      "learning_rate": 4.9933483632346164e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 530328,
      "step": 930
    },
    {
      "epoch": 16.4070796460177,
      "grad_norm": 0.000993463909253478,
      "learning_rate": 4.993276602785821e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 532936,
      "step": 935
    },
    {
      "epoch": 16.495575221238937,
      "grad_norm": 0.00037782100844196975,
      "learning_rate": 4.993204457842441e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 536168,
      "step": 940
    },
    {
      "epoch": 16.58407079646018,
      "grad_norm": 0.0001233874208992347,
      "learning_rate": 4.993131928415602e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 538728,
      "step": 945
    },
    {
      "epoch": 16.672566371681416,
      "grad_norm": 0.00029226118931546807,
      "learning_rate": 4.993059014516489e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 541832,
      "step": 950
    },
    {
      "epoch": 16.761061946902654,
      "grad_norm": 0.0004983594408258796,
      "learning_rate": 4.9929857161563464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 544440,
      "step": 955
    },
    {
      "epoch": 16.849557522123895,
      "grad_norm": 0.0011184505419805646,
      "learning_rate": 4.992912033346477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 547256,
      "step": 960
    },
    {
      "epoch": 16.938053097345133,
      "grad_norm": 0.0009019665303640068,
      "learning_rate": 4.992837966098245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 550360,
      "step": 965
    },
    {
      "epoch": 17.01769911504425,
      "grad_norm": 0.00035257049603387713,
      "learning_rate": 4.992763514423071e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 552464,
      "step": 970
    },
    {
      "epoch": 17.106194690265486,
      "grad_norm": 0.000289648596663028,
      "learning_rate": 4.992688678332437e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 555040,
      "step": 975
    },
    {
      "epoch": 17.194690265486727,
      "grad_norm": 0.0003278889926150441,
      "learning_rate": 4.992613457837884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 557968,
      "step": 980
    },
    {
      "epoch": 17.283185840707965,
      "grad_norm": 0.00045742731890641153,
      "learning_rate": 4.992537852951011e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 560768,
      "step": 985
    },
    {
      "epoch": 17.371681415929203,
      "grad_norm": 0.0007173553458414972,
      "learning_rate": 4.9924618636834785e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 563424,
      "step": 990
    },
    {
      "epoch": 17.460176991150444,
      "grad_norm": 0.0002477349480614066,
      "learning_rate": 4.9923854900470046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 566384,
      "step": 995
    },
    {
      "epoch": 17.548672566371682,
      "grad_norm": 0.00030144694028422236,
      "learning_rate": 4.992308732053367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 569504,
      "step": 1000
    },
    {
      "epoch": 17.548672566371682,
      "eval_loss": 0.32255837321281433,
      "eval_runtime": 1.0601,
      "eval_samples_per_second": 23.583,
      "eval_steps_per_second": 12.263,
      "num_input_tokens_seen": 569504,
      "step": 1000
    },
    {
      "epoch": 17.63716814159292,
      "grad_norm": 0.0008658593287691474,
      "learning_rate": 4.992231589714402e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 572608,
      "step": 1005
    },
    {
      "epoch": 17.72566371681416,
      "grad_norm": 0.0008811381994746625,
      "learning_rate": 4.992154063042007e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 575504,
      "step": 1010
    },
    {
      "epoch": 17.8141592920354,
      "grad_norm": 0.00029839613125659525,
      "learning_rate": 4.992076152048136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 578064,
      "step": 1015
    },
    {
      "epoch": 17.902654867256636,
      "grad_norm": 0.00012245471589267254,
      "learning_rate": 4.991997856744807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 581312,
      "step": 1020
    },
    {
      "epoch": 17.991150442477878,
      "grad_norm": 0.0003122404159512371,
      "learning_rate": 4.9919191771440905e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 584240,
      "step": 1025
    },
    {
      "epoch": 18.07079646017699,
      "grad_norm": 0.00044089319999329746,
      "learning_rate": 4.991840113258122e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 586296,
      "step": 1030
    },
    {
      "epoch": 18.15929203539823,
      "grad_norm": 0.0008907542214728892,
      "learning_rate": 4.9917606650990933e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 588936,
      "step": 1035
    },
    {
      "epoch": 18.24778761061947,
      "grad_norm": 0.001218626624904573,
      "learning_rate": 4.9916808326792566e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 591816,
      "step": 1040
    },
    {
      "epoch": 18.336283185840706,
      "grad_norm": 0.0011305089574307203,
      "learning_rate": 4.9916006160109235e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 595000,
      "step": 1045
    },
    {
      "epoch": 18.424778761061948,
      "grad_norm": 0.0006374447839334607,
      "learning_rate": 4.991520015106464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 597960,
      "step": 1050
    },
    {
      "epoch": 18.513274336283185,
      "grad_norm": 0.0019155779154971242,
      "learning_rate": 4.991439029978308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 600728,
      "step": 1055
    },
    {
      "epoch": 18.601769911504427,
      "grad_norm": 0.0013557582860812545,
      "learning_rate": 4.9913576606389434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 603496,
      "step": 1060
    },
    {
      "epoch": 18.690265486725664,
      "grad_norm": 0.001460827304981649,
      "learning_rate": 4.991275907100919e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 606104,
      "step": 1065
    },
    {
      "epoch": 18.778761061946902,
      "grad_norm": 0.0004133485199417919,
      "learning_rate": 4.9911937693768434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 609160,
      "step": 1070
    },
    {
      "epoch": 18.86725663716814,
      "grad_norm": 0.0003913461696356535,
      "learning_rate": 4.991111247479382e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 612264,
      "step": 1075
    },
    {
      "epoch": 18.95575221238938,
      "grad_norm": 0.0002515081432648003,
      "learning_rate": 4.9910283414212605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 615256,
      "step": 1080
    },
    {
      "epoch": 19.035398230088497,
      "grad_norm": 0.000579297193326056,
      "learning_rate": 4.990945051215265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 617880,
      "step": 1085
    },
    {
      "epoch": 19.123893805309734,
      "grad_norm": 0.00045741861686110497,
      "learning_rate": 4.99086137687424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 620360,
      "step": 1090
    },
    {
      "epoch": 19.212389380530972,
      "grad_norm": 0.0003122776688542217,
      "learning_rate": 4.9907773184110874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 623224,
      "step": 1095
    },
    {
      "epoch": 19.300884955752213,
      "grad_norm": 0.0002691705303732306,
      "learning_rate": 4.9906928758387715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 625560,
      "step": 1100
    },
    {
      "epoch": 19.38938053097345,
      "grad_norm": 0.00032366198138333857,
      "learning_rate": 4.9906080491703146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 628488,
      "step": 1105
    },
    {
      "epoch": 19.47787610619469,
      "grad_norm": 0.00013737140398006886,
      "learning_rate": 4.990522838418797e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 631400,
      "step": 1110
    },
    {
      "epoch": 19.56637168141593,
      "grad_norm": 0.000284130044747144,
      "learning_rate": 4.9904372435973604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 634408,
      "step": 1115
    },
    {
      "epoch": 19.654867256637168,
      "grad_norm": 0.0009706313139759004,
      "learning_rate": 4.990351264719203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 637448,
      "step": 1120
    },
    {
      "epoch": 19.743362831858406,
      "grad_norm": 0.0002732073771767318,
      "learning_rate": 4.990264901797586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 640280,
      "step": 1125
    },
    {
      "epoch": 19.831858407079647,
      "grad_norm": 0.0008934547659009695,
      "learning_rate": 4.990178154845826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 642984,
      "step": 1130
    },
    {
      "epoch": 19.920353982300885,
      "grad_norm": 0.0003161237691529095,
      "learning_rate": 4.9900910238773014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 646072,
      "step": 1135
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.00019830200471915305,
      "learning_rate": 4.990003508905448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 648736,
      "step": 1140
    },
    {
      "epoch": 20.088495575221238,
      "grad_norm": 0.0015212713042274117,
      "learning_rate": 4.989915609943763e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 651264,
      "step": 1145
    },
    {
      "epoch": 20.17699115044248,
      "grad_norm": 0.00010795858543133363,
      "learning_rate": 4.9898273270058e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 654352,
      "step": 1150
    },
    {
      "epoch": 20.265486725663717,
      "grad_norm": 0.0005103930016048253,
      "learning_rate": 4.989738660105174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 657088,
      "step": 1155
    },
    {
      "epoch": 20.353982300884955,
      "grad_norm": 0.00012371719640213996,
      "learning_rate": 4.989649609255559e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 659968,
      "step": 1160
    },
    {
      "epoch": 20.442477876106196,
      "grad_norm": 0.00011301910853944719,
      "learning_rate": 4.989560174470687e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 663152,
      "step": 1165
    },
    {
      "epoch": 20.530973451327434,
      "grad_norm": 0.019765768200159073,
      "learning_rate": 4.989470355764351e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 666240,
      "step": 1170
    },
    {
      "epoch": 20.61946902654867,
      "grad_norm": 0.000147049140650779,
      "learning_rate": 4.9893801531504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 668896,
      "step": 1175
    },
    {
      "epoch": 20.707964601769913,
      "grad_norm": 0.0004554605111479759,
      "learning_rate": 4.9892895666427475e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 671376,
      "step": 1180
    },
    {
      "epoch": 20.79646017699115,
      "grad_norm": 0.00013317918637767434,
      "learning_rate": 4.9891985962553606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 674304,
      "step": 1185
    },
    {
      "epoch": 20.884955752212388,
      "grad_norm": 0.00019648313173092902,
      "learning_rate": 4.989107242002269e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 677152,
      "step": 1190
    },
    {
      "epoch": 20.97345132743363,
      "grad_norm": 0.00017418782226741314,
      "learning_rate": 4.989015503897561e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 679840,
      "step": 1195
    },
    {
      "epoch": 21.053097345132745,
      "grad_norm": 0.00011750702833523974,
      "learning_rate": 4.988923381955383e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 682024,
      "step": 1200
    },
    {
      "epoch": 21.053097345132745,
      "eval_loss": 0.3285369575023651,
      "eval_runtime": 1.0594,
      "eval_samples_per_second": 23.597,
      "eval_steps_per_second": 12.271,
      "num_input_tokens_seen": 682024,
      "step": 1200
    },
    {
      "epoch": 21.141592920353983,
      "grad_norm": 0.0012259316863492131,
      "learning_rate": 4.988830876189942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 685320,
      "step": 1205
    },
    {
      "epoch": 21.23008849557522,
      "grad_norm": 0.0005495221703313291,
      "learning_rate": 4.988737986615503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 688200,
      "step": 1210
    },
    {
      "epoch": 21.31858407079646,
      "grad_norm": 0.0005997789558023214,
      "learning_rate": 4.988644713246391e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 691432,
      "step": 1215
    },
    {
      "epoch": 21.4070796460177,
      "grad_norm": 0.000248508236836642,
      "learning_rate": 4.988551056096991e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 694088,
      "step": 1220
    },
    {
      "epoch": 21.495575221238937,
      "grad_norm": 0.0007198575185611844,
      "learning_rate": 4.988457015181743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 696568,
      "step": 1225
    },
    {
      "epoch": 21.58407079646018,
      "grad_norm": 0.0006390251219272614,
      "learning_rate": 4.988362590515153e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 699128,
      "step": 1230
    },
    {
      "epoch": 21.672566371681416,
      "grad_norm": 0.0006380958366207778,
      "learning_rate": 4.9882677821117805e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 702936,
      "step": 1235
    },
    {
      "epoch": 21.761061946902654,
      "grad_norm": 0.00021856748207937926,
      "learning_rate": 4.988172589986246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 705560,
      "step": 1240
    },
    {
      "epoch": 21.849557522123895,
      "grad_norm": 0.00024148618103936315,
      "learning_rate": 4.9880770141532304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 708312,
      "step": 1245
    },
    {
      "epoch": 21.938053097345133,
      "grad_norm": 0.0001464893139200285,
      "learning_rate": 4.987981054627472e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 711224,
      "step": 1250
    },
    {
      "epoch": 22.01769911504425,
      "grad_norm": 0.0005165801849216223,
      "learning_rate": 4.987884711423769e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 713400,
      "step": 1255
    },
    {
      "epoch": 22.106194690265486,
      "grad_norm": 0.0009242875385098159,
      "learning_rate": 4.9877879845569784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 716440,
      "step": 1260
    },
    {
      "epoch": 22.194690265486727,
      "grad_norm": 0.00016186076391022652,
      "learning_rate": 4.9876908740420175e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 719112,
      "step": 1265
    },
    {
      "epoch": 22.283185840707965,
      "grad_norm": 0.0008543524891138077,
      "learning_rate": 4.987593379893861e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 722520,
      "step": 1270
    },
    {
      "epoch": 22.371681415929203,
      "grad_norm": 0.00012664805399253964,
      "learning_rate": 4.987495502127545e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 726040,
      "step": 1275
    },
    {
      "epoch": 22.460176991150444,
      "grad_norm": 0.00020888607832603157,
      "learning_rate": 4.987397240758162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 728712,
      "step": 1280
    },
    {
      "epoch": 22.548672566371682,
      "grad_norm": 0.000778821122366935,
      "learning_rate": 4.9872985958008664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 731848,
      "step": 1285
    },
    {
      "epoch": 22.63716814159292,
      "grad_norm": 0.0003232439048588276,
      "learning_rate": 4.987199567270871e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 734504,
      "step": 1290
    },
    {
      "epoch": 22.72566371681416,
      "grad_norm": 0.0006002117297612131,
      "learning_rate": 4.9871001551834444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 737240,
      "step": 1295
    },
    {
      "epoch": 22.8141592920354,
      "grad_norm": 0.0005520267877727747,
      "learning_rate": 4.98700035955392e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 740024,
      "step": 1300
    },
    {
      "epoch": 22.902654867256636,
      "grad_norm": 0.0006427133921533823,
      "learning_rate": 4.986900180397686e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 742568,
      "step": 1305
    },
    {
      "epoch": 22.991150442477878,
      "grad_norm": 0.0002589756913948804,
      "learning_rate": 4.9867996177301926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 745224,
      "step": 1310
    },
    {
      "epoch": 23.07079646017699,
      "grad_norm": 8.518361573806033e-05,
      "learning_rate": 4.9866986715669464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 747824,
      "step": 1315
    },
    {
      "epoch": 23.15929203539823,
      "grad_norm": 0.001051821862347424,
      "learning_rate": 4.9865973419235155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 750752,
      "step": 1320
    },
    {
      "epoch": 23.24778761061947,
      "grad_norm": 0.001523130340501666,
      "learning_rate": 4.986495628815526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 753872,
      "step": 1325
    },
    {
      "epoch": 23.336283185840706,
      "grad_norm": 0.00035833631409332156,
      "learning_rate": 4.986393532258663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 756752,
      "step": 1330
    },
    {
      "epoch": 23.424778761061948,
      "grad_norm": 0.0004111168091185391,
      "learning_rate": 4.986291052268671e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 759328,
      "step": 1335
    },
    {
      "epoch": 23.513274336283185,
      "grad_norm": 0.0003605491074267775,
      "learning_rate": 4.986188188861355e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 762224,
      "step": 1340
    },
    {
      "epoch": 23.601769911504427,
      "grad_norm": 0.0005755495512858033,
      "learning_rate": 4.9860849420525766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 765344,
      "step": 1345
    },
    {
      "epoch": 23.690265486725664,
      "grad_norm": 0.005162349436432123,
      "learning_rate": 4.9859813118582575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 768256,
      "step": 1350
    },
    {
      "epoch": 23.778761061946902,
      "grad_norm": 0.0002728042018134147,
      "learning_rate": 4.98587729829438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 770960,
      "step": 1355
    },
    {
      "epoch": 23.86725663716814,
      "grad_norm": 0.00024401921837124974,
      "learning_rate": 4.985772901376983e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 773584,
      "step": 1360
    },
    {
      "epoch": 23.95575221238938,
      "grad_norm": 0.0003736689977813512,
      "learning_rate": 4.9856681211221666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 776848,
      "step": 1365
    },
    {
      "epoch": 24.035398230088497,
      "grad_norm": 7.618107338203117e-05,
      "learning_rate": 4.985562957546089e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 779304,
      "step": 1370
    },
    {
      "epoch": 24.123893805309734,
      "grad_norm": 0.00043841602746397257,
      "learning_rate": 4.9854574106649686e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 782088,
      "step": 1375
    },
    {
      "epoch": 24.212389380530972,
      "grad_norm": 0.0004642882267944515,
      "learning_rate": 4.985351480495081e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 785080,
      "step": 1380
    },
    {
      "epoch": 24.300884955752213,
      "grad_norm": 0.0001836758165154606,
      "learning_rate": 4.985245167052762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 787800,
      "step": 1385
    },
    {
      "epoch": 24.38938053097345,
      "grad_norm": 7.74842337705195e-05,
      "learning_rate": 4.9851384703544066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 790424,
      "step": 1390
    },
    {
      "epoch": 24.47787610619469,
      "grad_norm": 7.478034967789426e-05,
      "learning_rate": 4.985031390416469e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 793032,
      "step": 1395
    },
    {
      "epoch": 24.56637168141593,
      "grad_norm": 0.00020559602126013488,
      "learning_rate": 4.984923927255461e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 796328,
      "step": 1400
    },
    {
      "epoch": 24.56637168141593,
      "eval_loss": 0.33615559339523315,
      "eval_runtime": 1.0614,
      "eval_samples_per_second": 23.554,
      "eval_steps_per_second": 12.248,
      "num_input_tokens_seen": 796328,
      "step": 1400
    },
    {
      "epoch": 24.654867256637168,
      "grad_norm": 0.000756134744733572,
      "learning_rate": 4.984816080887958e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 799176,
      "step": 1405
    },
    {
      "epoch": 24.743362831858406,
      "grad_norm": 0.0002768280392047018,
      "learning_rate": 4.9847078513305875e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 802072,
      "step": 1410
    },
    {
      "epoch": 24.831858407079647,
      "grad_norm": 0.0008831946179270744,
      "learning_rate": 4.984599238600043e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 804648,
      "step": 1415
    },
    {
      "epoch": 24.920353982300885,
      "grad_norm": 0.00027520820731297135,
      "learning_rate": 4.9844902427130716e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 807368,
      "step": 1420
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.00015700951917096972,
      "learning_rate": 4.984380863686482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 810192,
      "step": 1425
    },
    {
      "epoch": 25.088495575221238,
      "grad_norm": 0.00036796709173358977,
      "learning_rate": 4.984271101537143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 812960,
      "step": 1430
    },
    {
      "epoch": 25.17699115044248,
      "grad_norm": 0.0004097498022019863,
      "learning_rate": 4.9841609562819816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 816256,
      "step": 1435
    },
    {
      "epoch": 25.265486725663717,
      "grad_norm": 0.0005331021384336054,
      "learning_rate": 4.984050427937983e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 819232,
      "step": 1440
    },
    {
      "epoch": 25.353982300884955,
      "grad_norm": 0.0006165427039377391,
      "learning_rate": 4.983939516522191e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 822080,
      "step": 1445
    },
    {
      "epoch": 25.442477876106196,
      "grad_norm": 0.00015595488366670907,
      "learning_rate": 4.983828222051711e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 824688,
      "step": 1450
    },
    {
      "epoch": 25.530973451327434,
      "grad_norm": 0.00020890269661322236,
      "learning_rate": 4.983716544543705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 827360,
      "step": 1455
    },
    {
      "epoch": 25.61946902654867,
      "grad_norm": 0.00015757766959723085,
      "learning_rate": 4.983604484015395e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 830848,
      "step": 1460
    },
    {
      "epoch": 25.707964601769913,
      "grad_norm": 0.00036990962689742446,
      "learning_rate": 4.983492040484064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 833264,
      "step": 1465
    },
    {
      "epoch": 25.79646017699115,
      "grad_norm": 9.64353239396587e-05,
      "learning_rate": 4.98337921396705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 836480,
      "step": 1470
    },
    {
      "epoch": 25.884955752212388,
      "grad_norm": 0.0002745796518865973,
      "learning_rate": 4.983266004481753e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 839168,
      "step": 1475
    },
    {
      "epoch": 25.97345132743363,
      "grad_norm": 0.00023308982781600207,
      "learning_rate": 4.9831524120456316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 842160,
      "step": 1480
    },
    {
      "epoch": 26.053097345132745,
      "grad_norm": 0.00035840386408381164,
      "learning_rate": 4.9830384366762026e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 844392,
      "step": 1485
    },
    {
      "epoch": 26.141592920353983,
      "grad_norm": 0.00020141304412391037,
      "learning_rate": 4.9829240783910436e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 847464,
      "step": 1490
    },
    {
      "epoch": 26.23008849557522,
      "grad_norm": 0.00016767541819717735,
      "learning_rate": 4.982809337207789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 849752,
      "step": 1495
    },
    {
      "epoch": 26.31858407079646,
      "grad_norm": 0.00020944034622516483,
      "learning_rate": 4.9826942131441337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 852632,
      "step": 1500
    },
    {
      "epoch": 26.4070796460177,
      "grad_norm": 0.0001551635068608448,
      "learning_rate": 4.9825787062178315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 855640,
      "step": 1505
    },
    {
      "epoch": 26.495575221238937,
      "grad_norm": 0.0003660897782538086,
      "learning_rate": 4.9824628164466945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 858728,
      "step": 1510
    },
    {
      "epoch": 26.58407079646018,
      "grad_norm": 0.00014047816512174904,
      "learning_rate": 4.982346543848595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 861736,
      "step": 1515
    },
    {
      "epoch": 26.672566371681416,
      "grad_norm": 0.0003159353800583631,
      "learning_rate": 4.9822298884414626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 864248,
      "step": 1520
    },
    {
      "epoch": 26.761061946902654,
      "grad_norm": 0.0014063138514757156,
      "learning_rate": 4.982112850243288e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 867288,
      "step": 1525
    },
    {
      "epoch": 26.849557522123895,
      "grad_norm": 0.0006448616040870547,
      "learning_rate": 4.98199542927212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 870536,
      "step": 1530
    },
    {
      "epoch": 26.938053097345133,
      "grad_norm": 0.00012189028348075226,
      "learning_rate": 4.981877625546066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 873048,
      "step": 1535
    },
    {
      "epoch": 27.01769911504425,
      "grad_norm": 0.00032792025012895465,
      "learning_rate": 4.981759439083293e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 875304,
      "step": 1540
    },
    {
      "epoch": 27.106194690265486,
      "grad_norm": 0.0002749502018559724,
      "learning_rate": 4.981640869902027e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 878312,
      "step": 1545
    },
    {
      "epoch": 27.194690265486727,
      "grad_norm": 0.0009039886645041406,
      "learning_rate": 4.9815219180205517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 881192,
      "step": 1550
    },
    {
      "epoch": 27.283185840707965,
      "grad_norm": 0.00018558117153588682,
      "learning_rate": 4.9814025834572126e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 883896,
      "step": 1555
    },
    {
      "epoch": 27.371681415929203,
      "grad_norm": 0.0003715149941854179,
      "learning_rate": 4.981282866230411e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 886376,
      "step": 1560
    },
    {
      "epoch": 27.460176991150444,
      "grad_norm": 9.656500333221629e-05,
      "learning_rate": 4.981162766358611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 889384,
      "step": 1565
    },
    {
      "epoch": 27.548672566371682,
      "grad_norm": 0.00013709939958062023,
      "learning_rate": 4.9810422838603316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 892344,
      "step": 1570
    },
    {
      "epoch": 27.63716814159292,
      "grad_norm": 0.0003426656185183674,
      "learning_rate": 4.9809214187541533e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 895048,
      "step": 1575
    },
    {
      "epoch": 27.72566371681416,
      "grad_norm": 0.0007989946752786636,
      "learning_rate": 4.980800171058715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 897992,
      "step": 1580
    },
    {
      "epoch": 27.8141592920354,
      "grad_norm": 0.0007163723348639905,
      "learning_rate": 4.980678540792715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 900888,
      "step": 1585
    },
    {
      "epoch": 27.902654867256636,
      "grad_norm": 0.0002287779498146847,
      "learning_rate": 4.980556527974909e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 904072,
      "step": 1590
    },
    {
      "epoch": 27.991150442477878,
      "grad_norm": 0.0002399350341875106,
      "learning_rate": 4.980434132624114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 907016,
      "step": 1595
    },
    {
      "epoch": 28.07079646017699,
      "grad_norm": 0.00018373604689259082,
      "learning_rate": 4.980311354759205e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 909320,
      "step": 1600
    },
    {
      "epoch": 28.07079646017699,
      "eval_loss": 0.34196165204048157,
      "eval_runtime": 1.0644,
      "eval_samples_per_second": 23.487,
      "eval_steps_per_second": 12.213,
      "num_input_tokens_seen": 909320,
      "step": 1600
    },
    {
      "epoch": 28.15929203539823,
      "grad_norm": 0.0001457381440559402,
      "learning_rate": 4.980188194399116e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 912536,
      "step": 1605
    },
    {
      "epoch": 28.24778761061947,
      "grad_norm": 0.00027937511913478374,
      "learning_rate": 4.9800646515628384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 915480,
      "step": 1610
    },
    {
      "epoch": 28.336283185840706,
      "grad_norm": 0.0012296867789700627,
      "learning_rate": 4.979940726269426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 918344,
      "step": 1615
    },
    {
      "epoch": 28.424778761061948,
      "grad_norm": 0.00022247666493058205,
      "learning_rate": 4.979816418537988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 920808,
      "step": 1620
    },
    {
      "epoch": 28.513274336283185,
      "grad_norm": 0.00022811858798377216,
      "learning_rate": 4.979691728387696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 923976,
      "step": 1625
    },
    {
      "epoch": 28.601769911504427,
      "grad_norm": 0.0009099822491407394,
      "learning_rate": 4.979566655837776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 926568,
      "step": 1630
    },
    {
      "epoch": 28.690265486725664,
      "grad_norm": 0.00039756763726472855,
      "learning_rate": 4.9794412009075184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 929512,
      "step": 1635
    },
    {
      "epoch": 28.778761061946902,
      "grad_norm": 8.380794315598905e-05,
      "learning_rate": 4.979315363616269e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 932424,
      "step": 1640
    },
    {
      "epoch": 28.86725663716814,
      "grad_norm": 0.00012285279808565974,
      "learning_rate": 4.979189143983434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 935496,
      "step": 1645
    },
    {
      "epoch": 28.95575221238938,
      "grad_norm": 0.00044101959792897105,
      "learning_rate": 4.979062542028478e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 938664,
      "step": 1650
    },
    {
      "epoch": 29.035398230088497,
      "grad_norm": 0.0005608491483144462,
      "learning_rate": 4.978935557770923e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 940776,
      "step": 1655
    },
    {
      "epoch": 29.123893805309734,
      "grad_norm": 0.00046104000648483634,
      "learning_rate": 4.978808191230353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 943624,
      "step": 1660
    },
    {
      "epoch": 29.212389380530972,
      "grad_norm": 0.0002687009400688112,
      "learning_rate": 4.9786804424264085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 946728,
      "step": 1665
    },
    {
      "epoch": 29.300884955752213,
      "grad_norm": 0.00018068151257466525,
      "learning_rate": 4.978552311378792e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 949880,
      "step": 1670
    },
    {
      "epoch": 29.38938053097345,
      "grad_norm": 0.00018492106755729765,
      "learning_rate": 4.978423798107261e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 952376,
      "step": 1675
    },
    {
      "epoch": 29.47787610619469,
      "grad_norm": 0.0002485496806912124,
      "learning_rate": 4.978294902631635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 955272,
      "step": 1680
    },
    {
      "epoch": 29.56637168141593,
      "grad_norm": 0.00015166541561484337,
      "learning_rate": 4.9781656249717914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 958376,
      "step": 1685
    },
    {
      "epoch": 29.654867256637168,
      "grad_norm": 0.0004005020600743592,
      "learning_rate": 4.9780359651476645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 960952,
      "step": 1690
    },
    {
      "epoch": 29.743362831858406,
      "grad_norm": 0.00048121355939656496,
      "learning_rate": 4.977905923179251e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 963752,
      "step": 1695
    },
    {
      "epoch": 29.831858407079647,
      "grad_norm": 0.0003121072659268975,
      "learning_rate": 4.977775499086606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 967032,
      "step": 1700
    },
    {
      "epoch": 29.920353982300885,
      "grad_norm": 7.087828998919576e-05,
      "learning_rate": 4.97764469288984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 969672,
      "step": 1705
    },
    {
      "epoch": 30.0,
      "grad_norm": 5.543347288039513e-05,
      "learning_rate": 4.977513504609127e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 972184,
      "step": 1710
    },
    {
      "epoch": 30.088495575221238,
      "grad_norm": 9.004897583508864e-05,
      "learning_rate": 4.9773819342646965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 974888,
      "step": 1715
    },
    {
      "epoch": 30.17699115044248,
      "grad_norm": 0.00038345353095792234,
      "learning_rate": 4.97724998187684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 977368,
      "step": 1720
    },
    {
      "epoch": 30.265486725663717,
      "grad_norm": 0.00026625749887898564,
      "learning_rate": 4.9771176474659045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 980696,
      "step": 1725
    },
    {
      "epoch": 30.353982300884955,
      "grad_norm": 0.00030353316105902195,
      "learning_rate": 4.976984931052299e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 983288,
      "step": 1730
    },
    {
      "epoch": 30.442477876106196,
      "grad_norm": 0.0001503156527178362,
      "learning_rate": 4.976851832656489e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 986520,
      "step": 1735
    },
    {
      "epoch": 30.530973451327434,
      "grad_norm": 6.17704790784046e-05,
      "learning_rate": 4.9767183522990004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 989496,
      "step": 1740
    },
    {
      "epoch": 30.61946902654867,
      "grad_norm": 0.0017050260212272406,
      "learning_rate": 4.9765844900004176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 992552,
      "step": 1745
    },
    {
      "epoch": 30.707964601769913,
      "grad_norm": 0.00010810598178068176,
      "learning_rate": 4.9764502457813834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 995448,
      "step": 1750
    },
    {
      "epoch": 30.79646017699115,
      "grad_norm": 0.00021925194596406072,
      "learning_rate": 4.9763156196626005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 998504,
      "step": 1755
    },
    {
      "epoch": 30.884955752212388,
      "grad_norm": 3.9061011193552986e-05,
      "learning_rate": 4.97618061166483e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1001352,
      "step": 1760
    },
    {
      "epoch": 30.97345132743363,
      "grad_norm": 0.00018888968043029308,
      "learning_rate": 4.9760452218088915e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1003896,
      "step": 1765
    },
    {
      "epoch": 31.053097345132745,
      "grad_norm": 6.913545803399757e-05,
      "learning_rate": 4.975909450115663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1006384,
      "step": 1770
    },
    {
      "epoch": 31.141592920353983,
      "grad_norm": 0.0002730939886532724,
      "learning_rate": 4.975773296606084e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1009600,
      "step": 1775
    },
    {
      "epoch": 31.23008849557522,
      "grad_norm": 0.0004454023437574506,
      "learning_rate": 4.97563676130115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1012736,
      "step": 1780
    },
    {
      "epoch": 31.31858407079646,
      "grad_norm": 0.00014406244736164808,
      "learning_rate": 4.9754998442219166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1015632,
      "step": 1785
    },
    {
      "epoch": 31.4070796460177,
      "grad_norm": 9.349796891910955e-05,
      "learning_rate": 4.9753625453894984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1018624,
      "step": 1790
    },
    {
      "epoch": 31.495575221238937,
      "grad_norm": 0.0002831156016327441,
      "learning_rate": 4.975224864825068e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1021152,
      "step": 1795
    },
    {
      "epoch": 31.58407079646018,
      "grad_norm": 0.00010501465294510126,
      "learning_rate": 4.9750868025498576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1023696,
      "step": 1800
    },
    {
      "epoch": 31.58407079646018,
      "eval_loss": 0.34817826747894287,
      "eval_runtime": 1.0578,
      "eval_samples_per_second": 23.634,
      "eval_steps_per_second": 12.29,
      "num_input_tokens_seen": 1023696,
      "step": 1800
    },
    {
      "epoch": 31.672566371681416,
      "grad_norm": 0.00011696917499648407,
      "learning_rate": 4.974948358585158e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1026400,
      "step": 1805
    },
    {
      "epoch": 31.761061946902654,
      "grad_norm": 0.0014729638351127505,
      "learning_rate": 4.9748095329523205e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1029040,
      "step": 1810
    },
    {
      "epoch": 31.849557522123895,
      "grad_norm": 0.0011162806767970324,
      "learning_rate": 4.974670325672752e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1032032,
      "step": 1815
    },
    {
      "epoch": 31.938053097345133,
      "grad_norm": 0.0004767157952301204,
      "learning_rate": 4.974530736767921e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1034912,
      "step": 1820
    },
    {
      "epoch": 32.017699115044245,
      "grad_norm": 0.0007481095963157713,
      "learning_rate": 4.9743907662593524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1037416,
      "step": 1825
    },
    {
      "epoch": 32.10619469026549,
      "grad_norm": 4.264988092472777e-05,
      "learning_rate": 4.974250414168633e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1040040,
      "step": 1830
    },
    {
      "epoch": 32.19469026548673,
      "grad_norm": 0.0010251669446006417,
      "learning_rate": 4.974109680517407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1043288,
      "step": 1835
    },
    {
      "epoch": 32.283185840707965,
      "grad_norm": 7.593436021124944e-05,
      "learning_rate": 4.973968565327376e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1045896,
      "step": 1840
    },
    {
      "epoch": 32.3716814159292,
      "grad_norm": 0.00017767395183909684,
      "learning_rate": 4.973827068620303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1048792,
      "step": 1845
    },
    {
      "epoch": 32.46017699115044,
      "grad_norm": 0.0003477015998214483,
      "learning_rate": 4.973685190418008e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1051352,
      "step": 1850
    },
    {
      "epoch": 32.54867256637168,
      "grad_norm": 9.30609239730984e-05,
      "learning_rate": 4.97354293074237e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1054520,
      "step": 1855
    },
    {
      "epoch": 32.63716814159292,
      "grad_norm": 0.00033159792656078935,
      "learning_rate": 4.9734002896153276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1057768,
      "step": 1860
    },
    {
      "epoch": 32.72566371681416,
      "grad_norm": 0.0002392515743849799,
      "learning_rate": 4.973257267058877e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1060936,
      "step": 1865
    },
    {
      "epoch": 32.8141592920354,
      "grad_norm": 0.0007124161929823458,
      "learning_rate": 4.973113863095076e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1063880,
      "step": 1870
    },
    {
      "epoch": 32.902654867256636,
      "grad_norm": 0.00012581142073031515,
      "learning_rate": 4.9729700777460384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1066616,
      "step": 1875
    },
    {
      "epoch": 32.991150442477874,
      "grad_norm": 0.00022746283502783626,
      "learning_rate": 4.972825911033937e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1069080,
      "step": 1880
    },
    {
      "epoch": 33.07079646017699,
      "grad_norm": 0.0006355750374495983,
      "learning_rate": 4.9726813629810056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1071712,
      "step": 1885
    },
    {
      "epoch": 33.15929203539823,
      "grad_norm": 0.0009925408521667123,
      "learning_rate": 4.9725364336095326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1074608,
      "step": 1890
    },
    {
      "epoch": 33.24778761061947,
      "grad_norm": 0.0002284862130181864,
      "learning_rate": 4.972391122941871e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1077168,
      "step": 1895
    },
    {
      "epoch": 33.336283185840706,
      "grad_norm": 0.0008172227535396814,
      "learning_rate": 4.972245431000428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1079920,
      "step": 1900
    },
    {
      "epoch": 33.424778761061944,
      "grad_norm": 0.00019004671776201576,
      "learning_rate": 4.972099357807671e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1082816,
      "step": 1905
    },
    {
      "epoch": 33.51327433628319,
      "grad_norm": 0.00023623814922757447,
      "learning_rate": 4.971952903386127e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1085600,
      "step": 1910
    },
    {
      "epoch": 33.60176991150443,
      "grad_norm": 6.412656512111425e-05,
      "learning_rate": 4.971806067758381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1088464,
      "step": 1915
    },
    {
      "epoch": 33.690265486725664,
      "grad_norm": 0.00011040906247217208,
      "learning_rate": 4.971658850947076e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1091648,
      "step": 1920
    },
    {
      "epoch": 33.7787610619469,
      "grad_norm": 0.0001085743133444339,
      "learning_rate": 4.9715112529749165e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1094080,
      "step": 1925
    },
    {
      "epoch": 33.86725663716814,
      "grad_norm": 0.00023568385222461075,
      "learning_rate": 4.9713632738646624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1097344,
      "step": 1930
    },
    {
      "epoch": 33.95575221238938,
      "grad_norm": 5.288970351102762e-05,
      "learning_rate": 4.971214913639134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1100032,
      "step": 1935
    },
    {
      "epoch": 34.0353982300885,
      "grad_norm": 0.00013071700232103467,
      "learning_rate": 4.9710661723212104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1102528,
      "step": 1940
    },
    {
      "epoch": 34.123893805309734,
      "grad_norm": 0.0011460609966889024,
      "learning_rate": 4.9709170499338295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1105648,
      "step": 1945
    },
    {
      "epoch": 34.21238938053097,
      "grad_norm": 0.0005200951127335429,
      "learning_rate": 4.9707675464999895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1108656,
      "step": 1950
    },
    {
      "epoch": 34.30088495575221,
      "grad_norm": 6.915051926625893e-05,
      "learning_rate": 4.970617662042743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1111504,
      "step": 1955
    },
    {
      "epoch": 34.389380530973455,
      "grad_norm": 0.0006193352164700627,
      "learning_rate": 4.970467396585206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1114448,
      "step": 1960
    },
    {
      "epoch": 34.47787610619469,
      "grad_norm": 0.00025149688008241355,
      "learning_rate": 4.97031675015055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1118144,
      "step": 1965
    },
    {
      "epoch": 34.56637168141593,
      "grad_norm": 5.68042705708649e-05,
      "learning_rate": 4.9701657227620075e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1120592,
      "step": 1970
    },
    {
      "epoch": 34.65486725663717,
      "grad_norm": 6.106602086219937e-05,
      "learning_rate": 4.9700143144428685e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1123024,
      "step": 1975
    },
    {
      "epoch": 34.743362831858406,
      "grad_norm": 6.480042793555185e-05,
      "learning_rate": 4.969862525216482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1126016,
      "step": 1980
    },
    {
      "epoch": 34.83185840707964,
      "grad_norm": 0.00012696548947133124,
      "learning_rate": 4.9697103551062556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1129056,
      "step": 1985
    },
    {
      "epoch": 34.92035398230089,
      "grad_norm": 0.00011158191045979038,
      "learning_rate": 4.9695578041356565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1131600,
      "step": 1990
    },
    {
      "epoch": 35.0,
      "grad_norm": 9.207201946992427e-05,
      "learning_rate": 4.969404872328209e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1134032,
      "step": 1995
    },
    {
      "epoch": 35.08849557522124,
      "grad_norm": 0.00010991047747666016,
      "learning_rate": 4.969251559707498e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1137280,
      "step": 2000
    },
    {
      "epoch": 35.08849557522124,
      "eval_loss": 0.33682557940483093,
      "eval_runtime": 1.0576,
      "eval_samples_per_second": 23.639,
      "eval_steps_per_second": 12.292,
      "num_input_tokens_seen": 1137280,
      "step": 2000
    },
    {
      "epoch": 35.176991150442475,
      "grad_norm": 0.0003175128949806094,
      "learning_rate": 4.9690978662971674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1139856,
      "step": 2005
    },
    {
      "epoch": 35.26548672566372,
      "grad_norm": 0.00015859647828619927,
      "learning_rate": 4.968943792120916e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1142800,
      "step": 2010
    },
    {
      "epoch": 35.35398230088496,
      "grad_norm": 0.00042362947715446353,
      "learning_rate": 4.9687893372025046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1145824,
      "step": 2015
    },
    {
      "epoch": 35.442477876106196,
      "grad_norm": 0.00018572174303699285,
      "learning_rate": 4.9686345015657535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1148736,
      "step": 2020
    },
    {
      "epoch": 35.530973451327434,
      "grad_norm": 8.271670958492905e-05,
      "learning_rate": 4.968479285234538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1151760,
      "step": 2025
    },
    {
      "epoch": 35.61946902654867,
      "grad_norm": 0.00019040309416595846,
      "learning_rate": 4.9683236882327974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1154496,
      "step": 2030
    },
    {
      "epoch": 35.70796460176991,
      "grad_norm": 0.00012395439262036234,
      "learning_rate": 4.968167710584526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1157088,
      "step": 2035
    },
    {
      "epoch": 35.796460176991154,
      "grad_norm": 0.00041859495104290545,
      "learning_rate": 4.968011352313775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1159824,
      "step": 2040
    },
    {
      "epoch": 35.88495575221239,
      "grad_norm": 0.0004082810482941568,
      "learning_rate": 4.967854613444659e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1162432,
      "step": 2045
    },
    {
      "epoch": 35.97345132743363,
      "grad_norm": 0.00018492669914849102,
      "learning_rate": 4.967697494001349e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1165680,
      "step": 2050
    },
    {
      "epoch": 36.05309734513274,
      "grad_norm": 0.00015748274745419621,
      "learning_rate": 4.9675399940080736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1168136,
      "step": 2055
    },
    {
      "epoch": 36.14159292035398,
      "grad_norm": 6.012044832459651e-05,
      "learning_rate": 4.9673821134891226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1171576,
      "step": 2060
    },
    {
      "epoch": 36.230088495575224,
      "grad_norm": 4.9376918468624353e-05,
      "learning_rate": 4.967223852468842e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1174360,
      "step": 2065
    },
    {
      "epoch": 36.31858407079646,
      "grad_norm": 0.0005479396204464138,
      "learning_rate": 4.967065210971639e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1177240,
      "step": 2070
    },
    {
      "epoch": 36.4070796460177,
      "grad_norm": 0.00020344123186077923,
      "learning_rate": 4.966906189021977e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1179896,
      "step": 2075
    },
    {
      "epoch": 36.49557522123894,
      "grad_norm": 0.00035436474718153477,
      "learning_rate": 4.966746786644379e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1182664,
      "step": 2080
    },
    {
      "epoch": 36.584070796460175,
      "grad_norm": 3.900430601788685e-05,
      "learning_rate": 4.966587003863429e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1185304,
      "step": 2085
    },
    {
      "epoch": 36.67256637168141,
      "grad_norm": 0.0007942054653540254,
      "learning_rate": 4.966426840703765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1188536,
      "step": 2090
    },
    {
      "epoch": 36.76106194690266,
      "grad_norm": 4.430364424479194e-05,
      "learning_rate": 4.9662662971900875e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1191304,
      "step": 2095
    },
    {
      "epoch": 36.849557522123895,
      "grad_norm": 0.00013305534957908094,
      "learning_rate": 4.9661053733471534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1193864,
      "step": 2100
    },
    {
      "epoch": 36.93805309734513,
      "grad_norm": 0.0003479511942714453,
      "learning_rate": 4.965944069199781e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1197096,
      "step": 2105
    },
    {
      "epoch": 37.017699115044245,
      "grad_norm": 0.00042743232916109264,
      "learning_rate": 4.965782384772842e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1199704,
      "step": 2110
    },
    {
      "epoch": 37.10619469026549,
      "grad_norm": 9.94155925582163e-05,
      "learning_rate": 4.9656203200912734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1203080,
      "step": 2115
    },
    {
      "epoch": 37.19469026548673,
      "grad_norm": 0.00016902509378269315,
      "learning_rate": 4.965457875180067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1205912,
      "step": 2120
    },
    {
      "epoch": 37.283185840707965,
      "grad_norm": 0.000725648773368448,
      "learning_rate": 4.9652950500642724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1208792,
      "step": 2125
    },
    {
      "epoch": 37.3716814159292,
      "grad_norm": 0.00010538724745856598,
      "learning_rate": 4.965131844769001e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1211752,
      "step": 2130
    },
    {
      "epoch": 37.46017699115044,
      "grad_norm": 0.00036774881300516427,
      "learning_rate": 4.96496825931942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1214312,
      "step": 2135
    },
    {
      "epoch": 37.54867256637168,
      "grad_norm": 8.24701928650029e-05,
      "learning_rate": 4.9648042937407566e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1216968,
      "step": 2140
    },
    {
      "epoch": 37.63716814159292,
      "grad_norm": 7.571346941404045e-05,
      "learning_rate": 4.964639948058297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1220744,
      "step": 2145
    },
    {
      "epoch": 37.72566371681416,
      "grad_norm": 8.165508916135877e-05,
      "learning_rate": 4.9644752222973846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1223352,
      "step": 2150
    },
    {
      "epoch": 37.8141592920354,
      "grad_norm": 0.00023263858747668564,
      "learning_rate": 4.964310116483422e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1226424,
      "step": 2155
    },
    {
      "epoch": 37.902654867256636,
      "grad_norm": 0.00012380874250084162,
      "learning_rate": 4.964144630641872e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1228920,
      "step": 2160
    },
    {
      "epoch": 37.991150442477874,
      "grad_norm": 0.00015458020789083093,
      "learning_rate": 4.9639787647982525e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1231672,
      "step": 2165
    },
    {
      "epoch": 38.07079646017699,
      "grad_norm": 0.000224915289436467,
      "learning_rate": 4.963812518978143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1234104,
      "step": 2170
    },
    {
      "epoch": 38.15929203539823,
      "grad_norm": 0.00015232065925374627,
      "learning_rate": 4.963645893207182e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1237080,
      "step": 2175
    },
    {
      "epoch": 38.24778761061947,
      "grad_norm": 0.00014986122550908476,
      "learning_rate": 4.963478887511063e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1240248,
      "step": 2180
    },
    {
      "epoch": 38.336283185840706,
      "grad_norm": 0.0004464318626560271,
      "learning_rate": 4.963311501915542e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1242936,
      "step": 2185
    },
    {
      "epoch": 38.424778761061944,
      "grad_norm": 8.75714686117135e-05,
      "learning_rate": 4.963143736446432e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1245752,
      "step": 2190
    },
    {
      "epoch": 38.51327433628319,
      "grad_norm": 4.308891948312521e-05,
      "learning_rate": 4.962975591129603e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1248152,
      "step": 2195
    },
    {
      "epoch": 38.60176991150443,
      "grad_norm": 6.6621825681068e-05,
      "learning_rate": 4.962807065990986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1251592,
      "step": 2200
    },
    {
      "epoch": 38.60176991150443,
      "eval_loss": 0.34983158111572266,
      "eval_runtime": 1.0595,
      "eval_samples_per_second": 23.597,
      "eval_steps_per_second": 12.27,
      "num_input_tokens_seen": 1251592,
      "step": 2200
    },
    {
      "epoch": 38.690265486725664,
      "grad_norm": 0.00013585940178018063,
      "learning_rate": 4.9626381610565714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1254312,
      "step": 2205
    },
    {
      "epoch": 38.7787610619469,
      "grad_norm": 0.0004515045147854835,
      "learning_rate": 4.9624688763524043e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1257096,
      "step": 2210
    },
    {
      "epoch": 38.86725663716814,
      "grad_norm": 0.00036725657992064953,
      "learning_rate": 4.962299211904591e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1260184,
      "step": 2215
    },
    {
      "epoch": 38.95575221238938,
      "grad_norm": 0.00018853717483580112,
      "learning_rate": 4.962129167739296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1262888,
      "step": 2220
    },
    {
      "epoch": 39.0353982300885,
      "grad_norm": 0.000187097248272039,
      "learning_rate": 4.961958743882742e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1265464,
      "step": 2225
    },
    {
      "epoch": 39.123893805309734,
      "grad_norm": 0.0005766918184235692,
      "learning_rate": 4.961787940361211e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1268088,
      "step": 2230
    },
    {
      "epoch": 39.21238938053097,
      "grad_norm": 7.629850006196648e-05,
      "learning_rate": 4.961616757201043e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1271352,
      "step": 2235
    },
    {
      "epoch": 39.30088495575221,
      "grad_norm": 0.00018880373681895435,
      "learning_rate": 4.961445194428637e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1273736,
      "step": 2240
    },
    {
      "epoch": 39.389380530973455,
      "grad_norm": 0.00044931730371899903,
      "learning_rate": 4.9612732520704486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1276600,
      "step": 2245
    },
    {
      "epoch": 39.47787610619469,
      "grad_norm": 0.00020239490550011396,
      "learning_rate": 4.961100930152994e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1279800,
      "step": 2250
    },
    {
      "epoch": 39.56637168141593,
      "grad_norm": 5.904089266550727e-05,
      "learning_rate": 4.960928228702849e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1283144,
      "step": 2255
    },
    {
      "epoch": 39.65486725663717,
      "grad_norm": 0.00016151352610904723,
      "learning_rate": 4.960755147746645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1285816,
      "step": 2260
    },
    {
      "epoch": 39.743362831858406,
      "grad_norm": 0.0005621831514872611,
      "learning_rate": 4.9605816873110736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1288856,
      "step": 2265
    },
    {
      "epoch": 39.83185840707964,
      "grad_norm": 0.00038023083470761776,
      "learning_rate": 4.960407847422883e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1291544,
      "step": 2270
    },
    {
      "epoch": 39.92035398230089,
      "grad_norm": 4.867268944508396e-05,
      "learning_rate": 4.960233628108885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1294296,
      "step": 2275
    },
    {
      "epoch": 40.0,
      "grad_norm": 4.6419936552410945e-05,
      "learning_rate": 4.960059029395942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1296696,
      "step": 2280
    },
    {
      "epoch": 40.08849557522124,
      "grad_norm": 0.0001341633906122297,
      "learning_rate": 4.959884051310983e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1299640,
      "step": 2285
    },
    {
      "epoch": 40.176991150442475,
      "grad_norm": 7.908207044238225e-05,
      "learning_rate": 4.959708693880991e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1302520,
      "step": 2290
    },
    {
      "epoch": 40.26548672566372,
      "grad_norm": 0.0002487853926140815,
      "learning_rate": 4.9595329571330074e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1305784,
      "step": 2295
    },
    {
      "epoch": 40.35398230088496,
      "grad_norm": 0.00010209906758973375,
      "learning_rate": 4.9593568410941326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1309128,
      "step": 2300
    },
    {
      "epoch": 40.442477876106196,
      "grad_norm": 5.9376794524723664e-05,
      "learning_rate": 4.959180345791528e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1311592,
      "step": 2305
    },
    {
      "epoch": 40.530973451327434,
      "grad_norm": 3.720061431522481e-05,
      "learning_rate": 4.9590034712524086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1314104,
      "step": 2310
    },
    {
      "epoch": 40.61946902654867,
      "grad_norm": 0.0001543039979878813,
      "learning_rate": 4.958826217504053e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1316776,
      "step": 2315
    },
    {
      "epoch": 40.70796460176991,
      "grad_norm": 0.0003066086210310459,
      "learning_rate": 4.958648584573795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1319400,
      "step": 2320
    },
    {
      "epoch": 40.796460176991154,
      "grad_norm": 9.64335777098313e-05,
      "learning_rate": 4.958470572489028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1322408,
      "step": 2325
    },
    {
      "epoch": 40.88495575221239,
      "grad_norm": 0.00013059726916253567,
      "learning_rate": 4.958292181277203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1325256,
      "step": 2330
    },
    {
      "epoch": 40.97345132743363,
      "grad_norm": 0.00011612750677159056,
      "learning_rate": 4.958113410965832e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1327976,
      "step": 2335
    },
    {
      "epoch": 41.05309734513274,
      "grad_norm": 4.5467411837307736e-05,
      "learning_rate": 4.957934261582481e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1330648,
      "step": 2340
    },
    {
      "epoch": 41.14159292035398,
      "grad_norm": 0.000672188529279083,
      "learning_rate": 4.95775473315478e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1334088,
      "step": 2345
    },
    {
      "epoch": 41.230088495575224,
      "grad_norm": 0.0002594262477941811,
      "learning_rate": 4.9575748257104124e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1337000,
      "step": 2350
    },
    {
      "epoch": 41.31858407079646,
      "grad_norm": 0.00023683729523327202,
      "learning_rate": 4.9573945392771224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1339976,
      "step": 2355
    },
    {
      "epoch": 41.4070796460177,
      "grad_norm": 0.0002656909346114844,
      "learning_rate": 4.9572138738827134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1342792,
      "step": 2360
    },
    {
      "epoch": 41.49557522123894,
      "grad_norm": 0.00014086582814343274,
      "learning_rate": 4.957032829555046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1345480,
      "step": 2365
    },
    {
      "epoch": 41.584070796460175,
      "grad_norm": 0.0008349600830115378,
      "learning_rate": 4.956851406322039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1348472,
      "step": 2370
    },
    {
      "epoch": 41.67256637168141,
      "grad_norm": 0.000732863787561655,
      "learning_rate": 4.9566696042116704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1351432,
      "step": 2375
    },
    {
      "epoch": 41.76106194690266,
      "grad_norm": 0.0001576658250996843,
      "learning_rate": 4.9564874232519766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1354168,
      "step": 2380
    },
    {
      "epoch": 41.849557522123895,
      "grad_norm": 7.971344894031063e-05,
      "learning_rate": 4.9563048634710516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1356952,
      "step": 2385
    },
    {
      "epoch": 41.93805309734513,
      "grad_norm": 5.300787597661838e-05,
      "learning_rate": 4.956121924897049e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1359304,
      "step": 2390
    },
    {
      "epoch": 42.017699115044245,
      "grad_norm": 1.811354377423413e-05,
      "learning_rate": 4.955938607558181e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1361752,
      "step": 2395
    },
    {
      "epoch": 42.10619469026549,
      "grad_norm": 0.0005139414570294321,
      "learning_rate": 4.955754911482715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1364312,
      "step": 2400
    },
    {
      "epoch": 42.10619469026549,
      "eval_loss": 0.34935858845710754,
      "eval_runtime": 1.0711,
      "eval_samples_per_second": 23.341,
      "eval_steps_per_second": 12.137,
      "num_input_tokens_seen": 1364312,
      "step": 2400
    },
    {
      "epoch": 42.19469026548673,
      "grad_norm": 0.0004127823340240866,
      "learning_rate": 4.9555708366989804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1367144,
      "step": 2405
    },
    {
      "epoch": 42.283185840707965,
      "grad_norm": 7.618355448357761e-05,
      "learning_rate": 4.9553863832353655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1369864,
      "step": 2410
    },
    {
      "epoch": 42.3716814159292,
      "grad_norm": 0.00010405355715192854,
      "learning_rate": 4.955201551120313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1373304,
      "step": 2415
    },
    {
      "epoch": 42.46017699115044,
      "grad_norm": 9.86897575785406e-05,
      "learning_rate": 4.955016340382328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1376168,
      "step": 2420
    },
    {
      "epoch": 42.54867256637168,
      "grad_norm": 0.00014975882368162274,
      "learning_rate": 4.954830751049972e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1379080,
      "step": 2425
    },
    {
      "epoch": 42.63716814159292,
      "grad_norm": 8.347063703695312e-05,
      "learning_rate": 4.954644783151864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1382088,
      "step": 2430
    },
    {
      "epoch": 42.72566371681416,
      "grad_norm": 0.00035207445034757257,
      "learning_rate": 4.954458436716684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1384600,
      "step": 2435
    },
    {
      "epoch": 42.8141592920354,
      "grad_norm": 0.00019096385221928358,
      "learning_rate": 4.954271711773168e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1387128,
      "step": 2440
    },
    {
      "epoch": 42.902654867256636,
      "grad_norm": 7.878107135184109e-05,
      "learning_rate": 4.9540846083501115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1389896,
      "step": 2445
    },
    {
      "epoch": 42.991150442477874,
      "grad_norm": 0.00010926426330115646,
      "learning_rate": 4.953897126476369e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1393192,
      "step": 2450
    },
    {
      "epoch": 43.07079646017699,
      "grad_norm": 1.5719959264970385e-05,
      "learning_rate": 4.9537092661808514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1395600,
      "step": 2455
    },
    {
      "epoch": 43.15929203539823,
      "grad_norm": 0.00025073919096030295,
      "learning_rate": 4.9535210274925306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1398944,
      "step": 2460
    },
    {
      "epoch": 43.24778761061947,
      "grad_norm": 4.45429323008284e-05,
      "learning_rate": 4.953332410440435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1402160,
      "step": 2465
    },
    {
      "epoch": 43.336283185840706,
      "grad_norm": 0.0001331576640950516,
      "learning_rate": 4.9531434150536496e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1404784,
      "step": 2470
    },
    {
      "epoch": 43.424778761061944,
      "grad_norm": 9.848157787928358e-05,
      "learning_rate": 4.952954041361322e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1407344,
      "step": 2475
    },
    {
      "epoch": 43.51327433628319,
      "grad_norm": 0.00017651003145147115,
      "learning_rate": 4.952764289392655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1410656,
      "step": 2480
    },
    {
      "epoch": 43.60176991150443,
      "grad_norm": 0.00018139620078727603,
      "learning_rate": 4.952574159176912e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1413344,
      "step": 2485
    },
    {
      "epoch": 43.690265486725664,
      "grad_norm": 0.0003804354346357286,
      "learning_rate": 4.952383650743413e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1415904,
      "step": 2490
    },
    {
      "epoch": 43.7787610619469,
      "grad_norm": 3.744133573491126e-05,
      "learning_rate": 4.952192764121536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1418384,
      "step": 2495
    },
    {
      "epoch": 43.86725663716814,
      "grad_norm": 3.131960693281144e-05,
      "learning_rate": 4.9520014993407185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1421328,
      "step": 2500
    },
    {
      "epoch": 43.95575221238938,
      "grad_norm": 0.0002673002309165895,
      "learning_rate": 4.951809856430456e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1424304,
      "step": 2505
    },
    {
      "epoch": 44.0353982300885,
      "grad_norm": 6.77849748171866e-05,
      "learning_rate": 4.951617835420303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1426608,
      "step": 2510
    },
    {
      "epoch": 44.123893805309734,
      "grad_norm": 0.00019121421792078763,
      "learning_rate": 4.951425436339869e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1429392,
      "step": 2515
    },
    {
      "epoch": 44.21238938053097,
      "grad_norm": 4.9445377953816205e-05,
      "learning_rate": 4.9512326592188274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1432464,
      "step": 2520
    },
    {
      "epoch": 44.30088495575221,
      "grad_norm": 0.0003611265274230391,
      "learning_rate": 4.9510395040869054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1435200,
      "step": 2525
    },
    {
      "epoch": 44.389380530973455,
      "grad_norm": 6.96985371178016e-05,
      "learning_rate": 4.9508459709738905e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1438448,
      "step": 2530
    },
    {
      "epoch": 44.47787610619469,
      "grad_norm": 0.00020978836982976645,
      "learning_rate": 4.950652059909627e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1440976,
      "step": 2535
    },
    {
      "epoch": 44.56637168141593,
      "grad_norm": 0.00019709551997948438,
      "learning_rate": 4.95045777092402e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1444160,
      "step": 2540
    },
    {
      "epoch": 44.65486725663717,
      "grad_norm": 3.369479964021593e-05,
      "learning_rate": 4.950263104047031e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1446944,
      "step": 2545
    },
    {
      "epoch": 44.743362831858406,
      "grad_norm": 0.002129932399839163,
      "learning_rate": 4.9500680593086775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1450336,
      "step": 2550
    },
    {
      "epoch": 44.83185840707964,
      "grad_norm": 0.00012764347775373608,
      "learning_rate": 4.94987263673904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1453328,
      "step": 2555
    },
    {
      "epoch": 44.92035398230089,
      "grad_norm": 8.575858373660594e-05,
      "learning_rate": 4.949676836368256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1456208,
      "step": 2560
    },
    {
      "epoch": 45.0,
      "grad_norm": 0.0002012513723457232,
      "learning_rate": 4.949480658226518e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1458448,
      "step": 2565
    },
    {
      "epoch": 45.08849557522124,
      "grad_norm": 0.0001290444197366014,
      "learning_rate": 4.949284102344082e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1461616,
      "step": 2570
    },
    {
      "epoch": 45.176991150442475,
      "grad_norm": 4.0040242311079055e-05,
      "learning_rate": 4.9490871687512565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1464352,
      "step": 2575
    },
    {
      "epoch": 45.26548672566372,
      "grad_norm": 0.00045553743257187307,
      "learning_rate": 4.948889857478413e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1467232,
      "step": 2580
    },
    {
      "epoch": 45.35398230088496,
      "grad_norm": 0.0001255954266525805,
      "learning_rate": 4.948692168555978e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1470144,
      "step": 2585
    },
    {
      "epoch": 45.442477876106196,
      "grad_norm": 9.036884148372337e-05,
      "learning_rate": 4.94849410201444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1473120,
      "step": 2590
    },
    {
      "epoch": 45.530973451327434,
      "grad_norm": 0.0003102921473328024,
      "learning_rate": 4.948295657884341e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1475856,
      "step": 2595
    },
    {
      "epoch": 45.61946902654867,
      "grad_norm": 2.753458102233708e-05,
      "learning_rate": 4.9480968361962835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1478704,
      "step": 2600
    },
    {
      "epoch": 45.61946902654867,
      "eval_loss": 0.3589751124382019,
      "eval_runtime": 1.0624,
      "eval_samples_per_second": 23.531,
      "eval_steps_per_second": 12.236,
      "num_input_tokens_seen": 1478704,
      "step": 2600
    },
    {
      "epoch": 45.70796460176991,
      "grad_norm": 8.963972504716367e-05,
      "learning_rate": 4.9478976369809305e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1481120,
      "step": 2605
    },
    {
      "epoch": 45.796460176991154,
      "grad_norm": 0.00014809786807745695,
      "learning_rate": 4.947698060268999e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1484320,
      "step": 2610
    },
    {
      "epoch": 45.88495575221239,
      "grad_norm": 0.00027220527408644557,
      "learning_rate": 4.9474981060912665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1487568,
      "step": 2615
    },
    {
      "epoch": 45.97345132743363,
      "grad_norm": 3.048935286642518e-05,
      "learning_rate": 4.94729777447857e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1490304,
      "step": 2620
    },
    {
      "epoch": 46.05309734513274,
      "grad_norm": 7.962770905578509e-05,
      "learning_rate": 4.947097065461801e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1492544,
      "step": 2625
    },
    {
      "epoch": 46.14159292035398,
      "grad_norm": 8.126032480504364e-05,
      "learning_rate": 4.9468959790719125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1495632,
      "step": 2630
    },
    {
      "epoch": 46.230088495575224,
      "grad_norm": 0.0011224745539948344,
      "learning_rate": 4.9466945153399146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1498528,
      "step": 2635
    },
    {
      "epoch": 46.31858407079646,
      "grad_norm": 6.731283065164462e-05,
      "learning_rate": 4.9464926742968755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1501648,
      "step": 2640
    },
    {
      "epoch": 46.4070796460177,
      "grad_norm": 7.876844756538048e-05,
      "learning_rate": 4.946290455973921e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1504352,
      "step": 2645
    },
    {
      "epoch": 46.49557522123894,
      "grad_norm": 3.9954193198354915e-05,
      "learning_rate": 4.9460878604022365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1507360,
      "step": 2650
    },
    {
      "epoch": 46.584070796460175,
      "grad_norm": 0.0002688575186766684,
      "learning_rate": 4.945884887613065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1510032,
      "step": 2655
    },
    {
      "epoch": 46.67256637168141,
      "grad_norm": 4.984966653864831e-05,
      "learning_rate": 4.9456815376377055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1513136,
      "step": 2660
    },
    {
      "epoch": 46.76106194690266,
      "grad_norm": 0.0001184520879178308,
      "learning_rate": 4.9454778105075195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1515520,
      "step": 2665
    },
    {
      "epoch": 46.849557522123895,
      "grad_norm": 0.00016530165157746524,
      "learning_rate": 4.945273706253924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1518544,
      "step": 2670
    },
    {
      "epoch": 46.93805309734513,
      "grad_norm": 2.5189849111484364e-05,
      "learning_rate": 4.9450692249083925e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1521280,
      "step": 2675
    },
    {
      "epoch": 47.017699115044245,
      "grad_norm": 8.866257121553645e-05,
      "learning_rate": 4.9448643665024605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1523616,
      "step": 2680
    },
    {
      "epoch": 47.10619469026549,
      "grad_norm": 0.00019412091933190823,
      "learning_rate": 4.944659131067719e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1526816,
      "step": 2685
    },
    {
      "epoch": 47.19469026548673,
      "grad_norm": 3.241914964746684e-05,
      "learning_rate": 4.944453518635818e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1529776,
      "step": 2690
    },
    {
      "epoch": 47.283185840707965,
      "grad_norm": 1.9888131646439433e-05,
      "learning_rate": 4.944247529238465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1532624,
      "step": 2695
    },
    {
      "epoch": 47.3716814159292,
      "grad_norm": 4.390193134895526e-05,
      "learning_rate": 4.944041162907427e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1535440,
      "step": 2700
    },
    {
      "epoch": 47.46017699115044,
      "grad_norm": 7.359178562182933e-05,
      "learning_rate": 4.943834419674529e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1538336,
      "step": 2705
    },
    {
      "epoch": 47.54867256637168,
      "grad_norm": 0.00012109276576666161,
      "learning_rate": 4.9436272995716506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1541024,
      "step": 2710
    },
    {
      "epoch": 47.63716814159292,
      "grad_norm": 0.0001384573697578162,
      "learning_rate": 4.943419802630735e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1544160,
      "step": 2715
    },
    {
      "epoch": 47.72566371681416,
      "grad_norm": 0.00010374926932854578,
      "learning_rate": 4.94321192888378e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1546880,
      "step": 2720
    },
    {
      "epoch": 47.8141592920354,
      "grad_norm": 0.0002103819279000163,
      "learning_rate": 4.943003678362842e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1549216,
      "step": 2725
    },
    {
      "epoch": 47.902654867256636,
      "grad_norm": 0.00028063502395525575,
      "learning_rate": 4.942795051100036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1551648,
      "step": 2730
    },
    {
      "epoch": 47.991150442477874,
      "grad_norm": 3.99855307477992e-05,
      "learning_rate": 4.942586047127536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1554752,
      "step": 2735
    },
    {
      "epoch": 48.07079646017699,
      "grad_norm": 6.00901257712394e-05,
      "learning_rate": 4.942376666477571e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1557288,
      "step": 2740
    },
    {
      "epoch": 48.15929203539823,
      "grad_norm": 4.76755085401237e-05,
      "learning_rate": 4.9421669091824304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1560408,
      "step": 2745
    },
    {
      "epoch": 48.24778761061947,
      "grad_norm": 0.00010590557212708518,
      "learning_rate": 4.9419567752744634e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1563160,
      "step": 2750
    },
    {
      "epoch": 48.336283185840706,
      "grad_norm": 4.839613757212646e-05,
      "learning_rate": 4.941746264786074e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1565896,
      "step": 2755
    },
    {
      "epoch": 48.424778761061944,
      "grad_norm": 6.852606020402163e-05,
      "learning_rate": 4.9415353777497254e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1568776,
      "step": 2760
    },
    {
      "epoch": 48.51327433628319,
      "grad_norm": 0.00030728327692486346,
      "learning_rate": 4.9413241141979394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1571736,
      "step": 2765
    },
    {
      "epoch": 48.60176991150443,
      "grad_norm": 0.0001020580020849593,
      "learning_rate": 4.9411124741632956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1575016,
      "step": 2770
    },
    {
      "epoch": 48.690265486725664,
      "grad_norm": 0.0008348529227077961,
      "learning_rate": 4.940900457678431e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1577448,
      "step": 2775
    },
    {
      "epoch": 48.7787610619469,
      "grad_norm": 0.00010069608833873644,
      "learning_rate": 4.9406880647760425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1580056,
      "step": 2780
    },
    {
      "epoch": 48.86725663716814,
      "grad_norm": 0.0001935552863869816,
      "learning_rate": 4.9404752954888824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1583096,
      "step": 2785
    },
    {
      "epoch": 48.95575221238938,
      "grad_norm": 0.00011437126522650942,
      "learning_rate": 4.940262149849762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1585864,
      "step": 2790
    },
    {
      "epoch": 49.0353982300885,
      "grad_norm": 0.00032133125932887197,
      "learning_rate": 4.9400486278915526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1588512,
      "step": 2795
    },
    {
      "epoch": 49.123893805309734,
      "grad_norm": 0.0002695470757316798,
      "learning_rate": 4.939834729647181e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1591424,
      "step": 2800
    },
    {
      "epoch": 49.123893805309734,
      "eval_loss": 0.36007025837898254,
      "eval_runtime": 1.057,
      "eval_samples_per_second": 23.651,
      "eval_steps_per_second": 12.299,
      "num_input_tokens_seen": 1591424,
      "step": 2800
    },
    {
      "epoch": 49.21238938053097,
      "grad_norm": 7.507876580348238e-05,
      "learning_rate": 4.9396204551496326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1593840,
      "step": 2805
    },
    {
      "epoch": 49.30088495575221,
      "grad_norm": 0.0001835319126257673,
      "learning_rate": 4.939405804431952e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1597232,
      "step": 2810
    },
    {
      "epoch": 49.389380530973455,
      "grad_norm": 0.00013375883281696588,
      "learning_rate": 4.9391907775272414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1600144,
      "step": 2815
    },
    {
      "epoch": 49.47787610619469,
      "grad_norm": 2.5947403628379107e-05,
      "learning_rate": 4.9389753744686604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1602800,
      "step": 2820
    },
    {
      "epoch": 49.56637168141593,
      "grad_norm": 4.369993257569149e-05,
      "learning_rate": 4.938759595289426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1605872,
      "step": 2825
    },
    {
      "epoch": 49.65486725663717,
      "grad_norm": 7.899948832346126e-05,
      "learning_rate": 4.938543440022815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1608768,
      "step": 2830
    },
    {
      "epoch": 49.743362831858406,
      "grad_norm": 2.166207559639588e-05,
      "learning_rate": 4.938326908702161e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1611488,
      "step": 2835
    },
    {
      "epoch": 49.83185840707964,
      "grad_norm": 7.60223701945506e-05,
      "learning_rate": 4.9381100013608554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1614448,
      "step": 2840
    },
    {
      "epoch": 49.92035398230089,
      "grad_norm": 0.0001025620658765547,
      "learning_rate": 4.9378927180323485e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1616784,
      "step": 2845
    },
    {
      "epoch": 50.0,
      "grad_norm": 7.667194586247206e-05,
      "learning_rate": 4.937675058750148e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1619360,
      "step": 2850
    },
    {
      "epoch": 50.08849557522124,
      "grad_norm": 0.0012060635490342975,
      "learning_rate": 4.937457023547819e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1622448,
      "step": 2855
    },
    {
      "epoch": 50.176991150442475,
      "grad_norm": 7.039837510092184e-05,
      "learning_rate": 4.9372386124589876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1625536,
      "step": 2860
    },
    {
      "epoch": 50.26548672566372,
      "grad_norm": 3.341645424370654e-05,
      "learning_rate": 4.937019825517333e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1628176,
      "step": 2865
    },
    {
      "epoch": 50.35398230088496,
      "grad_norm": 0.00016810267698019743,
      "learning_rate": 4.9368006627565954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1630640,
      "step": 2870
    },
    {
      "epoch": 50.442477876106196,
      "grad_norm": 2.0561175915645435e-05,
      "learning_rate": 4.936581124210573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1633072,
      "step": 2875
    },
    {
      "epoch": 50.530973451327434,
      "grad_norm": 9.739668166730553e-05,
      "learning_rate": 4.9363612099131216e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1635664,
      "step": 2880
    },
    {
      "epoch": 50.61946902654867,
      "grad_norm": 5.8754143537953496e-05,
      "learning_rate": 4.936140919898155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1638816,
      "step": 2885
    },
    {
      "epoch": 50.70796460176991,
      "grad_norm": 4.313279714551754e-05,
      "learning_rate": 4.9359202541996426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1641968,
      "step": 2890
    },
    {
      "epoch": 50.796460176991154,
      "grad_norm": 4.687581895268522e-05,
      "learning_rate": 4.935699212851616e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1644960,
      "step": 2895
    },
    {
      "epoch": 50.88495575221239,
      "grad_norm": 0.00032100066891871393,
      "learning_rate": 4.935477795888162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1648272,
      "step": 2900
    },
    {
      "epoch": 50.97345132743363,
      "grad_norm": 0.0002436418435536325,
      "learning_rate": 4.935256003343426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1651152,
      "step": 2905
    },
    {
      "epoch": 51.05309734513274,
      "grad_norm": 2.7731361115002073e-05,
      "learning_rate": 4.93503383525161e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1653504,
      "step": 2910
    },
    {
      "epoch": 51.14159292035398,
      "grad_norm": 4.0662878745933995e-05,
      "learning_rate": 4.934811291646977e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1656624,
      "step": 2915
    },
    {
      "epoch": 51.230088495575224,
      "grad_norm": 0.0002460065297782421,
      "learning_rate": 4.934588372563845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1659424,
      "step": 2920
    },
    {
      "epoch": 51.31858407079646,
      "grad_norm": 9.938023140421137e-05,
      "learning_rate": 4.93436507803659e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1662528,
      "step": 2925
    },
    {
      "epoch": 51.4070796460177,
      "grad_norm": 3.853082307614386e-05,
      "learning_rate": 4.934141408099649e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1665632,
      "step": 2930
    },
    {
      "epoch": 51.49557522123894,
      "grad_norm": 2.0529683752101846e-05,
      "learning_rate": 4.9339173627875135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1668544,
      "step": 2935
    },
    {
      "epoch": 51.584070796460175,
      "grad_norm": 2.8431215469026938e-05,
      "learning_rate": 4.9336929421347335e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1671184,
      "step": 2940
    },
    {
      "epoch": 51.67256637168141,
      "grad_norm": 5.4487434681504965e-05,
      "learning_rate": 4.933468146175918e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1674320,
      "step": 2945
    },
    {
      "epoch": 51.76106194690266,
      "grad_norm": 9.136016160482541e-05,
      "learning_rate": 4.933242974945734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1677216,
      "step": 2950
    },
    {
      "epoch": 51.849557522123895,
      "grad_norm": 3.146239396301098e-05,
      "learning_rate": 4.933017428478906e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1679968,
      "step": 2955
    },
    {
      "epoch": 51.93805309734513,
      "grad_norm": 0.0019273534417152405,
      "learning_rate": 4.932791506810214e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1682480,
      "step": 2960
    },
    {
      "epoch": 52.017699115044245,
      "grad_norm": 0.0001030058047035709,
      "learning_rate": 4.932565209974499e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1685352,
      "step": 2965
    },
    {
      "epoch": 52.10619469026549,
      "grad_norm": 2.3151935238274746e-05,
      "learning_rate": 4.93233853800666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1687784,
      "step": 2970
    },
    {
      "epoch": 52.19469026548673,
      "grad_norm": 4.268770135240629e-05,
      "learning_rate": 4.932111490941651e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1691416,
      "step": 2975
    },
    {
      "epoch": 52.283185840707965,
      "grad_norm": 0.00029012307641096413,
      "learning_rate": 4.9318840688144876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1694344,
      "step": 2980
    },
    {
      "epoch": 52.3716814159292,
      "grad_norm": 7.11620959918946e-05,
      "learning_rate": 4.9316562716602387e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1696840,
      "step": 2985
    },
    {
      "epoch": 52.46017699115044,
      "grad_norm": 6.760319956811145e-05,
      "learning_rate": 4.9314280995140346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1699304,
      "step": 2990
    },
    {
      "epoch": 52.54867256637168,
      "grad_norm": 0.000147602753713727,
      "learning_rate": 4.931199552411063e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1701880,
      "step": 2995
    },
    {
      "epoch": 52.63716814159292,
      "grad_norm": 0.00010341807501390576,
      "learning_rate": 4.930970630386568e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1705000,
      "step": 3000
    },
    {
      "epoch": 52.63716814159292,
      "eval_loss": 0.35832923650741577,
      "eval_runtime": 1.0602,
      "eval_samples_per_second": 23.58,
      "eval_steps_per_second": 12.261,
      "num_input_tokens_seen": 1705000,
      "step": 3000
    },
    {
      "epoch": 52.72566371681416,
      "grad_norm": 0.0001000025513349101,
      "learning_rate": 4.9307413334758524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1707896,
      "step": 3005
    },
    {
      "epoch": 52.8141592920354,
      "grad_norm": 3.302075492683798e-05,
      "learning_rate": 4.930511661714276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1710840,
      "step": 3010
    },
    {
      "epoch": 52.902654867256636,
      "grad_norm": 0.00031587955891154706,
      "learning_rate": 4.9302816151372576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1713480,
      "step": 3015
    },
    {
      "epoch": 52.991150442477874,
      "grad_norm": 2.1422487407107837e-05,
      "learning_rate": 4.930051193780274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1716696,
      "step": 3020
    },
    {
      "epoch": 53.07079646017699,
      "grad_norm": 2.228464109066408e-05,
      "learning_rate": 4.929820397678858e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1719344,
      "step": 3025
    },
    {
      "epoch": 53.15929203539823,
      "grad_norm": 0.000982264638878405,
      "learning_rate": 4.9295892268686015e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1722352,
      "step": 3030
    },
    {
      "epoch": 53.24778761061947,
      "grad_norm": 2.71742155746324e-05,
      "learning_rate": 4.9293576813851536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1725312,
      "step": 3035
    },
    {
      "epoch": 53.336283185840706,
      "grad_norm": 0.00026019191136583686,
      "learning_rate": 4.929125761264223e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1728208,
      "step": 3040
    },
    {
      "epoch": 53.424778761061944,
      "grad_norm": 9.145139483734965e-05,
      "learning_rate": 4.928893466541573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1731056,
      "step": 3045
    },
    {
      "epoch": 53.51327433628319,
      "grad_norm": 7.490388088626787e-05,
      "learning_rate": 4.928660797253027e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1734304,
      "step": 3050
    },
    {
      "epoch": 53.60176991150443,
      "grad_norm": 7.965043187141418e-05,
      "learning_rate": 4.928427753434467e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1737168,
      "step": 3055
    },
    {
      "epoch": 53.690265486725664,
      "grad_norm": 7.99868575995788e-05,
      "learning_rate": 4.9281943351218286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1740048,
      "step": 3060
    },
    {
      "epoch": 53.7787610619469,
      "grad_norm": 7.36413276172243e-05,
      "learning_rate": 4.9279605423511095e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1742960,
      "step": 3065
    },
    {
      "epoch": 53.86725663716814,
      "grad_norm": 1.6911308193812147e-05,
      "learning_rate": 4.927726375158363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1745376,
      "step": 3070
    },
    {
      "epoch": 53.95575221238938,
      "grad_norm": 2.070706250378862e-05,
      "learning_rate": 4.9274918335797004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1748160,
      "step": 3075
    },
    {
      "epoch": 54.0353982300885,
      "grad_norm": 8.099222031887621e-05,
      "learning_rate": 4.927256917651292e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1750480,
      "step": 3080
    },
    {
      "epoch": 54.123893805309734,
      "grad_norm": 6.904276233399287e-05,
      "learning_rate": 4.927021627409364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1753504,
      "step": 3085
    },
    {
      "epoch": 54.21238938053097,
      "grad_norm": 1.3334944924281444e-05,
      "learning_rate": 4.9267859628902005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1756128,
      "step": 3090
    },
    {
      "epoch": 54.30088495575221,
      "grad_norm": 0.0001534375041956082,
      "learning_rate": 4.9265499241301454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1758800,
      "step": 3095
    },
    {
      "epoch": 54.389380530973455,
      "grad_norm": 2.104890336340759e-05,
      "learning_rate": 4.926313511165598e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1761888,
      "step": 3100
    },
    {
      "epoch": 54.47787610619469,
      "grad_norm": 0.00022852334950584918,
      "learning_rate": 4.926076724033016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1765200,
      "step": 3105
    },
    {
      "epoch": 54.56637168141593,
      "grad_norm": 0.00013317303091753274,
      "learning_rate": 4.9258395627689146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1767712,
      "step": 3110
    },
    {
      "epoch": 54.65486725663717,
      "grad_norm": 4.975281626684591e-05,
      "learning_rate": 4.925602027409868e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1769920,
      "step": 3115
    },
    {
      "epoch": 54.743362831858406,
      "grad_norm": 4.553399048745632e-05,
      "learning_rate": 4.925364117992507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1773312,
      "step": 3120
    },
    {
      "epoch": 54.83185840707964,
      "grad_norm": 0.0001455327874282375,
      "learning_rate": 4.92512583455352e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1776128,
      "step": 3125
    },
    {
      "epoch": 54.92035398230089,
      "grad_norm": 0.00037552069989033043,
      "learning_rate": 4.9248871771296536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1779088,
      "step": 3130
    },
    {
      "epoch": 55.0,
      "grad_norm": 1.958963730430696e-05,
      "learning_rate": 4.924648145757711e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1781944,
      "step": 3135
    },
    {
      "epoch": 55.08849557522124,
      "grad_norm": 7.074280438246205e-05,
      "learning_rate": 4.924408740474554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1784568,
      "step": 3140
    },
    {
      "epoch": 55.176991150442475,
      "grad_norm": 6.483389006461948e-05,
      "learning_rate": 4.924168961317103e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1787560,
      "step": 3145
    },
    {
      "epoch": 55.26548672566372,
      "grad_norm": 4.3193387682549655e-05,
      "learning_rate": 4.9239288083223334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1790824,
      "step": 3150
    },
    {
      "epoch": 55.35398230088496,
      "grad_norm": 3.102220216533169e-05,
      "learning_rate": 4.9236882815272803e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1793176,
      "step": 3155
    },
    {
      "epoch": 55.442477876106196,
      "grad_norm": 0.00014426041161641479,
      "learning_rate": 4.9234473809690365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1796008,
      "step": 3160
    },
    {
      "epoch": 55.530973451327434,
      "grad_norm": 3.648840720416047e-05,
      "learning_rate": 4.923206106684752e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1799096,
      "step": 3165
    },
    {
      "epoch": 55.61946902654867,
      "grad_norm": 2.101895552186761e-05,
      "learning_rate": 4.922964458711634e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1802104,
      "step": 3170
    },
    {
      "epoch": 55.70796460176991,
      "grad_norm": 7.525499677285552e-05,
      "learning_rate": 4.9227224370869474e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1804808,
      "step": 3175
    },
    {
      "epoch": 55.796460176991154,
      "grad_norm": 3.2525891583645716e-05,
      "learning_rate": 4.9224800418480155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1807704,
      "step": 3180
    },
    {
      "epoch": 55.88495575221239,
      "grad_norm": 0.00015037461707834154,
      "learning_rate": 4.9222372730322176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1810440,
      "step": 3185
    },
    {
      "epoch": 55.97345132743363,
      "grad_norm": 7.455499871866778e-05,
      "learning_rate": 4.921994130676993e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1813272,
      "step": 3190
    },
    {
      "epoch": 56.05309734513274,
      "grad_norm": 6.85955019434914e-05,
      "learning_rate": 4.9217506148198366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1815616,
      "step": 3195
    },
    {
      "epoch": 56.14159292035398,
      "grad_norm": 6.52492672088556e-05,
      "learning_rate": 4.921506725498302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1818688,
      "step": 3200
    },
    {
      "epoch": 56.14159292035398,
      "eval_loss": 0.35879766941070557,
      "eval_runtime": 1.0584,
      "eval_samples_per_second": 23.621,
      "eval_steps_per_second": 12.283,
      "num_input_tokens_seen": 1818688,
      "step": 3200
    },
    {
      "epoch": 56.230088495575224,
      "grad_norm": 7.139416265999898e-05,
      "learning_rate": 4.9212624627499994e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1821104,
      "step": 3205
    },
    {
      "epoch": 56.31858407079646,
      "grad_norm": 0.00028747873147949576,
      "learning_rate": 4.921017826612597e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1824144,
      "step": 3210
    },
    {
      "epoch": 56.4070796460177,
      "grad_norm": 2.3780310584697872e-05,
      "learning_rate": 4.9207728171238223e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1827536,
      "step": 3215
    },
    {
      "epoch": 56.49557522123894,
      "grad_norm": 3.605989331845194e-05,
      "learning_rate": 4.920527434321458e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1830624,
      "step": 3220
    },
    {
      "epoch": 56.584070796460175,
      "grad_norm": 0.00010730222857091576,
      "learning_rate": 4.920281678243345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1832992,
      "step": 3225
    },
    {
      "epoch": 56.67256637168141,
      "grad_norm": 6.066475179977715e-05,
      "learning_rate": 4.920035548927381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1835936,
      "step": 3230
    },
    {
      "epoch": 56.76106194690266,
      "grad_norm": 0.00010475570161361247,
      "learning_rate": 4.919789046411525e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1838768,
      "step": 3235
    },
    {
      "epoch": 56.849557522123895,
      "grad_norm": 6.902002496644855e-05,
      "learning_rate": 4.919542170733787e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1841488,
      "step": 3240
    },
    {
      "epoch": 56.93805309734513,
      "grad_norm": 4.7111221647355705e-05,
      "learning_rate": 4.919294921932242e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1844688,
      "step": 3245
    },
    {
      "epoch": 57.017699115044245,
      "grad_norm": 0.00013267651956994087,
      "learning_rate": 4.919047300045016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1847216,
      "step": 3250
    },
    {
      "epoch": 57.10619469026549,
      "grad_norm": 4.6445667976513505e-05,
      "learning_rate": 4.918799305110299e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1849856,
      "step": 3255
    },
    {
      "epoch": 57.19469026548673,
      "grad_norm": 6.205553654581308e-05,
      "learning_rate": 4.918550937166331e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1852576,
      "step": 3260
    },
    {
      "epoch": 57.283185840707965,
      "grad_norm": 0.00010871393897105008,
      "learning_rate": 4.918302196251415e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1855872,
      "step": 3265
    },
    {
      "epoch": 57.3716814159292,
      "grad_norm": 1.894982415251434e-05,
      "learning_rate": 4.91805308240391e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1858416,
      "step": 3270
    },
    {
      "epoch": 57.46017699115044,
      "grad_norm": 0.0007602673140354455,
      "learning_rate": 4.9178035956622326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1861024,
      "step": 3275
    },
    {
      "epoch": 57.54867256637168,
      "grad_norm": 4.9074486014433205e-05,
      "learning_rate": 4.917553736064857e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1863616,
      "step": 3280
    },
    {
      "epoch": 57.63716814159292,
      "grad_norm": 0.00019151791639160365,
      "learning_rate": 4.917303503650314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1866672,
      "step": 3285
    },
    {
      "epoch": 57.72566371681416,
      "grad_norm": 5.46238043170888e-05,
      "learning_rate": 4.917052898457194e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1869456,
      "step": 3290
    },
    {
      "epoch": 57.8141592920354,
      "grad_norm": 1.8251523215440102e-05,
      "learning_rate": 4.916801920524141e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1872032,
      "step": 3295
    },
    {
      "epoch": 57.902654867256636,
      "grad_norm": 0.0003076460852753371,
      "learning_rate": 4.916550569889862e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1875104,
      "step": 3300
    },
    {
      "epoch": 57.991150442477874,
      "grad_norm": 7.653979992028326e-05,
      "learning_rate": 4.916298846593116e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1878672,
      "step": 3305
    },
    {
      "epoch": 58.07079646017699,
      "grad_norm": 2.2112724764156155e-05,
      "learning_rate": 4.916046750672722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1881032,
      "step": 3310
    },
    {
      "epoch": 58.15929203539823,
      "grad_norm": 6.863084126962349e-05,
      "learning_rate": 4.915794282167559e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1883480,
      "step": 3315
    },
    {
      "epoch": 58.24778761061947,
      "grad_norm": 5.78256840526592e-05,
      "learning_rate": 4.915541441116558e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1886648,
      "step": 3320
    },
    {
      "epoch": 58.336283185840706,
      "grad_norm": 4.6437813580268994e-05,
      "learning_rate": 4.915288227558711e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1890216,
      "step": 3325
    },
    {
      "epoch": 58.424778761061944,
      "grad_norm": 0.0004961728118360043,
      "learning_rate": 4.915034641533066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1892888,
      "step": 3330
    },
    {
      "epoch": 58.51327433628319,
      "grad_norm": 2.639968261064496e-05,
      "learning_rate": 4.914780683078731e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1895944,
      "step": 3335
    },
    {
      "epoch": 58.60176991150443,
      "grad_norm": 2.2175810954649933e-05,
      "learning_rate": 4.9145263522348695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1898840,
      "step": 3340
    },
    {
      "epoch": 58.690265486725664,
      "grad_norm": 3.1899293389869854e-05,
      "learning_rate": 4.9142716490407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1901560,
      "step": 3345
    },
    {
      "epoch": 58.7787610619469,
      "grad_norm": 0.00017229038348887116,
      "learning_rate": 4.914016573535504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1904248,
      "step": 3350
    },
    {
      "epoch": 58.86725663716814,
      "grad_norm": 8.041339606279507e-05,
      "learning_rate": 4.9137611257586154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1906904,
      "step": 3355
    },
    {
      "epoch": 58.95575221238938,
      "grad_norm": 1.603244527359493e-05,
      "learning_rate": 4.9135053057494274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1909800,
      "step": 3360
    },
    {
      "epoch": 59.0353982300885,
      "grad_norm": 1.1897727745235898e-05,
      "learning_rate": 4.913249113547392e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1912184,
      "step": 3365
    },
    {
      "epoch": 59.123893805309734,
      "grad_norm": 0.00012359664833638817,
      "learning_rate": 4.912992549192016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1914504,
      "step": 3370
    },
    {
      "epoch": 59.21238938053097,
      "grad_norm": 0.000259844382526353,
      "learning_rate": 4.9127356127228665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1917192,
      "step": 3375
    },
    {
      "epoch": 59.30088495575221,
      "grad_norm": 0.00015762238763272762,
      "learning_rate": 4.912478304179564e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1920056,
      "step": 3380
    },
    {
      "epoch": 59.389380530973455,
      "grad_norm": 3.872895103995688e-05,
      "learning_rate": 4.9122206236017896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1923048,
      "step": 3385
    },
    {
      "epoch": 59.47787610619469,
      "grad_norm": 3.326830847072415e-05,
      "learning_rate": 4.911962571029282e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1926328,
      "step": 3390
    },
    {
      "epoch": 59.56637168141593,
      "grad_norm": 0.0002398610522504896,
      "learning_rate": 4.9117041465018353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1929640,
      "step": 3395
    },
    {
      "epoch": 59.65486725663717,
      "grad_norm": 0.00032528661540709436,
      "learning_rate": 4.911445350059302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1932248,
      "step": 3400
    },
    {
      "epoch": 59.65486725663717,
      "eval_loss": 0.3547935485839844,
      "eval_runtime": 1.0629,
      "eval_samples_per_second": 23.52,
      "eval_steps_per_second": 12.23,
      "num_input_tokens_seen": 1932248,
      "step": 3400
    },
    {
      "epoch": 59.743362831858406,
      "grad_norm": 5.8042696764459834e-05,
      "learning_rate": 4.9111861817415905e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1935208,
      "step": 3405
    },
    {
      "epoch": 59.83185840707964,
      "grad_norm": 5.449409945867956e-05,
      "learning_rate": 4.91092664158867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1938216,
      "step": 3410
    },
    {
      "epoch": 59.92035398230089,
      "grad_norm": 8.197502756956965e-05,
      "learning_rate": 4.910666729640563e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1941000,
      "step": 3415
    },
    {
      "epoch": 60.0,
      "grad_norm": 8.958328180597164e-06,
      "learning_rate": 4.910406445937353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1943544,
      "step": 3420
    },
    {
      "epoch": 60.08849557522124,
      "grad_norm": 3.787256355280988e-05,
      "learning_rate": 4.9101457905191774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1946152,
      "step": 3425
    },
    {
      "epoch": 60.176991150442475,
      "grad_norm": 7.11195170879364e-05,
      "learning_rate": 4.909884763426233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1949240,
      "step": 3430
    },
    {
      "epoch": 60.26548672566372,
      "grad_norm": 1.2485955267038662e-05,
      "learning_rate": 4.9096233646987736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1952248,
      "step": 3435
    },
    {
      "epoch": 60.35398230088496,
      "grad_norm": 0.00032117919181473553,
      "learning_rate": 4.9093615943771104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1955064,
      "step": 3440
    },
    {
      "epoch": 60.442477876106196,
      "grad_norm": 3.726742943399586e-05,
      "learning_rate": 4.909099452501611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1957896,
      "step": 3445
    },
    {
      "epoch": 60.530973451327434,
      "grad_norm": 3.1807248888071626e-05,
      "learning_rate": 4.908836939112702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1960632,
      "step": 3450
    },
    {
      "epoch": 60.61946902654867,
      "grad_norm": 0.00022837235883343965,
      "learning_rate": 4.908574054250865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1963416,
      "step": 3455
    },
    {
      "epoch": 60.70796460176991,
      "grad_norm": 6.80694793118164e-05,
      "learning_rate": 4.9083107979566414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1966376,
      "step": 3460
    },
    {
      "epoch": 60.796460176991154,
      "grad_norm": 0.0002631841925904155,
      "learning_rate": 4.908047170270628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1969272,
      "step": 3465
    },
    {
      "epoch": 60.88495575221239,
      "grad_norm": 5.464500281959772e-05,
      "learning_rate": 4.9077831712334784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1972168,
      "step": 3470
    },
    {
      "epoch": 60.97345132743363,
      "grad_norm": 1.590252941241488e-05,
      "learning_rate": 4.907518800885907e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1975432,
      "step": 3475
    },
    {
      "epoch": 61.05309734513274,
      "grad_norm": 1.5700936273788102e-05,
      "learning_rate": 4.907254059268681e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1978104,
      "step": 3480
    },
    {
      "epoch": 61.14159292035398,
      "grad_norm": 9.122015035245568e-05,
      "learning_rate": 4.906988946422628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1980872,
      "step": 3485
    },
    {
      "epoch": 61.230088495575224,
      "grad_norm": 7.129135337891057e-05,
      "learning_rate": 4.9067234623886315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1983864,
      "step": 3490
    },
    {
      "epoch": 61.31858407079646,
      "grad_norm": 1.7830740034696646e-05,
      "learning_rate": 4.9064576072076316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1986440,
      "step": 3495
    },
    {
      "epoch": 61.4070796460177,
      "grad_norm": 4.901445208815858e-05,
      "learning_rate": 4.906191380920628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1989336,
      "step": 3500
    },
    {
      "epoch": 61.49557522123894,
      "grad_norm": 0.00012169154797447845,
      "learning_rate": 4.905924783568675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1991624,
      "step": 3505
    },
    {
      "epoch": 61.584070796460175,
      "grad_norm": 2.540066998335533e-05,
      "learning_rate": 4.905657815192886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1994440,
      "step": 3510
    },
    {
      "epoch": 61.67256637168141,
      "grad_norm": 4.668137262342498e-05,
      "learning_rate": 4.90539047583443e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 1997704,
      "step": 3515
    },
    {
      "epoch": 61.76106194690266,
      "grad_norm": 5.0458056648494676e-05,
      "learning_rate": 4.905122765534534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2000312,
      "step": 3520
    },
    {
      "epoch": 61.849557522123895,
      "grad_norm": 0.000335978576913476,
      "learning_rate": 4.9048546843344846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2003096,
      "step": 3525
    },
    {
      "epoch": 61.93805309734513,
      "grad_norm": 1.0124652362719644e-05,
      "learning_rate": 4.9045862322756206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2006040,
      "step": 3530
    },
    {
      "epoch": 62.017699115044245,
      "grad_norm": 1.8111390090780333e-05,
      "learning_rate": 4.904317409399342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2008712,
      "step": 3535
    },
    {
      "epoch": 62.10619469026549,
      "grad_norm": 2.4464543457725085e-05,
      "learning_rate": 4.904048215747104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2011672,
      "step": 3540
    },
    {
      "epoch": 62.19469026548673,
      "grad_norm": 9.575260628480464e-05,
      "learning_rate": 4.90377865136042e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2014808,
      "step": 3545
    },
    {
      "epoch": 62.283185840707965,
      "grad_norm": 5.795953256892972e-05,
      "learning_rate": 4.90350871628086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2018024,
      "step": 3550
    },
    {
      "epoch": 62.3716814159292,
      "grad_norm": 6.78955257171765e-05,
      "learning_rate": 4.903238410550052e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2020360,
      "step": 3555
    },
    {
      "epoch": 62.46017699115044,
      "grad_norm": 1.32651039166376e-05,
      "learning_rate": 4.90296773420968e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2023064,
      "step": 3560
    },
    {
      "epoch": 62.54867256637168,
      "grad_norm": 4.5410073653329164e-05,
      "learning_rate": 4.902696687301486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2026040,
      "step": 3565
    },
    {
      "epoch": 62.63716814159292,
      "grad_norm": 4.734982940135524e-05,
      "learning_rate": 4.902425269867268e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2028696,
      "step": 3570
    },
    {
      "epoch": 62.72566371681416,
      "grad_norm": 4.629569957614876e-05,
      "learning_rate": 4.902153481948883e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2031720,
      "step": 3575
    },
    {
      "epoch": 62.8141592920354,
      "grad_norm": 2.0748344468302093e-05,
      "learning_rate": 4.901881323588244e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2034760,
      "step": 3580
    },
    {
      "epoch": 62.902654867256636,
      "grad_norm": 0.00011364412057446316,
      "learning_rate": 4.90160879482732e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2037736,
      "step": 3585
    },
    {
      "epoch": 62.991150442477874,
      "grad_norm": 0.0007809350499883294,
      "learning_rate": 4.9013358957081405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2040008,
      "step": 3590
    },
    {
      "epoch": 63.07079646017699,
      "grad_norm": 2.1866355382371694e-05,
      "learning_rate": 4.901062626272789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2042440,
      "step": 3595
    },
    {
      "epoch": 63.15929203539823,
      "grad_norm": 7.865933002904058e-05,
      "learning_rate": 4.900788986563406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2045464,
      "step": 3600
    },
    {
      "epoch": 63.15929203539823,
      "eval_loss": 0.36013948917388916,
      "eval_runtime": 1.057,
      "eval_samples_per_second": 23.652,
      "eval_steps_per_second": 12.299,
      "num_input_tokens_seen": 2045464,
      "step": 3600
    },
    {
      "epoch": 63.24778761061947,
      "grad_norm": 1.820084980863612e-05,
      "learning_rate": 4.9005149766221915e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2048200,
      "step": 3605
    },
    {
      "epoch": 63.336283185840706,
      "grad_norm": 0.00012301075912546366,
      "learning_rate": 4.9002405964914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2051080,
      "step": 3610
    },
    {
      "epoch": 63.424778761061944,
      "grad_norm": 0.0001106782365241088,
      "learning_rate": 4.899965846213346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2054040,
      "step": 3615
    },
    {
      "epoch": 63.51327433628319,
      "grad_norm": 7.73612700868398e-05,
      "learning_rate": 4.899690725830399e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2056968,
      "step": 3620
    },
    {
      "epoch": 63.60176991150443,
      "grad_norm": 3.678886787383817e-05,
      "learning_rate": 4.899415235384985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2059704,
      "step": 3625
    },
    {
      "epoch": 63.690265486725664,
      "grad_norm": 4.105166590306908e-05,
      "learning_rate": 4.899139374919589e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2062520,
      "step": 3630
    },
    {
      "epoch": 63.7787610619469,
      "grad_norm": 0.00023377638717647642,
      "learning_rate": 4.898863144476752e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2065688,
      "step": 3635
    },
    {
      "epoch": 63.86725663716814,
      "grad_norm": 3.6445104342419654e-05,
      "learning_rate": 4.898586544099072e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2068472,
      "step": 3640
    },
    {
      "epoch": 63.95575221238938,
      "grad_norm": 9.92932400549762e-05,
      "learning_rate": 4.898309573829204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2071192,
      "step": 3645
    },
    {
      "epoch": 64.03539823008849,
      "grad_norm": 6.637921615038067e-05,
      "learning_rate": 4.898032233709862e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2073424,
      "step": 3650
    },
    {
      "epoch": 64.12389380530973,
      "grad_norm": 3.1585579563397914e-05,
      "learning_rate": 4.8977545237838123e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2076352,
      "step": 3655
    },
    {
      "epoch": 64.21238938053098,
      "grad_norm": 0.00010990969894919544,
      "learning_rate": 4.8974764440938836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2078912,
      "step": 3660
    },
    {
      "epoch": 64.30088495575221,
      "grad_norm": 5.582016092375852e-05,
      "learning_rate": 4.897197994682959e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2081904,
      "step": 3665
    },
    {
      "epoch": 64.38938053097345,
      "grad_norm": 9.379841503687203e-05,
      "learning_rate": 4.8969191755939786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2084768,
      "step": 3670
    },
    {
      "epoch": 64.47787610619469,
      "grad_norm": 5.405470074038021e-05,
      "learning_rate": 4.8966399868699396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2087792,
      "step": 3675
    },
    {
      "epoch": 64.56637168141593,
      "grad_norm": 1.79699927684851e-05,
      "learning_rate": 4.8963604285538965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2090336,
      "step": 3680
    },
    {
      "epoch": 64.65486725663717,
      "grad_norm": 9.72160978562897e-06,
      "learning_rate": 4.8960805006889604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2093264,
      "step": 3685
    },
    {
      "epoch": 64.7433628318584,
      "grad_norm": 2.227640652563423e-05,
      "learning_rate": 4.8958002033183004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2095856,
      "step": 3690
    },
    {
      "epoch": 64.83185840707965,
      "grad_norm": 2.9457034543156624e-05,
      "learning_rate": 4.8955195364851414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2099568,
      "step": 3695
    },
    {
      "epoch": 64.92035398230088,
      "grad_norm": 0.00022611797612626106,
      "learning_rate": 4.895238500232766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2102688,
      "step": 3700
    },
    {
      "epoch": 65.0,
      "grad_norm": 3.479194492683746e-05,
      "learning_rate": 4.8949570946045143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2104960,
      "step": 3705
    },
    {
      "epoch": 65.08849557522124,
      "grad_norm": 4.333892502472736e-05,
      "learning_rate": 4.89467531964378e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2107680,
      "step": 3710
    },
    {
      "epoch": 65.17699115044248,
      "grad_norm": 2.8041982659487985e-05,
      "learning_rate": 4.894393175394019e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2110384,
      "step": 3715
    },
    {
      "epoch": 65.26548672566372,
      "grad_norm": 0.00017448014114052057,
      "learning_rate": 4.8941106618987406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2113184,
      "step": 3720
    },
    {
      "epoch": 65.35398230088495,
      "grad_norm": 1.5380854165414348e-05,
      "learning_rate": 4.893827779201512e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2116048,
      "step": 3725
    },
    {
      "epoch": 65.4424778761062,
      "grad_norm": 8.366975089302287e-05,
      "learning_rate": 4.893544527345957e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2119344,
      "step": 3730
    },
    {
      "epoch": 65.53097345132744,
      "grad_norm": 5.1585542678367347e-05,
      "learning_rate": 4.8932609063757563e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2122240,
      "step": 3735
    },
    {
      "epoch": 65.61946902654867,
      "grad_norm": 4.340472150943242e-05,
      "learning_rate": 4.8929769163346484e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2124912,
      "step": 3740
    },
    {
      "epoch": 65.70796460176992,
      "grad_norm": 4.2455605580471456e-05,
      "learning_rate": 4.892692557266429e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2128208,
      "step": 3745
    },
    {
      "epoch": 65.79646017699115,
      "grad_norm": 4.6908902731956914e-05,
      "learning_rate": 4.8924078292149464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2130848,
      "step": 3750
    },
    {
      "epoch": 65.88495575221239,
      "grad_norm": 4.5259686885401607e-05,
      "learning_rate": 4.892122732224114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2133728,
      "step": 3755
    },
    {
      "epoch": 65.97345132743362,
      "grad_norm": 6.708678847644478e-05,
      "learning_rate": 4.8918372663378944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2136624,
      "step": 3760
    },
    {
      "epoch": 66.05309734513274,
      "grad_norm": 2.4107723220367916e-05,
      "learning_rate": 4.89155143160031e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2139000,
      "step": 3765
    },
    {
      "epoch": 66.14159292035399,
      "grad_norm": 2.1527803255594335e-05,
      "learning_rate": 4.891265228055441e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2142296,
      "step": 3770
    },
    {
      "epoch": 66.23008849557522,
      "grad_norm": 3.755742727662437e-05,
      "learning_rate": 4.890978655747424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2145432,
      "step": 3775
    },
    {
      "epoch": 66.31858407079646,
      "grad_norm": 1.1278111742285546e-05,
      "learning_rate": 4.89069171472045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2147960,
      "step": 3780
    },
    {
      "epoch": 66.40707964601769,
      "grad_norm": 1.6600199160166085e-05,
      "learning_rate": 4.890404405018772e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2151112,
      "step": 3785
    },
    {
      "epoch": 66.49557522123894,
      "grad_norm": 1.4172598639561329e-05,
      "learning_rate": 4.8901167266866934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2153448,
      "step": 3790
    },
    {
      "epoch": 66.58407079646018,
      "grad_norm": 1.5992813132470474e-05,
      "learning_rate": 4.88982867976858e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2156280,
      "step": 3795
    },
    {
      "epoch": 66.67256637168141,
      "grad_norm": 4.2733870941447094e-05,
      "learning_rate": 4.889540264308852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2159128,
      "step": 3800
    },
    {
      "epoch": 66.67256637168141,
      "eval_loss": 0.36583343148231506,
      "eval_runtime": 1.0622,
      "eval_samples_per_second": 23.537,
      "eval_steps_per_second": 12.239,
      "num_input_tokens_seen": 2159128,
      "step": 3800
    },
    {
      "epoch": 66.76106194690266,
      "grad_norm": 3.0683178920298815e-05,
      "learning_rate": 4.889251480351986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2162344,
      "step": 3805
    },
    {
      "epoch": 66.84955752212389,
      "grad_norm": 2.9403878215816803e-05,
      "learning_rate": 4.888962327942517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2164920,
      "step": 3810
    },
    {
      "epoch": 66.93805309734513,
      "grad_norm": 2.6320381948607974e-05,
      "learning_rate": 4.8886728071250356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2168024,
      "step": 3815
    },
    {
      "epoch": 67.01769911504425,
      "grad_norm": 0.0001398851745761931,
      "learning_rate": 4.8883829179441884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2170720,
      "step": 3820
    },
    {
      "epoch": 67.10619469026548,
      "grad_norm": 0.00018407282186672091,
      "learning_rate": 4.888092660444682e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2173312,
      "step": 3825
    },
    {
      "epoch": 67.19469026548673,
      "grad_norm": 1.9264864022261463e-05,
      "learning_rate": 4.887802034671276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2175984,
      "step": 3830
    },
    {
      "epoch": 67.28318584070796,
      "grad_norm": 1.1777989129768685e-05,
      "learning_rate": 4.88751104066879e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2178624,
      "step": 3835
    },
    {
      "epoch": 67.3716814159292,
      "grad_norm": 2.3476011847378686e-05,
      "learning_rate": 4.887219678482098e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2181088,
      "step": 3840
    },
    {
      "epoch": 67.46017699115045,
      "grad_norm": 0.0003668139106594026,
      "learning_rate": 4.8869279481561316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2184736,
      "step": 3845
    },
    {
      "epoch": 67.54867256637168,
      "grad_norm": 6.838345871074125e-05,
      "learning_rate": 4.88663584973588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2187296,
      "step": 3850
    },
    {
      "epoch": 67.63716814159292,
      "grad_norm": 5.2331706683617085e-05,
      "learning_rate": 4.8863433832663874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2190432,
      "step": 3855
    },
    {
      "epoch": 67.72566371681415,
      "grad_norm": 1.876895476016216e-05,
      "learning_rate": 4.886050548792757e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2193568,
      "step": 3860
    },
    {
      "epoch": 67.8141592920354,
      "grad_norm": 4.194037683191709e-05,
      "learning_rate": 4.8857573463601465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2196464,
      "step": 3865
    },
    {
      "epoch": 67.90265486725664,
      "grad_norm": 0.00010976172052323818,
      "learning_rate": 4.885463776013772e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2199584,
      "step": 3870
    },
    {
      "epoch": 67.99115044247787,
      "grad_norm": 0.00015385162259917706,
      "learning_rate": 4.8851698377989056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2202432,
      "step": 3875
    },
    {
      "epoch": 68.070796460177,
      "grad_norm": 2.1928795831627212e-05,
      "learning_rate": 4.884875531760876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2205616,
      "step": 3880
    },
    {
      "epoch": 68.15929203539822,
      "grad_norm": 6.213218148332089e-05,
      "learning_rate": 4.88458085794507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2208608,
      "step": 3885
    },
    {
      "epoch": 68.24778761061947,
      "grad_norm": 9.123675408773124e-05,
      "learning_rate": 4.884285816396929e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2210912,
      "step": 3890
    },
    {
      "epoch": 68.33628318584071,
      "grad_norm": 4.7259440179914236e-05,
      "learning_rate": 4.8839904071619526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2213952,
      "step": 3895
    },
    {
      "epoch": 68.42477876106194,
      "grad_norm": 3.597157774493098e-05,
      "learning_rate": 4.8836946302856955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2217024,
      "step": 3900
    },
    {
      "epoch": 68.51327433628319,
      "grad_norm": 2.461105032125488e-05,
      "learning_rate": 4.8833984858137715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2219680,
      "step": 3905
    },
    {
      "epoch": 68.60176991150442,
      "grad_norm": 0.0001590948668308556,
      "learning_rate": 4.8831019737918494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2222384,
      "step": 3910
    },
    {
      "epoch": 68.69026548672566,
      "grad_norm": 2.871975812013261e-05,
      "learning_rate": 4.882805094265655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2224864,
      "step": 3915
    },
    {
      "epoch": 68.77876106194691,
      "grad_norm": 1.0833410669874866e-05,
      "learning_rate": 4.8825078472809706e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2228192,
      "step": 3920
    },
    {
      "epoch": 68.86725663716814,
      "grad_norm": 2.0664323528762907e-05,
      "learning_rate": 4.882210232883635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2230976,
      "step": 3925
    },
    {
      "epoch": 68.95575221238938,
      "grad_norm": 1.7511618352727965e-05,
      "learning_rate": 4.881912251119546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2233968,
      "step": 3930
    },
    {
      "epoch": 69.03539823008849,
      "grad_norm": 4.239720510668121e-05,
      "learning_rate": 4.881613902034654e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2236392,
      "step": 3935
    },
    {
      "epoch": 69.12389380530973,
      "grad_norm": 2.003320332732983e-05,
      "learning_rate": 4.88131518567497e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2239032,
      "step": 3940
    },
    {
      "epoch": 69.21238938053098,
      "grad_norm": 3.676183769130148e-05,
      "learning_rate": 4.881016102086558e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2241912,
      "step": 3945
    },
    {
      "epoch": 69.30088495575221,
      "grad_norm": 3.965793803217821e-05,
      "learning_rate": 4.8807166513155425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2245016,
      "step": 3950
    },
    {
      "epoch": 69.38938053097345,
      "grad_norm": 2.7216023227083497e-05,
      "learning_rate": 4.8804168334081004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2247720,
      "step": 3955
    },
    {
      "epoch": 69.47787610619469,
      "grad_norm": 0.0002732703578658402,
      "learning_rate": 4.880116648410468e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2250360,
      "step": 3960
    },
    {
      "epoch": 69.56637168141593,
      "grad_norm": 2.488026257196907e-05,
      "learning_rate": 4.879816096368939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2253272,
      "step": 3965
    },
    {
      "epoch": 69.65486725663717,
      "grad_norm": 2.2940035705687478e-05,
      "learning_rate": 4.879515177329861e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2256120,
      "step": 3970
    },
    {
      "epoch": 69.7433628318584,
      "grad_norm": 0.00026151916244998574,
      "learning_rate": 4.8792138913396394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2259400,
      "step": 3975
    },
    {
      "epoch": 69.83185840707965,
      "grad_norm": 7.597067451570183e-05,
      "learning_rate": 4.8789122384447374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2262344,
      "step": 3980
    },
    {
      "epoch": 69.92035398230088,
      "grad_norm": 1.991814497159794e-05,
      "learning_rate": 4.878610218691673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2265160,
      "step": 3985
    },
    {
      "epoch": 70.0,
      "grad_norm": 2.690645669645164e-05,
      "learning_rate": 4.87830783212702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2267528,
      "step": 3990
    },
    {
      "epoch": 70.08849557522124,
      "grad_norm": 0.0002911906922236085,
      "learning_rate": 4.878005078797413e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2270056,
      "step": 3995
    },
    {
      "epoch": 70.17699115044248,
      "grad_norm": 7.77260065660812e-05,
      "learning_rate": 4.877701958749539e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2272792,
      "step": 4000
    },
    {
      "epoch": 70.17699115044248,
      "eval_loss": 0.37287405133247375,
      "eval_runtime": 1.0592,
      "eval_samples_per_second": 23.602,
      "eval_steps_per_second": 12.273,
      "num_input_tokens_seen": 2272792,
      "step": 4000
    },
    {
      "epoch": 70.26548672566372,
      "grad_norm": 1.7480368114775047e-05,
      "learning_rate": 4.877398472030142e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2275432,
      "step": 4005
    },
    {
      "epoch": 70.35398230088495,
      "grad_norm": 8.91624586074613e-06,
      "learning_rate": 4.877094618686024e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2278280,
      "step": 4010
    },
    {
      "epoch": 70.4424778761062,
      "grad_norm": 2.050289731414523e-05,
      "learning_rate": 4.876790398764045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2281128,
      "step": 4015
    },
    {
      "epoch": 70.53097345132744,
      "grad_norm": 7.38712478778325e-05,
      "learning_rate": 4.8764858123111167e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2284376,
      "step": 4020
    },
    {
      "epoch": 70.61946902654867,
      "grad_norm": 9.570927795721218e-05,
      "learning_rate": 4.876180859374212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2287208,
      "step": 4025
    },
    {
      "epoch": 70.70796460176992,
      "grad_norm": 8.711057489563245e-06,
      "learning_rate": 4.875875540000357e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2290024,
      "step": 4030
    },
    {
      "epoch": 70.79646017699115,
      "grad_norm": 5.2813466027146205e-05,
      "learning_rate": 4.8755698542366376e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2292616,
      "step": 4035
    },
    {
      "epoch": 70.88495575221239,
      "grad_norm": 7.522765372414142e-06,
      "learning_rate": 4.875263802130193e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2296104,
      "step": 4040
    },
    {
      "epoch": 70.97345132743362,
      "grad_norm": 2.6827097826753743e-05,
      "learning_rate": 4.8749573837282207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2298888,
      "step": 4045
    },
    {
      "epoch": 71.05309734513274,
      "grad_norm": 4.825209907721728e-05,
      "learning_rate": 4.874650599077974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2301192,
      "step": 4050
    },
    {
      "epoch": 71.14159292035399,
      "grad_norm": 0.00016790704103186727,
      "learning_rate": 4.874343448226764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2304312,
      "step": 4055
    },
    {
      "epoch": 71.23008849557522,
      "grad_norm": 1.2416728168318514e-05,
      "learning_rate": 4.874035931221955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2306824,
      "step": 4060
    },
    {
      "epoch": 71.31858407079646,
      "grad_norm": 1.892712498374749e-05,
      "learning_rate": 4.8737280481109724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2309800,
      "step": 4065
    },
    {
      "epoch": 71.40707964601769,
      "grad_norm": 2.4891673092497513e-05,
      "learning_rate": 4.873419798941294e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2312728,
      "step": 4070
    },
    {
      "epoch": 71.49557522123894,
      "grad_norm": 2.9852360967197455e-05,
      "learning_rate": 4.873111183760458e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2315608,
      "step": 4075
    },
    {
      "epoch": 71.58407079646018,
      "grad_norm": 1.4443602594838012e-05,
      "learning_rate": 4.8728022026160537e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2318264,
      "step": 4080
    },
    {
      "epoch": 71.67256637168141,
      "grad_norm": 2.20944857574068e-05,
      "learning_rate": 4.872492855555732e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2321032,
      "step": 4085
    },
    {
      "epoch": 71.76106194690266,
      "grad_norm": 1.8545693819760345e-05,
      "learning_rate": 4.8721831426271956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2324568,
      "step": 4090
    },
    {
      "epoch": 71.84955752212389,
      "grad_norm": 0.0002730564447119832,
      "learning_rate": 4.87187306387821e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2327144,
      "step": 4095
    },
    {
      "epoch": 71.93805309734513,
      "grad_norm": 8.594322571298108e-05,
      "learning_rate": 4.87156261935659e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2330104,
      "step": 4100
    },
    {
      "epoch": 72.01769911504425,
      "grad_norm": 1.0080576430482324e-05,
      "learning_rate": 4.871251809110211e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2332656,
      "step": 4105
    },
    {
      "epoch": 72.10619469026548,
      "grad_norm": 2.6484145564609207e-05,
      "learning_rate": 4.8709406331870044e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2335584,
      "step": 4110
    },
    {
      "epoch": 72.19469026548673,
      "grad_norm": 6.761600525351241e-05,
      "learning_rate": 4.8706290916349574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2338048,
      "step": 4115
    },
    {
      "epoch": 72.28318584070796,
      "grad_norm": 2.7466390747576952e-05,
      "learning_rate": 4.8703171845021134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2340848,
      "step": 4120
    },
    {
      "epoch": 72.3716814159292,
      "grad_norm": 0.00010778058640426025,
      "learning_rate": 4.870004911836572e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2344112,
      "step": 4125
    },
    {
      "epoch": 72.46017699115045,
      "grad_norm": 2.1217116227489896e-05,
      "learning_rate": 4.869692273686489e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2346544,
      "step": 4130
    },
    {
      "epoch": 72.54867256637168,
      "grad_norm": 8.481163968099281e-05,
      "learning_rate": 4.869379270100079e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2349456,
      "step": 4135
    },
    {
      "epoch": 72.63716814159292,
      "grad_norm": 3.4898876037914306e-05,
      "learning_rate": 4.86906590112561e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2352160,
      "step": 4140
    },
    {
      "epoch": 72.72566371681415,
      "grad_norm": 5.343929660739377e-05,
      "learning_rate": 4.8687521668114064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2355104,
      "step": 4145
    },
    {
      "epoch": 72.8141592920354,
      "grad_norm": 7.327162165893242e-05,
      "learning_rate": 4.868438067205853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2358288,
      "step": 4150
    },
    {
      "epoch": 72.90265486725664,
      "grad_norm": 1.6481488273711875e-05,
      "learning_rate": 4.8681236023573844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2361056,
      "step": 4155
    },
    {
      "epoch": 72.99115044247787,
      "grad_norm": 2.922467319876887e-05,
      "learning_rate": 4.867808772314497e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2364160,
      "step": 4160
    },
    {
      "epoch": 73.070796460177,
      "grad_norm": 3.9821174141252413e-05,
      "learning_rate": 4.867493577125741e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2366384,
      "step": 4165
    },
    {
      "epoch": 73.15929203539822,
      "grad_norm": 1.2287973731872626e-05,
      "learning_rate": 4.867178016839725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2369632,
      "step": 4170
    },
    {
      "epoch": 73.24778761061947,
      "grad_norm": 1.6550766304135323e-05,
      "learning_rate": 4.8668620915051094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2372880,
      "step": 4175
    },
    {
      "epoch": 73.33628318584071,
      "grad_norm": 0.0003210812574252486,
      "learning_rate": 4.866545801170616e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2375344,
      "step": 4180
    },
    {
      "epoch": 73.42477876106194,
      "grad_norm": 3.975871004513465e-05,
      "learning_rate": 4.86622914588502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2378288,
      "step": 4185
    },
    {
      "epoch": 73.51327433628319,
      "grad_norm": 4.9271351599600166e-05,
      "learning_rate": 4.865912125697154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2381600,
      "step": 4190
    },
    {
      "epoch": 73.60176991150442,
      "grad_norm": 1.7828917407314293e-05,
      "learning_rate": 4.865594740655907e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2384096,
      "step": 4195
    },
    {
      "epoch": 73.69026548672566,
      "grad_norm": 3.523853956721723e-05,
      "learning_rate": 4.865276990810222e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2387344,
      "step": 4200
    },
    {
      "epoch": 73.69026548672566,
      "eval_loss": 0.37824875116348267,
      "eval_runtime": 1.0573,
      "eval_samples_per_second": 23.646,
      "eval_steps_per_second": 12.296,
      "num_input_tokens_seen": 2387344,
      "step": 4200
    },
    {
      "epoch": 73.77876106194691,
      "grad_norm": 0.00016553171735722572,
      "learning_rate": 4.8649588762091016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2390256,
      "step": 4205
    },
    {
      "epoch": 73.86725663716814,
      "grad_norm": 6.28663765382953e-05,
      "learning_rate": 4.8646403969016016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2393056,
      "step": 4210
    },
    {
      "epoch": 73.95575221238938,
      "grad_norm": 1.0182659934798721e-05,
      "learning_rate": 4.864321552936838e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2395600,
      "step": 4215
    },
    {
      "epoch": 74.03539823008849,
      "grad_norm": 2.9327280572033487e-05,
      "learning_rate": 4.864002344363978e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2397832,
      "step": 4220
    },
    {
      "epoch": 74.12389380530973,
      "grad_norm": 5.346569014363922e-05,
      "learning_rate": 4.863682771232248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2400872,
      "step": 4225
    },
    {
      "epoch": 74.21238938053098,
      "grad_norm": 0.00017243874026462436,
      "learning_rate": 4.8633628335909324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2403544,
      "step": 4230
    },
    {
      "epoch": 74.30088495575221,
      "grad_norm": 8.955665180110373e-06,
      "learning_rate": 4.8630425314893676e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2406744,
      "step": 4235
    },
    {
      "epoch": 74.38938053097345,
      "grad_norm": 3.982438283856027e-05,
      "learning_rate": 4.862721864976948e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2409880,
      "step": 4240
    },
    {
      "epoch": 74.47787610619469,
      "grad_norm": 2.0544166545732878e-05,
      "learning_rate": 4.862400834103125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2412408,
      "step": 4245
    },
    {
      "epoch": 74.56637168141593,
      "grad_norm": 0.00019618013175204396,
      "learning_rate": 4.862079438917406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2415304,
      "step": 4250
    },
    {
      "epoch": 74.65486725663717,
      "grad_norm": 8.818778951535933e-06,
      "learning_rate": 4.8617576794693536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2418088,
      "step": 4255
    },
    {
      "epoch": 74.7433628318584,
      "grad_norm": 2.2091486243880354e-05,
      "learning_rate": 4.8614355558085875e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2420872,
      "step": 4260
    },
    {
      "epoch": 74.83185840707965,
      "grad_norm": 1.587151018611621e-05,
      "learning_rate": 4.861113067984783e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2423592,
      "step": 4265
    },
    {
      "epoch": 74.92035398230088,
      "grad_norm": 2.4346223653992638e-05,
      "learning_rate": 4.860790216047671e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2426584,
      "step": 4270
    },
    {
      "epoch": 75.0,
      "grad_norm": 0.0014797589974477887,
      "learning_rate": 4.860467000047041e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2428728,
      "step": 4275
    },
    {
      "epoch": 75.08849557522124,
      "grad_norm": 1.4160216778691392e-05,
      "learning_rate": 4.860143420032737e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2431576,
      "step": 4280
    },
    {
      "epoch": 75.17699115044248,
      "grad_norm": 2.025627145485487e-05,
      "learning_rate": 4.859819476054657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2434488,
      "step": 4285
    },
    {
      "epoch": 75.26548672566372,
      "grad_norm": 6.855128594907e-05,
      "learning_rate": 4.859495168162758e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2437208,
      "step": 4290
    },
    {
      "epoch": 75.35398230088495,
      "grad_norm": 0.00010113940516021103,
      "learning_rate": 4.859170496407054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2440120,
      "step": 4295
    },
    {
      "epoch": 75.4424778761062,
      "grad_norm": 3.9412065234500915e-05,
      "learning_rate": 4.8588454608376114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2442856,
      "step": 4300
    },
    {
      "epoch": 75.53097345132744,
      "grad_norm": 1.745676127029583e-05,
      "learning_rate": 4.8585200615045555e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2445816,
      "step": 4305
    },
    {
      "epoch": 75.61946902654867,
      "grad_norm": 1.826192055887077e-05,
      "learning_rate": 4.8581942984580674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2448520,
      "step": 4310
    },
    {
      "epoch": 75.70796460176992,
      "grad_norm": 1.643314135435503e-05,
      "learning_rate": 4.857868171748384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2451704,
      "step": 4315
    },
    {
      "epoch": 75.79646017699115,
      "grad_norm": 1.272216559300432e-05,
      "learning_rate": 4.8575416814257976e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2454392,
      "step": 4320
    },
    {
      "epoch": 75.88495575221239,
      "grad_norm": 9.465852053835988e-05,
      "learning_rate": 4.857214827540657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2457304,
      "step": 4325
    },
    {
      "epoch": 75.97345132743362,
      "grad_norm": 2.3755386791890487e-05,
      "learning_rate": 4.856887610143367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2460728,
      "step": 4330
    },
    {
      "epoch": 76.05309734513274,
      "grad_norm": 3.188071059412323e-05,
      "learning_rate": 4.8565600292843896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2463304,
      "step": 4335
    },
    {
      "epoch": 76.14159292035399,
      "grad_norm": 3.7281293771229684e-05,
      "learning_rate": 4.856232085014241e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2465720,
      "step": 4340
    },
    {
      "epoch": 76.23008849557522,
      "grad_norm": 2.364702413615305e-05,
      "learning_rate": 4.855903777383495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2468824,
      "step": 4345
    },
    {
      "epoch": 76.31858407079646,
      "grad_norm": 2.6804116714629345e-05,
      "learning_rate": 4.85557510644278e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2472040,
      "step": 4350
    },
    {
      "epoch": 76.40707964601769,
      "grad_norm": 1.4832770830253139e-05,
      "learning_rate": 4.855246072242782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2474936,
      "step": 4355
    },
    {
      "epoch": 76.49557522123894,
      "grad_norm": 2.892062002501916e-05,
      "learning_rate": 4.8549166748342414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2477736,
      "step": 4360
    },
    {
      "epoch": 76.58407079646018,
      "grad_norm": 6.781006959499791e-06,
      "learning_rate": 4.8545869142679556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2480296,
      "step": 4365
    },
    {
      "epoch": 76.67256637168141,
      "grad_norm": 1.236034859175561e-05,
      "learning_rate": 4.8542567905947776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2483256,
      "step": 4370
    },
    {
      "epoch": 76.76106194690266,
      "grad_norm": 1.1151855687785428e-05,
      "learning_rate": 4.853926303865618e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2486360,
      "step": 4375
    },
    {
      "epoch": 76.84955752212389,
      "grad_norm": 1.1023205843230244e-05,
      "learning_rate": 4.853595454131441e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2489352,
      "step": 4380
    },
    {
      "epoch": 76.93805309734513,
      "grad_norm": 2.4363576812902465e-05,
      "learning_rate": 4.8532642414432674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2492136,
      "step": 4385
    },
    {
      "epoch": 77.01769911504425,
      "grad_norm": 1.7550504708196968e-05,
      "learning_rate": 4.8529326658521754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2494240,
      "step": 4390
    },
    {
      "epoch": 77.10619469026548,
      "grad_norm": 3.9754562749294564e-05,
      "learning_rate": 4.8526007274092965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2497360,
      "step": 4395
    },
    {
      "epoch": 77.19469026548673,
      "grad_norm": 8.509722647431772e-06,
      "learning_rate": 4.852268426165822e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2500160,
      "step": 4400
    },
    {
      "epoch": 77.19469026548673,
      "eval_loss": 0.38144075870513916,
      "eval_runtime": 1.0628,
      "eval_samples_per_second": 23.523,
      "eval_steps_per_second": 12.232,
      "num_input_tokens_seen": 2500160,
      "step": 4400
    },
    {
      "epoch": 77.28318584070796,
      "grad_norm": 2.5738785552675836e-05,
      "learning_rate": 4.851935762172995e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2503184,
      "step": 4405
    },
    {
      "epoch": 77.3716814159292,
      "grad_norm": 3.650345024652779e-05,
      "learning_rate": 4.8516027354821175e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2505808,
      "step": 4410
    },
    {
      "epoch": 77.46017699115045,
      "grad_norm": 1.625017830519937e-05,
      "learning_rate": 4.851269346144546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2508128,
      "step": 4415
    },
    {
      "epoch": 77.54867256637168,
      "grad_norm": 8.671006071381271e-05,
      "learning_rate": 4.850935594211693e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2510896,
      "step": 4420
    },
    {
      "epoch": 77.63716814159292,
      "grad_norm": 1.061580769601278e-05,
      "learning_rate": 4.850601479735029e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2513792,
      "step": 4425
    },
    {
      "epoch": 77.72566371681415,
      "grad_norm": 0.00023255517589859664,
      "learning_rate": 4.850267002766076e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2516512,
      "step": 4430
    },
    {
      "epoch": 77.8141592920354,
      "grad_norm": 0.00010167371510760859,
      "learning_rate": 4.849932163356417e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2520448,
      "step": 4435
    },
    {
      "epoch": 77.90265486725664,
      "grad_norm": 4.0936989535111934e-05,
      "learning_rate": 4.8495969615576864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2523264,
      "step": 4440
    },
    {
      "epoch": 77.99115044247787,
      "grad_norm": 1.8116859791916795e-05,
      "learning_rate": 4.849261397421577e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2525888,
      "step": 4445
    },
    {
      "epoch": 78.070796460177,
      "grad_norm": 1.870875530585181e-05,
      "learning_rate": 4.848925470999839e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2528640,
      "step": 4450
    },
    {
      "epoch": 78.15929203539822,
      "grad_norm": 4.5541477447841316e-05,
      "learning_rate": 4.848589182344273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2531472,
      "step": 4455
    },
    {
      "epoch": 78.24778761061947,
      "grad_norm": 1.3822659639117774e-05,
      "learning_rate": 4.848252531506742e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2533968,
      "step": 4460
    },
    {
      "epoch": 78.33628318584071,
      "grad_norm": 1.218127363245003e-05,
      "learning_rate": 4.847915518539161e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2536464,
      "step": 4465
    },
    {
      "epoch": 78.42477876106194,
      "grad_norm": 0.0002542641304899007,
      "learning_rate": 4.847578143493501e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2539008,
      "step": 4470
    },
    {
      "epoch": 78.51327433628319,
      "grad_norm": 0.0002546671312302351,
      "learning_rate": 4.847240406421789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2542368,
      "step": 4475
    },
    {
      "epoch": 78.60176991150442,
      "grad_norm": 2.811913873301819e-05,
      "learning_rate": 4.84690230737611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2545184,
      "step": 4480
    },
    {
      "epoch": 78.69026548672566,
      "grad_norm": 2.0814972231164575e-05,
      "learning_rate": 4.846563846408602e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2548224,
      "step": 4485
    },
    {
      "epoch": 78.77876106194691,
      "grad_norm": 8.646665264677722e-06,
      "learning_rate": 4.84622502357146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2551072,
      "step": 4490
    },
    {
      "epoch": 78.86725663716814,
      "grad_norm": 1.4519543583446648e-05,
      "learning_rate": 4.8458858389169345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2554016,
      "step": 4495
    },
    {
      "epoch": 78.95575221238938,
      "grad_norm": 1.9556800907594152e-05,
      "learning_rate": 4.8455462924973334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2556960,
      "step": 4500
    },
    {
      "epoch": 79.03539823008849,
      "grad_norm": 2.9565426302724518e-05,
      "learning_rate": 4.845206384365018e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2559240,
      "step": 4505
    },
    {
      "epoch": 79.12389380530973,
      "grad_norm": 1.611733932804782e-05,
      "learning_rate": 4.844866114572405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2562056,
      "step": 4510
    },
    {
      "epoch": 79.21238938053098,
      "grad_norm": 1.4524162907036953e-05,
      "learning_rate": 4.8445254831719706e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2564744,
      "step": 4515
    },
    {
      "epoch": 79.30088495575221,
      "grad_norm": 1.822403646656312e-05,
      "learning_rate": 4.8441844902162434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2568040,
      "step": 4520
    },
    {
      "epoch": 79.38938053097345,
      "grad_norm": 2.4496717742295004e-05,
      "learning_rate": 4.843843135757809e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2570808,
      "step": 4525
    },
    {
      "epoch": 79.47787610619469,
      "grad_norm": 0.00016970120486803353,
      "learning_rate": 4.843501419849308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2573992,
      "step": 4530
    },
    {
      "epoch": 79.56637168141593,
      "grad_norm": 8.531063213013113e-05,
      "learning_rate": 4.8431593425434386e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2576744,
      "step": 4535
    },
    {
      "epoch": 79.65486725663717,
      "grad_norm": 0.0001828560489229858,
      "learning_rate": 4.8428169038929526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2579864,
      "step": 4540
    },
    {
      "epoch": 79.7433628318584,
      "grad_norm": 2.3619038984179497e-05,
      "learning_rate": 4.8424741039506575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2582632,
      "step": 4545
    },
    {
      "epoch": 79.83185840707965,
      "grad_norm": 1.5953575712046586e-05,
      "learning_rate": 4.842130942769419e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2585688,
      "step": 4550
    },
    {
      "epoch": 79.92035398230088,
      "grad_norm": 1.6664333088556305e-05,
      "learning_rate": 4.841787420402156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2588504,
      "step": 4555
    },
    {
      "epoch": 80.0,
      "grad_norm": 1.108821015805006e-05,
      "learning_rate": 4.841443536901844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2590976,
      "step": 4560
    },
    {
      "epoch": 80.08849557522124,
      "grad_norm": 6.948180816834792e-05,
      "learning_rate": 4.841099292321514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2593808,
      "step": 4565
    },
    {
      "epoch": 80.17699115044248,
      "grad_norm": 3.885464684572071e-05,
      "learning_rate": 4.8407546867142525e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2596800,
      "step": 4570
    },
    {
      "epoch": 80.26548672566372,
      "grad_norm": 6.377705722115934e-05,
      "learning_rate": 4.840409720133203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2599616,
      "step": 4575
    },
    {
      "epoch": 80.35398230088495,
      "grad_norm": 4.090169750270434e-05,
      "learning_rate": 4.8400643926315634e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2602432,
      "step": 4580
    },
    {
      "epoch": 80.4424778761062,
      "grad_norm": 8.992806215246674e-06,
      "learning_rate": 4.839718704262587e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2605104,
      "step": 4585
    },
    {
      "epoch": 80.53097345132744,
      "grad_norm": 1.986593269975856e-05,
      "learning_rate": 4.839372655079585e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2607984,
      "step": 4590
    },
    {
      "epoch": 80.61946902654867,
      "grad_norm": 2.386245978414081e-05,
      "learning_rate": 4.83902624513592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2610832,
      "step": 4595
    },
    {
      "epoch": 80.70796460176992,
      "grad_norm": 1.1911199180758558e-05,
      "learning_rate": 4.838679474485014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2614032,
      "step": 4600
    },
    {
      "epoch": 80.70796460176992,
      "eval_loss": 0.362618625164032,
      "eval_runtime": 1.0642,
      "eval_samples_per_second": 23.491,
      "eval_steps_per_second": 12.215,
      "num_input_tokens_seen": 2614032,
      "step": 4600
    },
    {
      "epoch": 80.79646017699115,
      "grad_norm": 1.4681406355521176e-05,
      "learning_rate": 4.838332343180343e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2616800,
      "step": 4605
    },
    {
      "epoch": 80.88495575221239,
      "grad_norm": 4.750757852889365e-06,
      "learning_rate": 4.83798485127544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2619648,
      "step": 4610
    },
    {
      "epoch": 80.97345132743362,
      "grad_norm": 1.4876244676997885e-05,
      "learning_rate": 4.837636998823892e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2622864,
      "step": 4615
    },
    {
      "epoch": 81.05309734513274,
      "grad_norm": 2.3281694666366093e-05,
      "learning_rate": 4.8372887858793414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2625232,
      "step": 4620
    },
    {
      "epoch": 81.14159292035399,
      "grad_norm": 1.0698969163058791e-05,
      "learning_rate": 4.836940212495489e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2628320,
      "step": 4625
    },
    {
      "epoch": 81.23008849557522,
      "grad_norm": 3.0409961254918016e-05,
      "learning_rate": 4.836591278726087e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2631056,
      "step": 4630
    },
    {
      "epoch": 81.31858407079646,
      "grad_norm": 0.00012012160004815087,
      "learning_rate": 4.836241984624947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2633792,
      "step": 4635
    },
    {
      "epoch": 81.40707964601769,
      "grad_norm": 9.046163177117705e-05,
      "learning_rate": 4.8358923302459336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2636288,
      "step": 4640
    },
    {
      "epoch": 81.49557522123894,
      "grad_norm": 1.8183476640842855e-05,
      "learning_rate": 4.835542315642968e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2639536,
      "step": 4645
    },
    {
      "epoch": 81.58407079646018,
      "grad_norm": 1.5246249859046657e-05,
      "learning_rate": 4.8351919408700274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2642256,
      "step": 4650
    },
    {
      "epoch": 81.67256637168141,
      "grad_norm": 0.00018556263239588588,
      "learning_rate": 4.834841205981144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2645280,
      "step": 4655
    },
    {
      "epoch": 81.76106194690266,
      "grad_norm": 0.00012137554585933685,
      "learning_rate": 4.8344901110304054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2648624,
      "step": 4660
    },
    {
      "epoch": 81.84955752212389,
      "grad_norm": 8.073571370914578e-05,
      "learning_rate": 4.8341386560719534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2651824,
      "step": 4665
    },
    {
      "epoch": 81.93805309734513,
      "grad_norm": 8.1516182035557e-06,
      "learning_rate": 4.833786841159989e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2654352,
      "step": 4670
    },
    {
      "epoch": 82.01769911504425,
      "grad_norm": 1.7923599443747662e-05,
      "learning_rate": 4.833434666348765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2656984,
      "step": 4675
    },
    {
      "epoch": 82.10619469026548,
      "grad_norm": 0.00011555101082194597,
      "learning_rate": 4.833082131692592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2659736,
      "step": 4680
    },
    {
      "epoch": 82.19469026548673,
      "grad_norm": 1.4988442671892699e-05,
      "learning_rate": 4.832729237245835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2662392,
      "step": 4685
    },
    {
      "epoch": 82.28318584070796,
      "grad_norm": 2.615781886561308e-05,
      "learning_rate": 4.8323759830629145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2664808,
      "step": 4690
    },
    {
      "epoch": 82.3716814159292,
      "grad_norm": 5.685828364221379e-05,
      "learning_rate": 4.8320223691983066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2667560,
      "step": 4695
    },
    {
      "epoch": 82.46017699115045,
      "grad_norm": 7.593636382807745e-06,
      "learning_rate": 4.831668395706544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2670232,
      "step": 4700
    },
    {
      "epoch": 82.54867256637168,
      "grad_norm": 1.5049660760269035e-05,
      "learning_rate": 4.8313140626422125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2673512,
      "step": 4705
    },
    {
      "epoch": 82.63716814159292,
      "grad_norm": 0.00016179266094695777,
      "learning_rate": 4.830959370059956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2676472,
      "step": 4710
    },
    {
      "epoch": 82.72566371681415,
      "grad_norm": 4.619023093255237e-05,
      "learning_rate": 4.830604318014472e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2679736,
      "step": 4715
    },
    {
      "epoch": 82.8141592920354,
      "grad_norm": 3.4993845474673435e-05,
      "learning_rate": 4.830248906560514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2682456,
      "step": 4720
    },
    {
      "epoch": 82.90265486725664,
      "grad_norm": 1.1618955795711372e-05,
      "learning_rate": 4.829893135752891e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2685880,
      "step": 4725
    },
    {
      "epoch": 82.99115044247787,
      "grad_norm": 1.8544489648775198e-05,
      "learning_rate": 4.829537005646466e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2688776,
      "step": 4730
    },
    {
      "epoch": 83.070796460177,
      "grad_norm": 7.345178801188013e-06,
      "learning_rate": 4.8291805162961615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2691032,
      "step": 4735
    },
    {
      "epoch": 83.15929203539822,
      "grad_norm": 1.2449928362912033e-05,
      "learning_rate": 4.82882366775695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2693656,
      "step": 4740
    },
    {
      "epoch": 83.24778761061947,
      "grad_norm": 2.1745314370491542e-05,
      "learning_rate": 4.828466460083864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2696248,
      "step": 4745
    },
    {
      "epoch": 83.33628318584071,
      "grad_norm": 9.38958692131564e-05,
      "learning_rate": 4.8281088933319877e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2699672,
      "step": 4750
    },
    {
      "epoch": 83.42477876106194,
      "grad_norm": 8.044791320571676e-05,
      "learning_rate": 4.827750967556464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2702344,
      "step": 4755
    },
    {
      "epoch": 83.51327433628319,
      "grad_norm": 1.8184142390964553e-05,
      "learning_rate": 4.827392682812488e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2705304,
      "step": 4760
    },
    {
      "epoch": 83.60176991150442,
      "grad_norm": 8.615345723228529e-05,
      "learning_rate": 4.827034039155312e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2708328,
      "step": 4765
    },
    {
      "epoch": 83.69026548672566,
      "grad_norm": 3.316084621474147e-05,
      "learning_rate": 4.8266750366402445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2711352,
      "step": 4770
    },
    {
      "epoch": 83.77876106194691,
      "grad_norm": 2.9917506253696047e-05,
      "learning_rate": 4.8263156753226476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2714312,
      "step": 4775
    },
    {
      "epoch": 83.86725663716814,
      "grad_norm": 1.215706561197294e-05,
      "learning_rate": 4.8259559552579394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2717544,
      "step": 4780
    },
    {
      "epoch": 83.95575221238938,
      "grad_norm": 1.612363485037349e-05,
      "learning_rate": 4.825595876501593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2720344,
      "step": 4785
    },
    {
      "epoch": 84.03539823008849,
      "grad_norm": 2.221361137344502e-05,
      "learning_rate": 4.825235439109137e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2722840,
      "step": 4790
    },
    {
      "epoch": 84.12389380530973,
      "grad_norm": 1.1388034181436524e-05,
      "learning_rate": 4.824874643136156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2725656,
      "step": 4795
    },
    {
      "epoch": 84.21238938053098,
      "grad_norm": 1.2712949683191255e-05,
      "learning_rate": 4.824513488638288e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2728488,
      "step": 4800
    },
    {
      "epoch": 84.21238938053098,
      "eval_loss": 0.3679148554801941,
      "eval_runtime": 1.0594,
      "eval_samples_per_second": 23.599,
      "eval_steps_per_second": 12.272,
      "num_input_tokens_seen": 2728488,
      "step": 4800
    },
    {
      "epoch": 84.30088495575221,
      "grad_norm": 5.040190080762841e-05,
      "learning_rate": 4.8241519756712293e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2731112,
      "step": 4805
    },
    {
      "epoch": 84.38938053097345,
      "grad_norm": 7.937336340546608e-05,
      "learning_rate": 4.8237901042907285e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2734328,
      "step": 4810
    },
    {
      "epoch": 84.47787610619469,
      "grad_norm": 5.904380032006884e-06,
      "learning_rate": 4.823427874552591e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2737272,
      "step": 4815
    },
    {
      "epoch": 84.56637168141593,
      "grad_norm": 6.868739092169562e-06,
      "learning_rate": 4.823065286512677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2740040,
      "step": 4820
    },
    {
      "epoch": 84.65486725663717,
      "grad_norm": 1.1667830221995246e-05,
      "learning_rate": 4.8227023402269025e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2743128,
      "step": 4825
    },
    {
      "epoch": 84.7433628318584,
      "grad_norm": 3.242742604925297e-05,
      "learning_rate": 4.822339035751239e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2745656,
      "step": 4830
    },
    {
      "epoch": 84.83185840707965,
      "grad_norm": 4.023673682240769e-05,
      "learning_rate": 4.8219753731417104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2748120,
      "step": 4835
    },
    {
      "epoch": 84.92035398230088,
      "grad_norm": 2.1143092453712597e-05,
      "learning_rate": 4.821611352454401e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2751528,
      "step": 4840
    },
    {
      "epoch": 85.0,
      "grad_norm": 1.5251353943313006e-05,
      "learning_rate": 4.8212469737454444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2754000,
      "step": 4845
    },
    {
      "epoch": 85.08849557522124,
      "grad_norm": 7.74105228629196e-06,
      "learning_rate": 4.820882237071035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2756992,
      "step": 4850
    },
    {
      "epoch": 85.17699115044248,
      "grad_norm": 3.054903936572373e-05,
      "learning_rate": 4.820517142487417e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2760000,
      "step": 4855
    },
    {
      "epoch": 85.26548672566372,
      "grad_norm": 7.285597530426458e-05,
      "learning_rate": 4.8201516900508956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2762704,
      "step": 4860
    },
    {
      "epoch": 85.35398230088495,
      "grad_norm": 7.96116582932882e-05,
      "learning_rate": 4.819785879817827e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2765568,
      "step": 4865
    },
    {
      "epoch": 85.4424778761062,
      "grad_norm": 3.978814129368402e-05,
      "learning_rate": 4.8194197118446226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2768592,
      "step": 4870
    },
    {
      "epoch": 85.53097345132744,
      "grad_norm": 2.2673961211694404e-05,
      "learning_rate": 4.819053186187752e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2771504,
      "step": 4875
    },
    {
      "epoch": 85.61946902654867,
      "grad_norm": 1.6942123693297617e-05,
      "learning_rate": 4.818686302903736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2774592,
      "step": 4880
    },
    {
      "epoch": 85.70796460176992,
      "grad_norm": 5.2469793445197865e-05,
      "learning_rate": 4.818319062049154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2777552,
      "step": 4885
    },
    {
      "epoch": 85.79646017699115,
      "grad_norm": 8.06311254564207e-06,
      "learning_rate": 4.817951463680639e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2780272,
      "step": 4890
    },
    {
      "epoch": 85.88495575221239,
      "grad_norm": 1.3982979908178095e-05,
      "learning_rate": 4.817583507854879e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2783008,
      "step": 4895
    },
    {
      "epoch": 85.97345132743362,
      "grad_norm": 1.5440047718584538e-05,
      "learning_rate": 4.817215194628617e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2785936,
      "step": 4900
    },
    {
      "epoch": 86.05309734513274,
      "grad_norm": 7.723529415670782e-05,
      "learning_rate": 4.816846524058653e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2788192,
      "step": 4905
    },
    {
      "epoch": 86.14159292035399,
      "grad_norm": 0.00033142176107503474,
      "learning_rate": 4.816477496201839e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2791344,
      "step": 4910
    },
    {
      "epoch": 86.23008849557522,
      "grad_norm": 6.418038537958637e-05,
      "learning_rate": 4.8161081111150845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2793792,
      "step": 4915
    },
    {
      "epoch": 86.31858407079646,
      "grad_norm": 3.394552550162189e-05,
      "learning_rate": 4.815738368855354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2796320,
      "step": 4920
    },
    {
      "epoch": 86.40707964601769,
      "grad_norm": 6.358473910950124e-05,
      "learning_rate": 4.815368269479664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2799360,
      "step": 4925
    },
    {
      "epoch": 86.49557522123894,
      "grad_norm": 3.2106967410072684e-05,
      "learning_rate": 4.814997813045092e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2801824,
      "step": 4930
    },
    {
      "epoch": 86.58407079646018,
      "grad_norm": 1.1631816960289143e-05,
      "learning_rate": 4.814626999608764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2804704,
      "step": 4935
    },
    {
      "epoch": 86.67256637168141,
      "grad_norm": 2.253058846690692e-05,
      "learning_rate": 4.814255829227865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2807424,
      "step": 4940
    },
    {
      "epoch": 86.76106194690266,
      "grad_norm": 0.00012762704864144325,
      "learning_rate": 4.813884301959635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2810304,
      "step": 4945
    },
    {
      "epoch": 86.84955752212389,
      "grad_norm": 4.0756433008937165e-05,
      "learning_rate": 4.813512417861368e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2813328,
      "step": 4950
    },
    {
      "epoch": 86.93805309734513,
      "grad_norm": 5.6250202760566026e-05,
      "learning_rate": 4.813140176990411e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2816432,
      "step": 4955
    },
    {
      "epoch": 87.01769911504425,
      "grad_norm": 6.595565992029151e-06,
      "learning_rate": 4.8127675794041714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2819264,
      "step": 4960
    },
    {
      "epoch": 87.10619469026548,
      "grad_norm": 2.7121843231725506e-05,
      "learning_rate": 4.812394625160107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2821984,
      "step": 4965
    },
    {
      "epoch": 87.19469026548673,
      "grad_norm": 1.830983819672838e-05,
      "learning_rate": 4.812021314315732e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2824784,
      "step": 4970
    },
    {
      "epoch": 87.28318584070796,
      "grad_norm": 1.5116307622520253e-05,
      "learning_rate": 4.811647646928616e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2827456,
      "step": 4975
    },
    {
      "epoch": 87.3716814159292,
      "grad_norm": 3.0028999390196986e-05,
      "learning_rate": 4.8112736230563814e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2830592,
      "step": 4980
    },
    {
      "epoch": 87.46017699115045,
      "grad_norm": 7.161133908084594e-06,
      "learning_rate": 4.81089924275671e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2833328,
      "step": 4985
    },
    {
      "epoch": 87.54867256637168,
      "grad_norm": 2.442136246827431e-05,
      "learning_rate": 4.810524506087335e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2836752,
      "step": 4990
    },
    {
      "epoch": 87.63716814159292,
      "grad_norm": 1.553621950733941e-05,
      "learning_rate": 4.810149413106044e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2840000,
      "step": 4995
    },
    {
      "epoch": 87.72566371681415,
      "grad_norm": 4.045459354529157e-05,
      "learning_rate": 4.809773963870684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2842656,
      "step": 5000
    },
    {
      "epoch": 87.72566371681415,
      "eval_loss": 0.3791765868663788,
      "eval_runtime": 1.064,
      "eval_samples_per_second": 23.497,
      "eval_steps_per_second": 12.218,
      "num_input_tokens_seen": 2842656,
      "step": 5000
    },
    {
      "epoch": 87.8141592920354,
      "grad_norm": 4.2096035031136125e-05,
      "learning_rate": 4.809398158439151e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2845328,
      "step": 5005
    },
    {
      "epoch": 87.90265486725664,
      "grad_norm": 1.175963734567631e-05,
      "learning_rate": 4.8090219968694005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2848560,
      "step": 5010
    },
    {
      "epoch": 87.99115044247787,
      "grad_norm": 3.539510316841188e-06,
      "learning_rate": 4.808645479219442e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2851200,
      "step": 5015
    },
    {
      "epoch": 88.070796460177,
      "grad_norm": 9.67841042438522e-05,
      "learning_rate": 4.8082686055473375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2853392,
      "step": 5020
    },
    {
      "epoch": 88.15929203539822,
      "grad_norm": 1.0930628377536777e-05,
      "learning_rate": 4.8078913759112066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2856016,
      "step": 5025
    },
    {
      "epoch": 88.24778761061947,
      "grad_norm": 1.4674860722152516e-05,
      "learning_rate": 4.807513790369223e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2859104,
      "step": 5030
    },
    {
      "epoch": 88.33628318584071,
      "grad_norm": 1.3145571756467689e-05,
      "learning_rate": 4.8071358489796145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2862016,
      "step": 5035
    },
    {
      "epoch": 88.42477876106194,
      "grad_norm": 3.888148785335943e-05,
      "learning_rate": 4.806757551800665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2864800,
      "step": 5040
    },
    {
      "epoch": 88.51327433628319,
      "grad_norm": 4.882104985881597e-05,
      "learning_rate": 4.806378898890713e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2867296,
      "step": 5045
    },
    {
      "epoch": 88.60176991150442,
      "grad_norm": 6.312512414297089e-05,
      "learning_rate": 4.80599989030815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2870608,
      "step": 5050
    },
    {
      "epoch": 88.69026548672566,
      "grad_norm": 1.6287229300360195e-05,
      "learning_rate": 4.805620526111426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2873472,
      "step": 5055
    },
    {
      "epoch": 88.77876106194691,
      "grad_norm": 2.7109961592941545e-05,
      "learning_rate": 4.805240806359042e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2876592,
      "step": 5060
    },
    {
      "epoch": 88.86725663716814,
      "grad_norm": 1.7756648958311416e-05,
      "learning_rate": 4.804860731109557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2879440,
      "step": 5065
    },
    {
      "epoch": 88.95575221238938,
      "grad_norm": 3.408586417208426e-05,
      "learning_rate": 4.804480300421581e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2882560,
      "step": 5070
    },
    {
      "epoch": 89.03539823008849,
      "grad_norm": 1.4548199033015408e-05,
      "learning_rate": 4.804099514353784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2884808,
      "step": 5075
    },
    {
      "epoch": 89.12389380530973,
      "grad_norm": 1.607229933142662e-05,
      "learning_rate": 4.8037183729648867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2887656,
      "step": 5080
    },
    {
      "epoch": 89.21238938053098,
      "grad_norm": 4.241734495735727e-05,
      "learning_rate": 4.803336876313666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2890552,
      "step": 5085
    },
    {
      "epoch": 89.30088495575221,
      "grad_norm": 2.2726804672856815e-05,
      "learning_rate": 4.802955024458953e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2893288,
      "step": 5090
    },
    {
      "epoch": 89.38938053097345,
      "grad_norm": 8.082717977231368e-05,
      "learning_rate": 4.802572817459634e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2896008,
      "step": 5095
    },
    {
      "epoch": 89.47787610619469,
      "grad_norm": 2.3283571863430552e-05,
      "learning_rate": 4.802190255374651e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2899096,
      "step": 5100
    },
    {
      "epoch": 89.56637168141593,
      "grad_norm": 2.4595767172286287e-05,
      "learning_rate": 4.801807338263e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2901544,
      "step": 5105
    },
    {
      "epoch": 89.65486725663717,
      "grad_norm": 5.713747668778524e-06,
      "learning_rate": 4.8014240661837306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2904632,
      "step": 5110
    },
    {
      "epoch": 89.7433628318584,
      "grad_norm": 5.9837035223608837e-05,
      "learning_rate": 4.80104043919595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2907992,
      "step": 5115
    },
    {
      "epoch": 89.83185840707965,
      "grad_norm": 5.357680947781773e-06,
      "learning_rate": 4.800656457358815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2911192,
      "step": 5120
    },
    {
      "epoch": 89.92035398230088,
      "grad_norm": 0.00010601630492601544,
      "learning_rate": 4.800272120731544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2913912,
      "step": 5125
    },
    {
      "epoch": 90.0,
      "grad_norm": 3.4932963899336755e-05,
      "learning_rate": 4.799887429373404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2916440,
      "step": 5130
    },
    {
      "epoch": 90.08849557522124,
      "grad_norm": 2.4866481908247806e-05,
      "learning_rate": 4.79950238334372e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2919000,
      "step": 5135
    },
    {
      "epoch": 90.17699115044248,
      "grad_norm": 2.852681245713029e-05,
      "learning_rate": 4.799116982701872e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2922520,
      "step": 5140
    },
    {
      "epoch": 90.26548672566372,
      "grad_norm": 5.2218412747606635e-05,
      "learning_rate": 4.7987312275072926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2925144,
      "step": 5145
    },
    {
      "epoch": 90.35398230088495,
      "grad_norm": 1.55226716742618e-05,
      "learning_rate": 4.79834511781947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2928216,
      "step": 5150
    },
    {
      "epoch": 90.4424778761062,
      "grad_norm": 5.423145921668038e-05,
      "learning_rate": 4.797958653697947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2931416,
      "step": 5155
    },
    {
      "epoch": 90.53097345132744,
      "grad_norm": 1.795066418708302e-05,
      "learning_rate": 4.7975718352023225e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2934360,
      "step": 5160
    },
    {
      "epoch": 90.61946902654867,
      "grad_norm": 1.4121006643108558e-05,
      "learning_rate": 4.7971846623922476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2937096,
      "step": 5165
    },
    {
      "epoch": 90.70796460176992,
      "grad_norm": 1.377313856210094e-05,
      "learning_rate": 4.7967971353274294e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2939720,
      "step": 5170
    },
    {
      "epoch": 90.79646017699115,
      "grad_norm": 5.845126725034788e-05,
      "learning_rate": 4.79640925406763e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2942648,
      "step": 5175
    },
    {
      "epoch": 90.88495575221239,
      "grad_norm": 4.910877032671124e-05,
      "learning_rate": 4.796021018672664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2945032,
      "step": 5180
    },
    {
      "epoch": 90.97345132743362,
      "grad_norm": 3.606249811127782e-05,
      "learning_rate": 4.795632429202405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2948184,
      "step": 5185
    },
    {
      "epoch": 91.05309734513274,
      "grad_norm": 4.422223355504684e-05,
      "learning_rate": 4.795243485716775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2950648,
      "step": 5190
    },
    {
      "epoch": 91.14159292035399,
      "grad_norm": 2.3311447876039892e-05,
      "learning_rate": 4.794854188275757e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2954088,
      "step": 5195
    },
    {
      "epoch": 91.23008849557522,
      "grad_norm": 1.1759426342905499e-05,
      "learning_rate": 4.794464536939384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2956824,
      "step": 5200
    },
    {
      "epoch": 91.23008849557522,
      "eval_loss": 0.3791448175907135,
      "eval_runtime": 1.0591,
      "eval_samples_per_second": 23.605,
      "eval_steps_per_second": 12.274,
      "num_input_tokens_seen": 2956824,
      "step": 5200
    },
    {
      "epoch": 91.31858407079646,
      "grad_norm": 3.365647353348322e-05,
      "learning_rate": 4.794074531767745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2959496,
      "step": 5205
    },
    {
      "epoch": 91.40707964601769,
      "grad_norm": 8.782851182331797e-06,
      "learning_rate": 4.7936841728209834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2962744,
      "step": 5210
    },
    {
      "epoch": 91.49557522123894,
      "grad_norm": 1.979790613404475e-05,
      "learning_rate": 4.7932934601593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2965576,
      "step": 5215
    },
    {
      "epoch": 91.58407079646018,
      "grad_norm": 1.2212344699946698e-05,
      "learning_rate": 4.792902393842943e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2968264,
      "step": 5220
    },
    {
      "epoch": 91.67256637168141,
      "grad_norm": 7.494259534723824e-06,
      "learning_rate": 4.792510973932225e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2971000,
      "step": 5225
    },
    {
      "epoch": 91.76106194690266,
      "grad_norm": 1.825256367737893e-05,
      "learning_rate": 4.7921192004875036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2973608,
      "step": 5230
    },
    {
      "epoch": 91.84955752212389,
      "grad_norm": 1.3610292626253795e-05,
      "learning_rate": 4.791727073569198e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2976184,
      "step": 5235
    },
    {
      "epoch": 91.93805309734513,
      "grad_norm": 2.9179303965065628e-05,
      "learning_rate": 4.7913345932377775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2979208,
      "step": 5240
    },
    {
      "epoch": 92.01769911504425,
      "grad_norm": 4.413542410475202e-05,
      "learning_rate": 4.790941759553769e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2981840,
      "step": 5245
    },
    {
      "epoch": 92.10619469026548,
      "grad_norm": 3.76336247427389e-05,
      "learning_rate": 4.79054857257775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2984448,
      "step": 5250
    },
    {
      "epoch": 92.19469026548673,
      "grad_norm": 1.8513643226469867e-05,
      "learning_rate": 4.790155032370357e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2987360,
      "step": 5255
    },
    {
      "epoch": 92.28318584070796,
      "grad_norm": 1.3011131159146316e-05,
      "learning_rate": 4.789761138992278e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2990048,
      "step": 5260
    },
    {
      "epoch": 92.3716814159292,
      "grad_norm": 1.4519916476274375e-05,
      "learning_rate": 4.7893668925042565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2992992,
      "step": 5265
    },
    {
      "epoch": 92.46017699115045,
      "grad_norm": 3.7923680793028325e-05,
      "learning_rate": 4.78897229296709e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2995824,
      "step": 5270
    },
    {
      "epoch": 92.54867256637168,
      "grad_norm": 8.074504876276478e-05,
      "learning_rate": 4.7885773404416315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 2998688,
      "step": 5275
    },
    {
      "epoch": 92.63716814159292,
      "grad_norm": 4.9451031372882426e-05,
      "learning_rate": 4.788182034988786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3001248,
      "step": 5280
    },
    {
      "epoch": 92.72566371681415,
      "grad_norm": 2.012345430557616e-05,
      "learning_rate": 4.787786376669516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3003856,
      "step": 5285
    },
    {
      "epoch": 92.8141592920354,
      "grad_norm": 1.1158133929711767e-05,
      "learning_rate": 4.787390365544837e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3007216,
      "step": 5290
    },
    {
      "epoch": 92.90265486725664,
      "grad_norm": 3.264218685217202e-05,
      "learning_rate": 4.786994001675818e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3010432,
      "step": 5295
    },
    {
      "epoch": 92.99115044247787,
      "grad_norm": 5.818254521727795e-06,
      "learning_rate": 4.786597285123584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3013056,
      "step": 5300
    },
    {
      "epoch": 93.070796460177,
      "grad_norm": 2.6617255571181886e-05,
      "learning_rate": 4.7862002159493135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3015736,
      "step": 5305
    },
    {
      "epoch": 93.15929203539822,
      "grad_norm": 4.8571160732535645e-05,
      "learning_rate": 4.785802794214239e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3018552,
      "step": 5310
    },
    {
      "epoch": 93.24778761061947,
      "grad_norm": 1.2826485544792376e-05,
      "learning_rate": 4.7854050199796495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3021416,
      "step": 5315
    },
    {
      "epoch": 93.33628318584071,
      "grad_norm": 7.14857378625311e-05,
      "learning_rate": 4.7850068933068845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3024168,
      "step": 5320
    },
    {
      "epoch": 93.42477876106194,
      "grad_norm": 2.6217334379907697e-05,
      "learning_rate": 4.7846084142573425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3026856,
      "step": 5325
    },
    {
      "epoch": 93.51327433628319,
      "grad_norm": 0.00013069502892903984,
      "learning_rate": 4.7842095828924725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3029864,
      "step": 5330
    },
    {
      "epoch": 93.60176991150442,
      "grad_norm": 1.4429737348109484e-05,
      "learning_rate": 4.783810399273779e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3032888,
      "step": 5335
    },
    {
      "epoch": 93.69026548672566,
      "grad_norm": 3.340968760312535e-05,
      "learning_rate": 4.7834108634628226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3035608,
      "step": 5340
    },
    {
      "epoch": 93.77876106194691,
      "grad_norm": 1.8348249795963056e-05,
      "learning_rate": 4.783010975521216e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3038536,
      "step": 5345
    },
    {
      "epoch": 93.86725663716814,
      "grad_norm": 9.572889211995061e-06,
      "learning_rate": 4.782610735510626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3040920,
      "step": 5350
    },
    {
      "epoch": 93.95575221238938,
      "grad_norm": 1.374269322695909e-05,
      "learning_rate": 4.782210143492776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3043688,
      "step": 5355
    },
    {
      "epoch": 94.03539823008849,
      "grad_norm": 7.907110557425767e-06,
      "learning_rate": 4.781809199529442e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3046240,
      "step": 5360
    },
    {
      "epoch": 94.12389380530973,
      "grad_norm": 4.691958110925043e-06,
      "learning_rate": 4.781407903682454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3049072,
      "step": 5365
    },
    {
      "epoch": 94.21238938053098,
      "grad_norm": 5.348283229977824e-05,
      "learning_rate": 4.781006256013698e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3052000,
      "step": 5370
    },
    {
      "epoch": 94.30088495575221,
      "grad_norm": 5.97368307353463e-05,
      "learning_rate": 4.7806042565851115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3054752,
      "step": 5375
    },
    {
      "epoch": 94.38938053097345,
      "grad_norm": 2.357465746172238e-05,
      "learning_rate": 4.7802019054586895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3057792,
      "step": 5380
    },
    {
      "epoch": 94.47787610619469,
      "grad_norm": 1.412731216987595e-05,
      "learning_rate": 4.779799202696479e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3060912,
      "step": 5385
    },
    {
      "epoch": 94.56637168141593,
      "grad_norm": 1.8307055142940953e-05,
      "learning_rate": 4.779396148360581e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3064096,
      "step": 5390
    },
    {
      "epoch": 94.65486725663717,
      "grad_norm": 5.0190526962978765e-05,
      "learning_rate": 4.7789927425131517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3066896,
      "step": 5395
    },
    {
      "epoch": 94.7433628318584,
      "grad_norm": 9.864623280009255e-05,
      "learning_rate": 4.778588985216403e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3069840,
      "step": 5400
    },
    {
      "epoch": 94.7433628318584,
      "eval_loss": 0.40042489767074585,
      "eval_runtime": 1.0597,
      "eval_samples_per_second": 23.591,
      "eval_steps_per_second": 12.267,
      "num_input_tokens_seen": 3069840,
      "step": 5400
    },
    {
      "epoch": 94.83185840707965,
      "grad_norm": 4.154091948294081e-05,
      "learning_rate": 4.778184876532598e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3072432,
      "step": 5405
    },
    {
      "epoch": 94.92035398230088,
      "grad_norm": 1.3770302757620811e-05,
      "learning_rate": 4.7777804165240556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3075152,
      "step": 5410
    },
    {
      "epoch": 95.0,
      "grad_norm": 6.43398889224045e-05,
      "learning_rate": 4.7773756052531485e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3077616,
      "step": 5415
    },
    {
      "epoch": 95.08849557522124,
      "grad_norm": 2.1216088498476893e-05,
      "learning_rate": 4.7769704427823035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3080304,
      "step": 5420
    },
    {
      "epoch": 95.17699115044248,
      "grad_norm": 6.695571937598288e-05,
      "learning_rate": 4.776564929174003e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3083168,
      "step": 5425
    },
    {
      "epoch": 95.26548672566372,
      "grad_norm": 7.0477362896781415e-06,
      "learning_rate": 4.7761590644907806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3086608,
      "step": 5430
    },
    {
      "epoch": 95.35398230088495,
      "grad_norm": 1.0432704584673047e-05,
      "learning_rate": 4.7757528487952263e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3089840,
      "step": 5435
    },
    {
      "epoch": 95.4424778761062,
      "grad_norm": 4.343217824498424e-06,
      "learning_rate": 4.7753462821499836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3092464,
      "step": 5440
    },
    {
      "epoch": 95.53097345132744,
      "grad_norm": 2.138469608325977e-05,
      "learning_rate": 4.774939364617751e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3095376,
      "step": 5445
    },
    {
      "epoch": 95.61946902654867,
      "grad_norm": 5.136169056640938e-05,
      "learning_rate": 4.7745320962612795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3098352,
      "step": 5450
    },
    {
      "epoch": 95.70796460176992,
      "grad_norm": 2.4683051378815435e-05,
      "learning_rate": 4.7741244771433756e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3100992,
      "step": 5455
    },
    {
      "epoch": 95.79646017699115,
      "grad_norm": 1.1291719602013472e-05,
      "learning_rate": 4.7737165073268985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3104016,
      "step": 5460
    },
    {
      "epoch": 95.88495575221239,
      "grad_norm": 1.6496873286087066e-05,
      "learning_rate": 4.7733081868747626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3107088,
      "step": 5465
    },
    {
      "epoch": 95.97345132743362,
      "grad_norm": 3.351855048094876e-05,
      "learning_rate": 4.772899515849936e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3109696,
      "step": 5470
    },
    {
      "epoch": 96.05309734513274,
      "grad_norm": 5.748741386923939e-06,
      "learning_rate": 4.7724904943154414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3111880,
      "step": 5475
    },
    {
      "epoch": 96.14159292035399,
      "grad_norm": 1.9345754481037147e-05,
      "learning_rate": 4.772081122334354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3114696,
      "step": 5480
    },
    {
      "epoch": 96.23008849557522,
      "grad_norm": 1.3439039321383461e-05,
      "learning_rate": 4.771671399969806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3118040,
      "step": 5485
    },
    {
      "epoch": 96.31858407079646,
      "grad_norm": 2.5619841835577972e-05,
      "learning_rate": 4.7712613272849794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3120856,
      "step": 5490
    },
    {
      "epoch": 96.40707964601769,
      "grad_norm": 1.760073791956529e-05,
      "learning_rate": 4.770850904343114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3123560,
      "step": 5495
    },
    {
      "epoch": 96.49557522123894,
      "grad_norm": 8.377748599741608e-06,
      "learning_rate": 4.770440131207502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3126200,
      "step": 5500
    },
    {
      "epoch": 96.58407079646018,
      "grad_norm": 1.4422633284993935e-05,
      "learning_rate": 4.7700290079414896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3129336,
      "step": 5505
    },
    {
      "epoch": 96.67256637168141,
      "grad_norm": 5.663756837748224e-06,
      "learning_rate": 4.769617534608477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3132296,
      "step": 5510
    },
    {
      "epoch": 96.76106194690266,
      "grad_norm": 1.2365678230708e-05,
      "learning_rate": 4.7692057112719193e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3135160,
      "step": 5515
    },
    {
      "epoch": 96.84955752212389,
      "grad_norm": 6.519709131680429e-06,
      "learning_rate": 4.7687935379953234e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3137912,
      "step": 5520
    },
    {
      "epoch": 96.93805309734513,
      "grad_norm": 1.1308909961371683e-05,
      "learning_rate": 4.7683810148422534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3140872,
      "step": 5525
    },
    {
      "epoch": 97.01769911504425,
      "grad_norm": 2.649181078595575e-05,
      "learning_rate": 4.767968141876324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3143568,
      "step": 5530
    },
    {
      "epoch": 97.10619469026548,
      "grad_norm": 5.190384035813622e-05,
      "learning_rate": 4.767554919161207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3146368,
      "step": 5535
    },
    {
      "epoch": 97.19469026548673,
      "grad_norm": 5.0931807891174685e-06,
      "learning_rate": 4.767141346760624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3149040,
      "step": 5540
    },
    {
      "epoch": 97.28318584070796,
      "grad_norm": 6.718244094372494e-06,
      "learning_rate": 4.766727424738356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3152224,
      "step": 5545
    },
    {
      "epoch": 97.3716814159292,
      "grad_norm": 6.433582166209817e-05,
      "learning_rate": 4.7663131531582325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3155216,
      "step": 5550
    },
    {
      "epoch": 97.46017699115045,
      "grad_norm": 1.481386243540328e-05,
      "learning_rate": 4.765898532084142e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3157984,
      "step": 5555
    },
    {
      "epoch": 97.54867256637168,
      "grad_norm": 4.322625318309292e-05,
      "learning_rate": 4.765483561580022e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3160928,
      "step": 5560
    },
    {
      "epoch": 97.63716814159292,
      "grad_norm": 9.257161582354456e-05,
      "learning_rate": 4.7650682417098666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3163920,
      "step": 5565
    },
    {
      "epoch": 97.72566371681415,
      "grad_norm": 1.1740916306735016e-05,
      "learning_rate": 4.7646525725377244e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3166416,
      "step": 5570
    },
    {
      "epoch": 97.8141592920354,
      "grad_norm": 2.4235252567450516e-05,
      "learning_rate": 4.764236554127696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3169456,
      "step": 5575
    },
    {
      "epoch": 97.90265486725664,
      "grad_norm": 7.174142956500873e-05,
      "learning_rate": 4.7638201865439356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3172208,
      "step": 5580
    },
    {
      "epoch": 97.99115044247787,
      "grad_norm": 1.0749962711997796e-05,
      "learning_rate": 4.7634034698506545e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3175584,
      "step": 5585
    },
    {
      "epoch": 98.070796460177,
      "grad_norm": 6.481282616732642e-05,
      "learning_rate": 4.762986404112115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3178096,
      "step": 5590
    },
    {
      "epoch": 98.15929203539822,
      "grad_norm": 9.557444172969554e-06,
      "learning_rate": 4.762568989392633e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3181152,
      "step": 5595
    },
    {
      "epoch": 98.24778761061947,
      "grad_norm": 0.00011830249422928318,
      "learning_rate": 4.76215122575658e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3183600,
      "step": 5600
    },
    {
      "epoch": 98.24778761061947,
      "eval_loss": 0.3896600306034088,
      "eval_runtime": 1.0623,
      "eval_samples_per_second": 23.533,
      "eval_steps_per_second": 12.237,
      "num_input_tokens_seen": 3183600,
      "step": 5600
    },
    {
      "epoch": 98.33628318584071,
      "grad_norm": 8.488675666740164e-05,
      "learning_rate": 4.7617331132683795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3186528,
      "step": 5605
    },
    {
      "epoch": 98.42477876106194,
      "grad_norm": 3.1886243959888816e-05,
      "learning_rate": 4.7613146519925105e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3189840,
      "step": 5610
    },
    {
      "epoch": 98.51327433628319,
      "grad_norm": 2.3559088731417432e-05,
      "learning_rate": 4.7608958419935045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3192512,
      "step": 5615
    },
    {
      "epoch": 98.60176991150442,
      "grad_norm": 6.522976036649197e-06,
      "learning_rate": 4.760476683335948e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3195648,
      "step": 5620
    },
    {
      "epoch": 98.69026548672566,
      "grad_norm": 1.3234563084552065e-05,
      "learning_rate": 4.760057176084479e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3198160,
      "step": 5625
    },
    {
      "epoch": 98.77876106194691,
      "grad_norm": 3.844524690066464e-05,
      "learning_rate": 4.759637320303793e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3201184,
      "step": 5630
    },
    {
      "epoch": 98.86725663716814,
      "grad_norm": 1.0644357644196134e-05,
      "learning_rate": 4.759217116058635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3204016,
      "step": 5635
    },
    {
      "epoch": 98.95575221238938,
      "grad_norm": 5.853835318703204e-06,
      "learning_rate": 4.758796563413807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3206624,
      "step": 5640
    },
    {
      "epoch": 99.03539823008849,
      "grad_norm": 8.31638590170769e-06,
      "learning_rate": 4.758375662434163e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3208856,
      "step": 5645
    },
    {
      "epoch": 99.12389380530973,
      "grad_norm": 8.623896974313539e-06,
      "learning_rate": 4.7579544131846114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3211992,
      "step": 5650
    },
    {
      "epoch": 99.21238938053098,
      "grad_norm": 2.1051338990218937e-05,
      "learning_rate": 4.757532815730114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3215064,
      "step": 5655
    },
    {
      "epoch": 99.30088495575221,
      "grad_norm": 1.671749123488553e-05,
      "learning_rate": 4.7571108701356865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3217736,
      "step": 5660
    },
    {
      "epoch": 99.38938053097345,
      "grad_norm": 2.8265611035749316e-05,
      "learning_rate": 4.756688576466398e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3220600,
      "step": 5665
    },
    {
      "epoch": 99.47787610619469,
      "grad_norm": 6.922745797055541e-06,
      "learning_rate": 4.756265934787372e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3223592,
      "step": 5670
    },
    {
      "epoch": 99.56637168141593,
      "grad_norm": 1.6587553545832634e-05,
      "learning_rate": 4.755842945163785e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3226248,
      "step": 5675
    },
    {
      "epoch": 99.65486725663717,
      "grad_norm": 4.542443548416486e-06,
      "learning_rate": 4.755419607660867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3229432,
      "step": 5680
    },
    {
      "epoch": 99.7433628318584,
      "grad_norm": 9.394648259331007e-06,
      "learning_rate": 4.7549959223439016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3231896,
      "step": 5685
    },
    {
      "epoch": 99.83185840707965,
      "grad_norm": 6.802508323744405e-06,
      "learning_rate": 4.754571889278228e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3235032,
      "step": 5690
    },
    {
      "epoch": 99.92035398230088,
      "grad_norm": 2.026318725256715e-05,
      "learning_rate": 4.754147508529235e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3237752,
      "step": 5695
    },
    {
      "epoch": 100.0,
      "grad_norm": 4.2263567593181506e-05,
      "learning_rate": 4.75372278016237e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3239888,
      "step": 5700
    },
    {
      "epoch": 100.08849557522124,
      "grad_norm": 3.1615487387171015e-05,
      "learning_rate": 4.753297704243129e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3242672,
      "step": 5705
    },
    {
      "epoch": 100.17699115044248,
      "grad_norm": 9.407552170159761e-06,
      "learning_rate": 4.752872280837066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3245408,
      "step": 5710
    },
    {
      "epoch": 100.26548672566372,
      "grad_norm": 9.638358460506424e-05,
      "learning_rate": 4.752446510009786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3248592,
      "step": 5715
    },
    {
      "epoch": 100.35398230088495,
      "grad_norm": 1.7459769878769293e-05,
      "learning_rate": 4.7520203918269476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3251072,
      "step": 5720
    },
    {
      "epoch": 100.4424778761062,
      "grad_norm": 3.0455888918368146e-05,
      "learning_rate": 4.751593926354265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3253872,
      "step": 5725
    },
    {
      "epoch": 100.53097345132744,
      "grad_norm": 1.7095733710448258e-05,
      "learning_rate": 4.751167113657503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3256752,
      "step": 5730
    },
    {
      "epoch": 100.61946902654867,
      "grad_norm": 9.835976015892811e-06,
      "learning_rate": 4.7507399538024834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3259584,
      "step": 5735
    },
    {
      "epoch": 100.70796460176992,
      "grad_norm": 1.0335240403946955e-05,
      "learning_rate": 4.750312446855077e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3263120,
      "step": 5740
    },
    {
      "epoch": 100.79646017699115,
      "grad_norm": 1.645578049647156e-05,
      "learning_rate": 4.749884592881212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3266032,
      "step": 5745
    },
    {
      "epoch": 100.88495575221239,
      "grad_norm": 4.043260560138151e-06,
      "learning_rate": 4.74945639194687e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3268736,
      "step": 5750
    },
    {
      "epoch": 100.97345132743362,
      "grad_norm": 1.3589383343060035e-05,
      "learning_rate": 4.749027844118083e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3272016,
      "step": 5755
    },
    {
      "epoch": 101.05309734513274,
      "grad_norm": 1.700009306659922e-05,
      "learning_rate": 4.7485989494609395e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3274392,
      "step": 5760
    },
    {
      "epoch": 101.14159292035399,
      "grad_norm": 3.7920566683169454e-05,
      "learning_rate": 4.748169708041581e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3276952,
      "step": 5765
    },
    {
      "epoch": 101.23008849557522,
      "grad_norm": 2.2796521079726517e-05,
      "learning_rate": 4.7477401199262004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3280104,
      "step": 5770
    },
    {
      "epoch": 101.31858407079646,
      "grad_norm": 2.8589773137355223e-05,
      "learning_rate": 4.747310185181048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3282840,
      "step": 5775
    },
    {
      "epoch": 101.40707964601769,
      "grad_norm": 1.6590882296441123e-05,
      "learning_rate": 4.746879903872422e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3285672,
      "step": 5780
    },
    {
      "epoch": 101.49557522123894,
      "grad_norm": 5.515978955372702e-06,
      "learning_rate": 4.746449276066679e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3288872,
      "step": 5785
    },
    {
      "epoch": 101.58407079646018,
      "grad_norm": 4.0277940570376813e-05,
      "learning_rate": 4.746018301830227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3292264,
      "step": 5790
    },
    {
      "epoch": 101.67256637168141,
      "grad_norm": 1.1031048416043632e-05,
      "learning_rate": 4.7455869812295275e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3294808,
      "step": 5795
    },
    {
      "epoch": 101.76106194690266,
      "grad_norm": 4.572404577629641e-05,
      "learning_rate": 4.7451553143310964e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3297896,
      "step": 5800
    },
    {
      "epoch": 101.76106194690266,
      "eval_loss": 0.38237905502319336,
      "eval_runtime": 1.0598,
      "eval_samples_per_second": 23.59,
      "eval_steps_per_second": 12.267,
      "num_input_tokens_seen": 3297896,
      "step": 5800
    },
    {
      "epoch": 101.84955752212389,
      "grad_norm": 4.717907813756028e-06,
      "learning_rate": 4.744723301201501e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3300648,
      "step": 5805
    },
    {
      "epoch": 101.93805309734513,
      "grad_norm": 1.2737408724206034e-05,
      "learning_rate": 4.744290941907364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3303400,
      "step": 5810
    },
    {
      "epoch": 102.01769911504425,
      "grad_norm": 8.844261174090207e-06,
      "learning_rate": 4.7438582365153594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3305792,
      "step": 5815
    },
    {
      "epoch": 102.10619469026548,
      "grad_norm": 1.909842831082642e-05,
      "learning_rate": 4.743425185092217e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3308880,
      "step": 5820
    },
    {
      "epoch": 102.19469026548673,
      "grad_norm": 9.510061317996588e-06,
      "learning_rate": 4.742991787704719e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3312368,
      "step": 5825
    },
    {
      "epoch": 102.28318584070796,
      "grad_norm": 0.00010796192509587854,
      "learning_rate": 4.7425580444196994e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3315072,
      "step": 5830
    },
    {
      "epoch": 102.3716814159292,
      "grad_norm": 2.196704372181557e-05,
      "learning_rate": 4.742123955304048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3317728,
      "step": 5835
    },
    {
      "epoch": 102.46017699115045,
      "grad_norm": 4.387954959383933e-06,
      "learning_rate": 4.741689520424706e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3320592,
      "step": 5840
    },
    {
      "epoch": 102.54867256637168,
      "grad_norm": 1.766483546816744e-05,
      "learning_rate": 4.741254739848669e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3322976,
      "step": 5845
    },
    {
      "epoch": 102.63716814159292,
      "grad_norm": 9.123917152464855e-06,
      "learning_rate": 4.740819613642987e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3325872,
      "step": 5850
    },
    {
      "epoch": 102.72566371681415,
      "grad_norm": 2.4916011170716956e-05,
      "learning_rate": 4.74038414187476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3328432,
      "step": 5855
    },
    {
      "epoch": 102.8141592920354,
      "grad_norm": 2.3639850041945465e-05,
      "learning_rate": 4.739948324611144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3331600,
      "step": 5860
    },
    {
      "epoch": 102.90265486725664,
      "grad_norm": 1.2081855857104529e-05,
      "learning_rate": 4.7395121619193465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3335040,
      "step": 5865
    },
    {
      "epoch": 102.99115044247787,
      "grad_norm": 8.577361768402625e-06,
      "learning_rate": 4.7390756538666313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3337920,
      "step": 5870
    },
    {
      "epoch": 103.070796460177,
      "grad_norm": 8.881267604010645e-06,
      "learning_rate": 4.738638800520311e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3340552,
      "step": 5875
    },
    {
      "epoch": 103.15929203539822,
      "grad_norm": 1.3608810149889905e-05,
      "learning_rate": 4.738201601947757e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3342904,
      "step": 5880
    },
    {
      "epoch": 103.24778761061947,
      "grad_norm": 8.550871825718787e-06,
      "learning_rate": 4.7377640582163876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3345480,
      "step": 5885
    },
    {
      "epoch": 103.33628318584071,
      "grad_norm": 6.212937296368182e-05,
      "learning_rate": 4.7373261693936786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3348456,
      "step": 5890
    },
    {
      "epoch": 103.42477876106194,
      "grad_norm": 7.340172032854753e-06,
      "learning_rate": 4.7368879355471595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3351000,
      "step": 5895
    },
    {
      "epoch": 103.51327433628319,
      "grad_norm": 1.5633642760803923e-05,
      "learning_rate": 4.736449356744409e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3353624,
      "step": 5900
    },
    {
      "epoch": 103.60176991150442,
      "grad_norm": 0.00012720983067993075,
      "learning_rate": 4.736010433053064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3356504,
      "step": 5905
    },
    {
      "epoch": 103.69026548672566,
      "grad_norm": 5.69731992072775e-06,
      "learning_rate": 4.73557116454081e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3359080,
      "step": 5910
    },
    {
      "epoch": 103.77876106194691,
      "grad_norm": 1.2290623999433592e-05,
      "learning_rate": 4.735131551275389e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3362216,
      "step": 5915
    },
    {
      "epoch": 103.86725663716814,
      "grad_norm": 1.5073249414854217e-05,
      "learning_rate": 4.734691593324594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3365960,
      "step": 5920
    },
    {
      "epoch": 103.95575221238938,
      "grad_norm": 4.5537003643403295e-06,
      "learning_rate": 4.734251290756272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3369192,
      "step": 5925
    },
    {
      "epoch": 104.03539823008849,
      "grad_norm": 9.66015795711428e-06,
      "learning_rate": 4.7338106436383246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3371848,
      "step": 5930
    },
    {
      "epoch": 104.12389380530973,
      "grad_norm": 5.031836735724937e-06,
      "learning_rate": 4.733369652038703e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3374792,
      "step": 5935
    },
    {
      "epoch": 104.21238938053098,
      "grad_norm": 1.2557297850435134e-05,
      "learning_rate": 4.7329283160254156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3377816,
      "step": 5940
    },
    {
      "epoch": 104.30088495575221,
      "grad_norm": 1.0829078746610321e-05,
      "learning_rate": 4.732486635666521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3380808,
      "step": 5945
    },
    {
      "epoch": 104.38938053097345,
      "grad_norm": 1.864486694103107e-05,
      "learning_rate": 4.732044611030132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3384008,
      "step": 5950
    },
    {
      "epoch": 104.47787610619469,
      "grad_norm": 0.00013746856711804867,
      "learning_rate": 4.731602242184414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3386472,
      "step": 5955
    },
    {
      "epoch": 104.56637168141593,
      "grad_norm": 8.560599235352129e-05,
      "learning_rate": 4.7311595291975864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3389272,
      "step": 5960
    },
    {
      "epoch": 104.65486725663717,
      "grad_norm": 6.508545084216166e-06,
      "learning_rate": 4.7307164721379216e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3392392,
      "step": 5965
    },
    {
      "epoch": 104.7433628318584,
      "grad_norm": 1.0374439625593368e-05,
      "learning_rate": 4.730273071073743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3395304,
      "step": 5970
    },
    {
      "epoch": 104.83185840707965,
      "grad_norm": 1.1811460353783332e-05,
      "learning_rate": 4.729829326073429e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3397896,
      "step": 5975
    },
    {
      "epoch": 104.92035398230088,
      "grad_norm": 4.131334935664199e-05,
      "learning_rate": 4.7293852372054126e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3400552,
      "step": 5980
    },
    {
      "epoch": 105.0,
      "grad_norm": 7.966938028403092e-06,
      "learning_rate": 4.728940804538176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3402936,
      "step": 5985
    },
    {
      "epoch": 105.08849557522124,
      "grad_norm": 3.310256579425186e-05,
      "learning_rate": 4.7284960281402556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3405640,
      "step": 5990
    },
    {
      "epoch": 105.17699115044248,
      "grad_norm": 5.4261668083199766e-06,
      "learning_rate": 4.728050908080244e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3408232,
      "step": 5995
    },
    {
      "epoch": 105.26548672566372,
      "grad_norm": 7.524590273533249e-06,
      "learning_rate": 4.727605444426782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3411544,
      "step": 6000
    },
    {
      "epoch": 105.26548672566372,
      "eval_loss": 0.38351133465766907,
      "eval_runtime": 1.0618,
      "eval_samples_per_second": 23.544,
      "eval_steps_per_second": 12.243,
      "num_input_tokens_seen": 3411544,
      "step": 6000
    },
    {
      "epoch": 105.35398230088495,
      "grad_norm": 1.948706994880922e-05,
      "learning_rate": 4.727159637248567e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3414664,
      "step": 6005
    },
    {
      "epoch": 105.4424778761062,
      "grad_norm": 1.0981922969222069e-05,
      "learning_rate": 4.7267134866143474e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3417528,
      "step": 6010
    },
    {
      "epoch": 105.53097345132744,
      "grad_norm": 7.860393452574499e-06,
      "learning_rate": 4.726266992592926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3420408,
      "step": 6015
    },
    {
      "epoch": 105.61946902654867,
      "grad_norm": 5.101440183352679e-05,
      "learning_rate": 4.725820155253157e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3423272,
      "step": 6020
    },
    {
      "epoch": 105.70796460176992,
      "grad_norm": 1.0177225703955628e-05,
      "learning_rate": 4.725372974663948e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3426264,
      "step": 6025
    },
    {
      "epoch": 105.79646017699115,
      "grad_norm": 6.668170044576982e-06,
      "learning_rate": 4.724925450894262e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3429304,
      "step": 6030
    },
    {
      "epoch": 105.88495575221239,
      "grad_norm": 7.531664323323639e-06,
      "learning_rate": 4.72447758401311e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3432120,
      "step": 6035
    },
    {
      "epoch": 105.97345132743362,
      "grad_norm": 7.362277392530814e-05,
      "learning_rate": 4.7240293740895616e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3434552,
      "step": 6040
    },
    {
      "epoch": 106.05309734513274,
      "grad_norm": 1.9240425899624825e-05,
      "learning_rate": 4.723580821192733e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3436568,
      "step": 6045
    },
    {
      "epoch": 106.14159292035399,
      "grad_norm": 7.921392352727707e-06,
      "learning_rate": 4.7231319253917996e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3440088,
      "step": 6050
    },
    {
      "epoch": 106.23008849557522,
      "grad_norm": 5.790775503555778e-06,
      "learning_rate": 4.722682686755986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3443400,
      "step": 6055
    },
    {
      "epoch": 106.31858407079646,
      "grad_norm": 5.420154593593907e-06,
      "learning_rate": 4.722233105354569e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3446120,
      "step": 6060
    },
    {
      "epoch": 106.40707964601769,
      "grad_norm": 7.841806109354366e-06,
      "learning_rate": 4.7217831812568815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3448520,
      "step": 6065
    },
    {
      "epoch": 106.49557522123894,
      "grad_norm": 3.772136915358715e-05,
      "learning_rate": 4.721332914532307e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3451192,
      "step": 6070
    },
    {
      "epoch": 106.58407079646018,
      "grad_norm": 6.516958364954917e-06,
      "learning_rate": 4.720882305250281e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3454376,
      "step": 6075
    },
    {
      "epoch": 106.67256637168141,
      "grad_norm": 3.961460606660694e-05,
      "learning_rate": 4.720431353480295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3456872,
      "step": 6080
    },
    {
      "epoch": 106.76106194690266,
      "grad_norm": 5.712998245144263e-06,
      "learning_rate": 4.719980059291891e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3459512,
      "step": 6085
    },
    {
      "epoch": 106.84955752212389,
      "grad_norm": 2.1741310774814337e-05,
      "learning_rate": 4.7195284227546634e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3462472,
      "step": 6090
    },
    {
      "epoch": 106.93805309734513,
      "grad_norm": 1.2835181223636027e-05,
      "learning_rate": 4.7190764439382604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3465096,
      "step": 6095
    },
    {
      "epoch": 107.01769911504425,
      "grad_norm": 4.321036612964235e-06,
      "learning_rate": 4.7186241229123826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3467568,
      "step": 6100
    },
    {
      "epoch": 107.10619469026548,
      "grad_norm": 2.4422590286121704e-05,
      "learning_rate": 4.718171459746785e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3470832,
      "step": 6105
    },
    {
      "epoch": 107.19469026548673,
      "grad_norm": 4.812156475964002e-06,
      "learning_rate": 4.717718454511273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3474032,
      "step": 6110
    },
    {
      "epoch": 107.28318584070796,
      "grad_norm": 6.999730885581812e-06,
      "learning_rate": 4.7172651072757056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3476944,
      "step": 6115
    },
    {
      "epoch": 107.3716814159292,
      "grad_norm": 2.569380285422085e-06,
      "learning_rate": 4.7168114181099945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3479840,
      "step": 6120
    },
    {
      "epoch": 107.46017699115045,
      "grad_norm": 1.8267222912982106e-05,
      "learning_rate": 4.716357387084105e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3482960,
      "step": 6125
    },
    {
      "epoch": 107.54867256637168,
      "grad_norm": 2.5757448383956216e-05,
      "learning_rate": 4.715903014268054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3485280,
      "step": 6130
    },
    {
      "epoch": 107.63716814159292,
      "grad_norm": 5.855270956089953e-06,
      "learning_rate": 4.715448299731911e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3488304,
      "step": 6135
    },
    {
      "epoch": 107.72566371681415,
      "grad_norm": 8.74169563758187e-06,
      "learning_rate": 4.7149932435457986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3491232,
      "step": 6140
    },
    {
      "epoch": 107.8141592920354,
      "grad_norm": 8.729279215913266e-06,
      "learning_rate": 4.714537845779894e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3494368,
      "step": 6145
    },
    {
      "epoch": 107.90265486725664,
      "grad_norm": 3.3053515835490543e-06,
      "learning_rate": 4.714082106504423e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3496912,
      "step": 6150
    },
    {
      "epoch": 107.99115044247787,
      "grad_norm": 3.395170324438368e-06,
      "learning_rate": 4.713626025789667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3499632,
      "step": 6155
    },
    {
      "epoch": 108.070796460177,
      "grad_norm": 6.291446879913565e-06,
      "learning_rate": 4.7131696037059606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3502320,
      "step": 6160
    },
    {
      "epoch": 108.15929203539822,
      "grad_norm": 1.1250343050051015e-05,
      "learning_rate": 4.712712840323689e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3505104,
      "step": 6165
    },
    {
      "epoch": 108.24778761061947,
      "grad_norm": 1.826532752602361e-05,
      "learning_rate": 4.71225573571329e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3507744,
      "step": 6170
    },
    {
      "epoch": 108.33628318584071,
      "grad_norm": 3.278957592556253e-06,
      "learning_rate": 4.711798289945256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3510848,
      "step": 6175
    },
    {
      "epoch": 108.42477876106194,
      "grad_norm": 6.852121714473469e-06,
      "learning_rate": 4.71134050309013e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3513952,
      "step": 6180
    },
    {
      "epoch": 108.51327433628319,
      "grad_norm": 2.623460750328377e-05,
      "learning_rate": 4.710882375218509e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3516992,
      "step": 6185
    },
    {
      "epoch": 108.60176991150442,
      "grad_norm": 4.540392183116637e-06,
      "learning_rate": 4.7104239064010424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3519952,
      "step": 6190
    },
    {
      "epoch": 108.69026548672566,
      "grad_norm": 2.5419726625841577e-06,
      "learning_rate": 4.709965096708432e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3522688,
      "step": 6195
    },
    {
      "epoch": 108.77876106194691,
      "grad_norm": 2.3001886802376248e-05,
      "learning_rate": 4.709505946211431e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3525472,
      "step": 6200
    },
    {
      "epoch": 108.77876106194691,
      "eval_loss": 0.3907127380371094,
      "eval_runtime": 1.0577,
      "eval_samples_per_second": 23.636,
      "eval_steps_per_second": 12.291,
      "num_input_tokens_seen": 3525472,
      "step": 6200
    },
    {
      "epoch": 108.86725663716814,
      "grad_norm": 7.186144557635998e-06,
      "learning_rate": 4.709046454980846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3528720,
      "step": 6205
    },
    {
      "epoch": 108.95575221238938,
      "grad_norm": 9.92341938399477e-06,
      "learning_rate": 4.708586623087538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3531264,
      "step": 6210
    },
    {
      "epoch": 109.03539823008849,
      "grad_norm": 7.653531611140352e-06,
      "learning_rate": 4.708126450602418e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3533456,
      "step": 6215
    },
    {
      "epoch": 109.12389380530973,
      "grad_norm": 9.38008633966092e-06,
      "learning_rate": 4.7076659375964495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3536624,
      "step": 6220
    },
    {
      "epoch": 109.21238938053098,
      "grad_norm": 2.3135555238695815e-05,
      "learning_rate": 4.707205084140651e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3539520,
      "step": 6225
    },
    {
      "epoch": 109.30088495575221,
      "grad_norm": 1.6371499441447668e-05,
      "learning_rate": 4.7067438903060904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3542160,
      "step": 6230
    },
    {
      "epoch": 109.38938053097345,
      "grad_norm": 8.586283911427017e-06,
      "learning_rate": 4.70628235616389e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3544944,
      "step": 6235
    },
    {
      "epoch": 109.47787610619469,
      "grad_norm": 2.834166934917448e-06,
      "learning_rate": 4.7058204817852256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3547424,
      "step": 6240
    },
    {
      "epoch": 109.56637168141593,
      "grad_norm": 4.9272104661213234e-05,
      "learning_rate": 4.705358267241322e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3550416,
      "step": 6245
    },
    {
      "epoch": 109.65486725663717,
      "grad_norm": 1.9514814994181506e-05,
      "learning_rate": 4.704895712603459e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3553728,
      "step": 6250
    },
    {
      "epoch": 109.7433628318584,
      "grad_norm": 5.006486844649771e-06,
      "learning_rate": 4.704432817942969e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3556992,
      "step": 6255
    },
    {
      "epoch": 109.83185840707965,
      "grad_norm": 4.1510897972329985e-06,
      "learning_rate": 4.703969583331236e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3559776,
      "step": 6260
    },
    {
      "epoch": 109.92035398230088,
      "grad_norm": 3.1010570182843367e-06,
      "learning_rate": 4.7035060088396965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3562640,
      "step": 6265
    },
    {
      "epoch": 110.0,
      "grad_norm": 3.3161541068693623e-06,
      "learning_rate": 4.703042094539839e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3565040,
      "step": 6270
    },
    {
      "epoch": 110.08849557522124,
      "grad_norm": 1.774953125277534e-05,
      "learning_rate": 4.702577840503206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3567632,
      "step": 6275
    },
    {
      "epoch": 110.17699115044248,
      "grad_norm": 1.142226938100066e-05,
      "learning_rate": 4.70211324680139e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3570816,
      "step": 6280
    },
    {
      "epoch": 110.26548672566372,
      "grad_norm": 7.527910838689422e-06,
      "learning_rate": 4.7016483135060386e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3573392,
      "step": 6285
    },
    {
      "epoch": 110.35398230088495,
      "grad_norm": 4.026141141366679e-06,
      "learning_rate": 4.701183040688849e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3576192,
      "step": 6290
    },
    {
      "epoch": 110.4424778761062,
      "grad_norm": 6.408405170077458e-05,
      "learning_rate": 4.700717428421573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3578928,
      "step": 6295
    },
    {
      "epoch": 110.53097345132744,
      "grad_norm": 8.95859284355538e-06,
      "learning_rate": 4.700251476776014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3581680,
      "step": 6300
    },
    {
      "epoch": 110.61946902654867,
      "grad_norm": 6.169352218421409e-06,
      "learning_rate": 4.699785185824026e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3584880,
      "step": 6305
    },
    {
      "epoch": 110.70796460176992,
      "grad_norm": 2.632413452374749e-05,
      "learning_rate": 4.699318555637519e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3587760,
      "step": 6310
    },
    {
      "epoch": 110.79646017699115,
      "grad_norm": 3.919938535545953e-06,
      "learning_rate": 4.6988515862884525e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3590752,
      "step": 6315
    },
    {
      "epoch": 110.88495575221239,
      "grad_norm": 6.077697889850242e-06,
      "learning_rate": 4.698384277848838e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3593648,
      "step": 6320
    },
    {
      "epoch": 110.97345132743362,
      "grad_norm": 2.0001782104372978e-05,
      "learning_rate": 4.6979166303907425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3596384,
      "step": 6325
    },
    {
      "epoch": 111.05309734513274,
      "grad_norm": 4.147494837525301e-05,
      "learning_rate": 4.697448643986281e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3598640,
      "step": 6330
    },
    {
      "epoch": 111.14159292035399,
      "grad_norm": 3.271522291470319e-05,
      "learning_rate": 4.696980318707624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3601328,
      "step": 6335
    },
    {
      "epoch": 111.23008849557522,
      "grad_norm": 9.23617699299939e-06,
      "learning_rate": 4.6965116546269924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3604192,
      "step": 6340
    },
    {
      "epoch": 111.31858407079646,
      "grad_norm": 2.576905899331905e-05,
      "learning_rate": 4.6960426518166615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3606800,
      "step": 6345
    },
    {
      "epoch": 111.40707964601769,
      "grad_norm": 1.1237275430175941e-05,
      "learning_rate": 4.6955733103489556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3609728,
      "step": 6350
    },
    {
      "epoch": 111.49557522123894,
      "grad_norm": 1.1655658454401419e-05,
      "learning_rate": 4.695103630296255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3612480,
      "step": 6355
    },
    {
      "epoch": 111.58407079646018,
      "grad_norm": 8.28174870548537e-06,
      "learning_rate": 4.694633611730988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3615760,
      "step": 6360
    },
    {
      "epoch": 111.67256637168141,
      "grad_norm": 6.964392923691776e-06,
      "learning_rate": 4.694163254725639e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3618656,
      "step": 6365
    },
    {
      "epoch": 111.76106194690266,
      "grad_norm": 1.5723180695204064e-05,
      "learning_rate": 4.693692559352743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3621264,
      "step": 6370
    },
    {
      "epoch": 111.84955752212389,
      "grad_norm": 1.451710249966709e-05,
      "learning_rate": 4.693221525684886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3624000,
      "step": 6375
    },
    {
      "epoch": 111.93805309734513,
      "grad_norm": 3.132229494440253e-06,
      "learning_rate": 4.6927501537947084e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3627072,
      "step": 6380
    },
    {
      "epoch": 112.01769911504425,
      "grad_norm": 6.129375833552331e-06,
      "learning_rate": 4.692278443754901e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3629896,
      "step": 6385
    },
    {
      "epoch": 112.10619469026548,
      "grad_norm": 4.462412107386626e-06,
      "learning_rate": 4.691806395638208e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3632520,
      "step": 6390
    },
    {
      "epoch": 112.19469026548673,
      "grad_norm": 1.4904771887813695e-05,
      "learning_rate": 4.6913340095174255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3635432,
      "step": 6395
    },
    {
      "epoch": 112.28318584070796,
      "grad_norm": 9.727341421239544e-06,
      "learning_rate": 4.690861285465399e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3638584,
      "step": 6400
    },
    {
      "epoch": 112.28318584070796,
      "eval_loss": 0.40295475721359253,
      "eval_runtime": 1.0587,
      "eval_samples_per_second": 23.615,
      "eval_steps_per_second": 12.28,
      "num_input_tokens_seen": 3638584,
      "step": 6400
    },
    {
      "epoch": 112.3716814159292,
      "grad_norm": 4.021453150926391e-06,
      "learning_rate": 4.690388223555031e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3641288,
      "step": 6405
    },
    {
      "epoch": 112.46017699115045,
      "grad_norm": 8.21608591650147e-06,
      "learning_rate": 4.689914823859273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3643800,
      "step": 6410
    },
    {
      "epoch": 112.54867256637168,
      "grad_norm": 1.8902985175373033e-05,
      "learning_rate": 4.689441086451129e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3646792,
      "step": 6415
    },
    {
      "epoch": 112.63716814159292,
      "grad_norm": 1.7366730389767326e-05,
      "learning_rate": 4.688967011403655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3649960,
      "step": 6420
    },
    {
      "epoch": 112.72566371681415,
      "grad_norm": 3.912212378054392e-06,
      "learning_rate": 4.68849259878996e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3653080,
      "step": 6425
    },
    {
      "epoch": 112.8141592920354,
      "grad_norm": 1.587484439369291e-05,
      "learning_rate": 4.6880178486832036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3655432,
      "step": 6430
    },
    {
      "epoch": 112.90265486725664,
      "grad_norm": 8.397369128942955e-06,
      "learning_rate": 4.687542761156598e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3658296,
      "step": 6435
    },
    {
      "epoch": 112.99115044247787,
      "grad_norm": 7.964235919644125e-06,
      "learning_rate": 4.6870673362834096e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3661048,
      "step": 6440
    },
    {
      "epoch": 113.070796460177,
      "grad_norm": 8.262255869340152e-05,
      "learning_rate": 4.6865915741369526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3663752,
      "step": 6445
    },
    {
      "epoch": 113.15929203539822,
      "grad_norm": 6.4177238527918234e-06,
      "learning_rate": 4.686115474790597e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3666504,
      "step": 6450
    },
    {
      "epoch": 113.24778761061947,
      "grad_norm": 7.441127763740951e-06,
      "learning_rate": 4.685639038317762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3669464,
      "step": 6455
    },
    {
      "epoch": 113.33628318584071,
      "grad_norm": 4.014477781311143e-06,
      "learning_rate": 4.685162264791921e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3672584,
      "step": 6460
    },
    {
      "epoch": 113.42477876106194,
      "grad_norm": 3.4820229757315246e-06,
      "learning_rate": 4.684685154286599e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3675400,
      "step": 6465
    },
    {
      "epoch": 113.51327433628319,
      "grad_norm": 9.664789104135707e-06,
      "learning_rate": 4.684207706875371e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3678488,
      "step": 6470
    },
    {
      "epoch": 113.60176991150442,
      "grad_norm": 2.1673235096386634e-05,
      "learning_rate": 4.683729922631866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3681240,
      "step": 6475
    },
    {
      "epoch": 113.69026548672566,
      "grad_norm": 1.691067882347852e-05,
      "learning_rate": 4.683251801629765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3684456,
      "step": 6480
    },
    {
      "epoch": 113.77876106194691,
      "grad_norm": 9.624272934161127e-06,
      "learning_rate": 4.6827733439428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3687240,
      "step": 6485
    },
    {
      "epoch": 113.86725663716814,
      "grad_norm": 2.511373395464034e-06,
      "learning_rate": 4.682294549644754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3689736,
      "step": 6490
    },
    {
      "epoch": 113.95575221238938,
      "grad_norm": 6.187151029735105e-06,
      "learning_rate": 4.681815418809464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3692584,
      "step": 6495
    },
    {
      "epoch": 114.03539823008849,
      "grad_norm": 1.0452435162733309e-05,
      "learning_rate": 4.681335951510819e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3694592,
      "step": 6500
    },
    {
      "epoch": 114.12389380530973,
      "grad_norm": 6.924224635440623e-06,
      "learning_rate": 4.6808561478227576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3697552,
      "step": 6505
    },
    {
      "epoch": 114.21238938053098,
      "grad_norm": 2.59162356996967e-06,
      "learning_rate": 4.680376007819271e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3700240,
      "step": 6510
    },
    {
      "epoch": 114.30088495575221,
      "grad_norm": 1.2993844393349718e-05,
      "learning_rate": 4.679895531574405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3703280,
      "step": 6515
    },
    {
      "epoch": 114.38938053097345,
      "grad_norm": 2.9984205411892617e-06,
      "learning_rate": 4.679414719162253e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3706368,
      "step": 6520
    },
    {
      "epoch": 114.47787610619469,
      "grad_norm": 6.295568255154649e-06,
      "learning_rate": 4.6789335706569635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3708960,
      "step": 6525
    },
    {
      "epoch": 114.56637168141593,
      "grad_norm": 3.750574614969082e-05,
      "learning_rate": 4.678452086132734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3712064,
      "step": 6530
    },
    {
      "epoch": 114.65486725663717,
      "grad_norm": 1.759421320457477e-05,
      "learning_rate": 4.677970265663818e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3715072,
      "step": 6535
    },
    {
      "epoch": 114.7433628318584,
      "grad_norm": 5.141166184330359e-06,
      "learning_rate": 4.677488109324517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3718304,
      "step": 6540
    },
    {
      "epoch": 114.83185840707965,
      "grad_norm": 9.309684173786081e-06,
      "learning_rate": 4.6770056171891846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3720928,
      "step": 6545
    },
    {
      "epoch": 114.92035398230088,
      "grad_norm": 5.182433142181253e-06,
      "learning_rate": 4.6765227893322286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3723792,
      "step": 6550
    },
    {
      "epoch": 115.0,
      "grad_norm": 3.6454111977946013e-06,
      "learning_rate": 4.676039625828107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3726240,
      "step": 6555
    },
    {
      "epoch": 115.08849557522124,
      "grad_norm": 3.3866749618027825e-06,
      "learning_rate": 4.675556126751328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3729184,
      "step": 6560
    },
    {
      "epoch": 115.17699115044248,
      "grad_norm": 1.5773301129229367e-05,
      "learning_rate": 4.6750722921764556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3732400,
      "step": 6565
    },
    {
      "epoch": 115.26548672566372,
      "grad_norm": 8.858786713972222e-06,
      "learning_rate": 4.674588122178102e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3735344,
      "step": 6570
    },
    {
      "epoch": 115.35398230088495,
      "grad_norm": 3.0023777526366757e-06,
      "learning_rate": 4.674103616830931e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3738208,
      "step": 6575
    },
    {
      "epoch": 115.4424778761062,
      "grad_norm": 3.7518127555813408e-06,
      "learning_rate": 4.673618776209663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3741184,
      "step": 6580
    },
    {
      "epoch": 115.53097345132744,
      "grad_norm": 5.530533144337824e-06,
      "learning_rate": 4.673133600389063e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3743648,
      "step": 6585
    },
    {
      "epoch": 115.61946902654867,
      "grad_norm": 7.612314220750704e-06,
      "learning_rate": 4.672648089443953e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3746544,
      "step": 6590
    },
    {
      "epoch": 115.70796460176992,
      "grad_norm": 3.7542029076576e-06,
      "learning_rate": 4.672162243449204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3749008,
      "step": 6595
    },
    {
      "epoch": 115.79646017699115,
      "grad_norm": 2.150010368495714e-05,
      "learning_rate": 4.67167606247974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3752608,
      "step": 6600
    },
    {
      "epoch": 115.79646017699115,
      "eval_loss": 0.40091925859451294,
      "eval_runtime": 1.0583,
      "eval_samples_per_second": 23.622,
      "eval_steps_per_second": 12.284,
      "num_input_tokens_seen": 3752608,
      "step": 6600
    },
    {
      "epoch": 115.88495575221239,
      "grad_norm": 1.280471133213723e-05,
      "learning_rate": 4.671189546610536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3755152,
      "step": 6605
    },
    {
      "epoch": 115.97345132743362,
      "grad_norm": 5.520615559362341e-06,
      "learning_rate": 4.67070269591662e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3758048,
      "step": 6610
    },
    {
      "epoch": 116.05309734513274,
      "grad_norm": 4.29612300649751e-05,
      "learning_rate": 4.670215510473068e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3760768,
      "step": 6615
    },
    {
      "epoch": 116.14159292035399,
      "grad_norm": 7.745978109596763e-06,
      "learning_rate": 4.669727990355013e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3763296,
      "step": 6620
    },
    {
      "epoch": 116.23008849557522,
      "grad_norm": 4.751200322061777e-06,
      "learning_rate": 4.669240135637635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3766656,
      "step": 6625
    },
    {
      "epoch": 116.31858407079646,
      "grad_norm": 3.6066062421014067e-06,
      "learning_rate": 4.6687519463961675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3769568,
      "step": 6630
    },
    {
      "epoch": 116.40707964601769,
      "grad_norm": 1.7270682292291895e-05,
      "learning_rate": 4.668263422705896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3772480,
      "step": 6635
    },
    {
      "epoch": 116.49557522123894,
      "grad_norm": 2.313971344847232e-05,
      "learning_rate": 4.667774564642156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3775360,
      "step": 6640
    },
    {
      "epoch": 116.58407079646018,
      "grad_norm": 1.6368077922379598e-05,
      "learning_rate": 4.6672853722803365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3778352,
      "step": 6645
    },
    {
      "epoch": 116.67256637168141,
      "grad_norm": 4.149494543526089e-06,
      "learning_rate": 4.666795845695877e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3781216,
      "step": 6650
    },
    {
      "epoch": 116.76106194690266,
      "grad_norm": 4.98972167406464e-06,
      "learning_rate": 4.666305984964269e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3783664,
      "step": 6655
    },
    {
      "epoch": 116.84955752212389,
      "grad_norm": 9.964713171939366e-06,
      "learning_rate": 4.6658157901610535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3786576,
      "step": 6660
    },
    {
      "epoch": 116.93805309734513,
      "grad_norm": 4.593842459144071e-05,
      "learning_rate": 4.665325261361826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3789504,
      "step": 6665
    },
    {
      "epoch": 117.01769911504425,
      "grad_norm": 6.5336730585841e-06,
      "learning_rate": 4.664834398642232e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3791752,
      "step": 6670
    },
    {
      "epoch": 117.10619469026548,
      "grad_norm": 1.3488926015270408e-05,
      "learning_rate": 4.6643432020779686e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3794520,
      "step": 6675
    },
    {
      "epoch": 117.19469026548673,
      "grad_norm": 2.9191814974183217e-06,
      "learning_rate": 4.663851671744786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3797096,
      "step": 6680
    },
    {
      "epoch": 117.28318584070796,
      "grad_norm": 9.702255738375243e-06,
      "learning_rate": 4.6633598077184815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3799576,
      "step": 6685
    },
    {
      "epoch": 117.3716814159292,
      "grad_norm": 4.466790869628312e-06,
      "learning_rate": 4.662867610074908e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3802632,
      "step": 6690
    },
    {
      "epoch": 117.46017699115045,
      "grad_norm": 4.811392955161864e-06,
      "learning_rate": 4.6623750788899696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3805864,
      "step": 6695
    },
    {
      "epoch": 117.54867256637168,
      "grad_norm": 3.188740947734914e-06,
      "learning_rate": 4.6618822142396195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3809096,
      "step": 6700
    },
    {
      "epoch": 117.63716814159292,
      "grad_norm": 8.962545507529285e-06,
      "learning_rate": 4.661389016199864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3811912,
      "step": 6705
    },
    {
      "epoch": 117.72566371681415,
      "grad_norm": 5.325776783138281e-06,
      "learning_rate": 4.660895484846761e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3814424,
      "step": 6710
    },
    {
      "epoch": 117.8141592920354,
      "grad_norm": 5.21161564392969e-06,
      "learning_rate": 4.660401620256418e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3817496,
      "step": 6715
    },
    {
      "epoch": 117.90265486725664,
      "grad_norm": 5.3328403737396e-06,
      "learning_rate": 4.659907422504997e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3820184,
      "step": 6720
    },
    {
      "epoch": 117.99115044247787,
      "grad_norm": 1.7550493794260547e-05,
      "learning_rate": 4.6594128916687074e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3823400,
      "step": 6725
    },
    {
      "epoch": 118.070796460177,
      "grad_norm": 1.5404280929942615e-05,
      "learning_rate": 4.658918027823813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3825952,
      "step": 6730
    },
    {
      "epoch": 118.15929203539822,
      "grad_norm": 4.185929810773814e-06,
      "learning_rate": 4.658422831046628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3828912,
      "step": 6735
    },
    {
      "epoch": 118.24778761061947,
      "grad_norm": 4.168191935605137e-06,
      "learning_rate": 4.657927301413518e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3831824,
      "step": 6740
    },
    {
      "epoch": 118.33628318584071,
      "grad_norm": 1.1735222869901918e-05,
      "learning_rate": 4.657431439000901e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3834688,
      "step": 6745
    },
    {
      "epoch": 118.42477876106194,
      "grad_norm": 7.141483365558088e-06,
      "learning_rate": 4.656935243885243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3837680,
      "step": 6750
    },
    {
      "epoch": 118.51327433628319,
      "grad_norm": 2.669153900569654e-06,
      "learning_rate": 4.656438716143066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3840176,
      "step": 6755
    },
    {
      "epoch": 118.60176991150442,
      "grad_norm": 6.980435500736348e-06,
      "learning_rate": 4.6559418558509384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3843536,
      "step": 6760
    },
    {
      "epoch": 118.69026548672566,
      "grad_norm": 4.064914719492663e-06,
      "learning_rate": 4.6554446630854833e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3846512,
      "step": 6765
    },
    {
      "epoch": 118.77876106194691,
      "grad_norm": 5.23139124197769e-06,
      "learning_rate": 4.654947137923374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3849184,
      "step": 6770
    },
    {
      "epoch": 118.86725663716814,
      "grad_norm": 4.207561687508132e-06,
      "learning_rate": 4.654449280441335e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3852064,
      "step": 6775
    },
    {
      "epoch": 118.95575221238938,
      "grad_norm": 3.6477724734140793e-06,
      "learning_rate": 4.653951090716143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3854688,
      "step": 6780
    },
    {
      "epoch": 119.03539823008849,
      "grad_norm": 7.2772531893861014e-06,
      "learning_rate": 4.653452568824625e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3856896,
      "step": 6785
    },
    {
      "epoch": 119.12389380530973,
      "grad_norm": 1.6649753888486885e-05,
      "learning_rate": 4.6529537148436585e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3859152,
      "step": 6790
    },
    {
      "epoch": 119.21238938053098,
      "grad_norm": 3.6777280911337584e-05,
      "learning_rate": 4.6524545288501734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3862256,
      "step": 6795
    },
    {
      "epoch": 119.30088495575221,
      "grad_norm": 4.410515884956112e-06,
      "learning_rate": 4.6519550109211506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3865376,
      "step": 6800
    },
    {
      "epoch": 119.30088495575221,
      "eval_loss": 0.4005506932735443,
      "eval_runtime": 1.0617,
      "eval_samples_per_second": 23.548,
      "eval_steps_per_second": 12.245,
      "num_input_tokens_seen": 3865376,
      "step": 6800
    },
    {
      "epoch": 119.38938053097345,
      "grad_norm": 3.898836894222768e-06,
      "learning_rate": 4.651455161133622e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3868048,
      "step": 6805
    },
    {
      "epoch": 119.47787610619469,
      "grad_norm": 5.081612016510917e-06,
      "learning_rate": 4.6509549795646704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3870816,
      "step": 6810
    },
    {
      "epoch": 119.56637168141593,
      "grad_norm": 5.1395320042502135e-05,
      "learning_rate": 4.6504544662914306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3874048,
      "step": 6815
    },
    {
      "epoch": 119.65486725663717,
      "grad_norm": 4.2374111217213795e-05,
      "learning_rate": 4.6499536213910876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3876832,
      "step": 6820
    },
    {
      "epoch": 119.7433628318584,
      "grad_norm": 5.1987290135002695e-06,
      "learning_rate": 4.6494524449408786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3879296,
      "step": 6825
    },
    {
      "epoch": 119.83185840707965,
      "grad_norm": 7.953462954901624e-06,
      "learning_rate": 4.6489509370180903e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3882592,
      "step": 6830
    },
    {
      "epoch": 119.92035398230088,
      "grad_norm": 3.1104286790650804e-06,
      "learning_rate": 4.648449097700063e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3885584,
      "step": 6835
    },
    {
      "epoch": 120.0,
      "grad_norm": 1.0075274076370988e-05,
      "learning_rate": 4.647946927064185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3888056,
      "step": 6840
    },
    {
      "epoch": 120.08849557522124,
      "grad_norm": 3.249194378440734e-06,
      "learning_rate": 4.647444425187898e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3890744,
      "step": 6845
    },
    {
      "epoch": 120.17699115044248,
      "grad_norm": 5.14702514919918e-05,
      "learning_rate": 4.646941592148695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3893640,
      "step": 6850
    },
    {
      "epoch": 120.26548672566372,
      "grad_norm": 3.5252817269793013e-06,
      "learning_rate": 4.646438428024117e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3896264,
      "step": 6855
    },
    {
      "epoch": 120.35398230088495,
      "grad_norm": 6.329035841190489e-06,
      "learning_rate": 4.64593493289176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3899192,
      "step": 6860
    },
    {
      "epoch": 120.4424778761062,
      "grad_norm": 1.684550807112828e-05,
      "learning_rate": 4.64543110682927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3901880,
      "step": 6865
    },
    {
      "epoch": 120.53097345132744,
      "grad_norm": 4.693876235251082e-06,
      "learning_rate": 4.644926949914341e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3904824,
      "step": 6870
    },
    {
      "epoch": 120.61946902654867,
      "grad_norm": 9.03130967344623e-06,
      "learning_rate": 4.644422462224722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3907720,
      "step": 6875
    },
    {
      "epoch": 120.70796460176992,
      "grad_norm": 5.139859422342852e-05,
      "learning_rate": 4.643917643838211e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3911064,
      "step": 6880
    },
    {
      "epoch": 120.79646017699115,
      "grad_norm": 1.5951194654917344e-05,
      "learning_rate": 4.6434124948326564e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3913784,
      "step": 6885
    },
    {
      "epoch": 120.88495575221239,
      "grad_norm": 2.67141149379313e-05,
      "learning_rate": 4.6429070152859594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3916504,
      "step": 6890
    },
    {
      "epoch": 120.97345132743362,
      "grad_norm": 7.723090675426647e-06,
      "learning_rate": 4.6424012052760714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3919400,
      "step": 6895
    },
    {
      "epoch": 121.05309734513274,
      "grad_norm": 1.9115221220999956e-05,
      "learning_rate": 4.6418950648809945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3921912,
      "step": 6900
    },
    {
      "epoch": 121.14159292035399,
      "grad_norm": 4.936076948069967e-06,
      "learning_rate": 4.641388594178782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3925128,
      "step": 6905
    },
    {
      "epoch": 121.23008849557522,
      "grad_norm": 7.638302122359164e-06,
      "learning_rate": 4.640881793247538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3928264,
      "step": 6910
    },
    {
      "epoch": 121.31858407079646,
      "grad_norm": 7.976500455697533e-06,
      "learning_rate": 4.6403746621654173e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3930824,
      "step": 6915
    },
    {
      "epoch": 121.40707964601769,
      "grad_norm": 4.879271273239283e-06,
      "learning_rate": 4.639867201010626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3933864,
      "step": 6920
    },
    {
      "epoch": 121.49557522123894,
      "grad_norm": 2.5621668555686483e-06,
      "learning_rate": 4.6393594098614204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3936504,
      "step": 6925
    },
    {
      "epoch": 121.58407079646018,
      "grad_norm": 2.0964641862519784e-06,
      "learning_rate": 4.63885128879611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3939752,
      "step": 6930
    },
    {
      "epoch": 121.67256637168141,
      "grad_norm": 2.238320575997932e-06,
      "learning_rate": 4.638342837893052e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3942872,
      "step": 6935
    },
    {
      "epoch": 121.76106194690266,
      "grad_norm": 4.7145334974629804e-05,
      "learning_rate": 4.6378340572306565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3945368,
      "step": 6940
    },
    {
      "epoch": 121.84955752212389,
      "grad_norm": 3.96911264033406e-06,
      "learning_rate": 4.6373249468873833e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3948120,
      "step": 6945
    },
    {
      "epoch": 121.93805309734513,
      "grad_norm": 3.3564228942850605e-06,
      "learning_rate": 4.636815506941744e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3951208,
      "step": 6950
    },
    {
      "epoch": 122.01769911504425,
      "grad_norm": 2.4155112896551145e-06,
      "learning_rate": 4.6363057374723004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3953560,
      "step": 6955
    },
    {
      "epoch": 122.10619469026548,
      "grad_norm": 3.483888349364861e-06,
      "learning_rate": 4.635795638557666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3955960,
      "step": 6960
    },
    {
      "epoch": 122.19469026548673,
      "grad_norm": 4.471212832868332e-06,
      "learning_rate": 4.635285210276504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3958984,
      "step": 6965
    },
    {
      "epoch": 122.28318584070796,
      "grad_norm": 6.9085831455595326e-06,
      "learning_rate": 4.6347744527075295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3962200,
      "step": 6970
    },
    {
      "epoch": 122.3716814159292,
      "grad_norm": 9.181264431390446e-06,
      "learning_rate": 4.634263365929506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3964696,
      "step": 6975
    },
    {
      "epoch": 122.46017699115045,
      "grad_norm": 9.516963473288342e-06,
      "learning_rate": 4.6337519500212515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3967752,
      "step": 6980
    },
    {
      "epoch": 122.54867256637168,
      "grad_norm": 3.103186600128538e-06,
      "learning_rate": 4.633240205061632e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3970712,
      "step": 6985
    },
    {
      "epoch": 122.63716814159292,
      "grad_norm": 3.210237264283933e-05,
      "learning_rate": 4.632728131129565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3974008,
      "step": 6990
    },
    {
      "epoch": 122.72566371681415,
      "grad_norm": 5.301857163431123e-05,
      "learning_rate": 4.632215728304018e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3976472,
      "step": 6995
    },
    {
      "epoch": 122.8141592920354,
      "grad_norm": 5.016665454604663e-06,
      "learning_rate": 4.63170299666401e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3979464,
      "step": 7000
    },
    {
      "epoch": 122.8141592920354,
      "eval_loss": 0.40326330065727234,
      "eval_runtime": 1.0597,
      "eval_samples_per_second": 23.592,
      "eval_steps_per_second": 12.268,
      "num_input_tokens_seen": 3979464,
      "step": 7000
    },
    {
      "epoch": 122.90265486725664,
      "grad_norm": 2.4759707230259664e-05,
      "learning_rate": 4.631189936288612e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3982408,
      "step": 7005
    },
    {
      "epoch": 122.99115044247787,
      "grad_norm": 2.309582669113297e-06,
      "learning_rate": 4.630676547256944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3985096,
      "step": 7010
    },
    {
      "epoch": 123.070796460177,
      "grad_norm": 2.0650554688472766e-06,
      "learning_rate": 4.630162829648176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3988384,
      "step": 7015
    },
    {
      "epoch": 123.15929203539822,
      "grad_norm": 8.102343599603046e-06,
      "learning_rate": 4.629648783541531e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3991360,
      "step": 7020
    },
    {
      "epoch": 123.24778761061947,
      "grad_norm": 8.155776413332205e-06,
      "learning_rate": 4.6291344090162804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3994176,
      "step": 7025
    },
    {
      "epoch": 123.33628318584071,
      "grad_norm": 5.504792625288246e-06,
      "learning_rate": 4.628619706151748e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3996816,
      "step": 7030
    },
    {
      "epoch": 123.42477876106194,
      "grad_norm": 8.871857971826103e-06,
      "learning_rate": 4.628104675027306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 3999552,
      "step": 7035
    },
    {
      "epoch": 123.51327433628319,
      "grad_norm": 4.387824901641579e-06,
      "learning_rate": 4.6275893157223805e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4002288,
      "step": 7040
    },
    {
      "epoch": 123.60176991150442,
      "grad_norm": 3.819180165010039e-06,
      "learning_rate": 4.627073628316445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4005264,
      "step": 7045
    },
    {
      "epoch": 123.69026548672566,
      "grad_norm": 4.3724572606151924e-05,
      "learning_rate": 4.626557612889026e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4008352,
      "step": 7050
    },
    {
      "epoch": 123.77876106194691,
      "grad_norm": 3.444352159931441e-06,
      "learning_rate": 4.626041269519699e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4010816,
      "step": 7055
    },
    {
      "epoch": 123.86725663716814,
      "grad_norm": 1.1585078027565032e-05,
      "learning_rate": 4.6255245982880905e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4013584,
      "step": 7060
    },
    {
      "epoch": 123.95575221238938,
      "grad_norm": 2.5396000182809075e-06,
      "learning_rate": 4.625007599273879e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4016400,
      "step": 7065
    },
    {
      "epoch": 124.03539823008849,
      "grad_norm": 4.59042030342971e-06,
      "learning_rate": 4.6244902725567895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4019080,
      "step": 7070
    },
    {
      "epoch": 124.12389380530973,
      "grad_norm": 1.1443159564805683e-05,
      "learning_rate": 4.6239726182166024e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4022488,
      "step": 7075
    },
    {
      "epoch": 124.21238938053098,
      "grad_norm": 3.520895006658975e-06,
      "learning_rate": 4.623454636333147e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4025128,
      "step": 7080
    },
    {
      "epoch": 124.30088495575221,
      "grad_norm": 4.341607564128935e-06,
      "learning_rate": 4.622936326986301e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4027976,
      "step": 7085
    },
    {
      "epoch": 124.38938053097345,
      "grad_norm": 5.4697575251339e-06,
      "learning_rate": 4.6224176902559946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4031496,
      "step": 7090
    },
    {
      "epoch": 124.47787610619469,
      "grad_norm": 2.5713139621075243e-05,
      "learning_rate": 4.621898726222209e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4034536,
      "step": 7095
    },
    {
      "epoch": 124.56637168141593,
      "grad_norm": 4.776743935508421e-06,
      "learning_rate": 4.6213794349649744e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4037096,
      "step": 7100
    },
    {
      "epoch": 124.65486725663717,
      "grad_norm": 2.952496834041085e-05,
      "learning_rate": 4.6208598165643715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4039448,
      "step": 7105
    },
    {
      "epoch": 124.7433628318584,
      "grad_norm": 3.5608516100182896e-06,
      "learning_rate": 4.620339871100533e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4042760,
      "step": 7110
    },
    {
      "epoch": 124.83185840707965,
      "grad_norm": 4.578625976137118e-06,
      "learning_rate": 4.6198195986536394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4045176,
      "step": 7115
    },
    {
      "epoch": 124.92035398230088,
      "grad_norm": 4.168620307609672e-06,
      "learning_rate": 4.619298999303926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4047864,
      "step": 7120
    },
    {
      "epoch": 125.0,
      "grad_norm": 4.2447113628441e-06,
      "learning_rate": 4.618778073131673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4050248,
      "step": 7125
    },
    {
      "epoch": 125.08849557522124,
      "grad_norm": 5.743712608818896e-05,
      "learning_rate": 4.618256820217215e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4053352,
      "step": 7130
    },
    {
      "epoch": 125.17699115044248,
      "grad_norm": 0.00010205978469457477,
      "learning_rate": 4.617735240640936e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4056088,
      "step": 7135
    },
    {
      "epoch": 125.26548672566372,
      "grad_norm": 3.463963366812095e-06,
      "learning_rate": 4.6172133344832705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4058968,
      "step": 7140
    },
    {
      "epoch": 125.35398230088495,
      "grad_norm": 5.266894731903449e-05,
      "learning_rate": 4.6166911018247004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4061976,
      "step": 7145
    },
    {
      "epoch": 125.4424778761062,
      "grad_norm": 1.9744145447475603e-06,
      "learning_rate": 4.616168542745764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4064696,
      "step": 7150
    },
    {
      "epoch": 125.53097345132744,
      "grad_norm": 2.132478402927518e-05,
      "learning_rate": 4.6156456573270446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4067592,
      "step": 7155
    },
    {
      "epoch": 125.61946902654867,
      "grad_norm": 1.575703572598286e-05,
      "learning_rate": 4.615122445649177e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4070696,
      "step": 7160
    },
    {
      "epoch": 125.70796460176992,
      "grad_norm": 3.940596343454672e-06,
      "learning_rate": 4.6145989077928486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4073112,
      "step": 7165
    },
    {
      "epoch": 125.79646017699115,
      "grad_norm": 4.0757736314844806e-06,
      "learning_rate": 4.6140750438387953e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4076072,
      "step": 7170
    },
    {
      "epoch": 125.88495575221239,
      "grad_norm": 5.451752258522902e-06,
      "learning_rate": 4.613550853867803e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4078920,
      "step": 7175
    },
    {
      "epoch": 125.97345132743362,
      "grad_norm": 1.398105359839974e-05,
      "learning_rate": 4.613026337960708e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4082120,
      "step": 7180
    },
    {
      "epoch": 126.05309734513274,
      "grad_norm": 5.846713975188322e-06,
      "learning_rate": 4.612501496198398e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4084544,
      "step": 7185
    },
    {
      "epoch": 126.14159292035399,
      "grad_norm": 4.959859506925568e-05,
      "learning_rate": 4.61197632866181e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4087728,
      "step": 7190
    },
    {
      "epoch": 126.23008849557522,
      "grad_norm": 3.0972248623584164e-06,
      "learning_rate": 4.611450835431931e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4090704,
      "step": 7195
    },
    {
      "epoch": 126.31858407079646,
      "grad_norm": 2.100631490975502e-06,
      "learning_rate": 4.6109250165898e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4093296,
      "step": 7200
    },
    {
      "epoch": 126.31858407079646,
      "eval_loss": 0.40944311022758484,
      "eval_runtime": 1.0638,
      "eval_samples_per_second": 23.5,
      "eval_steps_per_second": 12.22,
      "num_input_tokens_seen": 4093296,
      "step": 7200
    },
    {
      "epoch": 126.40707964601769,
      "grad_norm": 5.605777914752252e-06,
      "learning_rate": 4.610398872216503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4096144,
      "step": 7205
    },
    {
      "epoch": 126.49557522123894,
      "grad_norm": 6.1811692830815446e-06,
      "learning_rate": 4.6098724023931796e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4098608,
      "step": 7210
    },
    {
      "epoch": 126.58407079646018,
      "grad_norm": 4.008789801446255e-06,
      "learning_rate": 4.609345607201017e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4101792,
      "step": 7215
    },
    {
      "epoch": 126.67256637168141,
      "grad_norm": 1.784139158189646e-06,
      "learning_rate": 4.608818486721254e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4105184,
      "step": 7220
    },
    {
      "epoch": 126.76106194690266,
      "grad_norm": 5.0515932343842e-06,
      "learning_rate": 4.608291041035179e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4107952,
      "step": 7225
    },
    {
      "epoch": 126.84955752212389,
      "grad_norm": 2.1250998543109745e-05,
      "learning_rate": 4.607763270224132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4110480,
      "step": 7230
    },
    {
      "epoch": 126.93805309734513,
      "grad_norm": 2.099377752529108e-06,
      "learning_rate": 4.6072351743695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4113280,
      "step": 7235
    },
    {
      "epoch": 127.01769911504425,
      "grad_norm": 2.365942236792762e-06,
      "learning_rate": 4.606706753552723e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4115912,
      "step": 7240
    },
    {
      "epoch": 127.10619469026548,
      "grad_norm": 1.1753283615689725e-05,
      "learning_rate": 4.6061780078552906e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4118840,
      "step": 7245
    },
    {
      "epoch": 127.19469026548673,
      "grad_norm": 7.376882422249764e-06,
      "learning_rate": 4.605648937358742e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4121624,
      "step": 7250
    },
    {
      "epoch": 127.28318584070796,
      "grad_norm": 7.945823199406732e-06,
      "learning_rate": 4.605119542144665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4124152,
      "step": 7255
    },
    {
      "epoch": 127.3716814159292,
      "grad_norm": 2.7083347049483564e-06,
      "learning_rate": 4.604589822294701e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4127016,
      "step": 7260
    },
    {
      "epoch": 127.46017699115045,
      "grad_norm": 3.434039626881713e-06,
      "learning_rate": 4.604059777890537e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4129592,
      "step": 7265
    },
    {
      "epoch": 127.54867256637168,
      "grad_norm": 6.0060447140131146e-06,
      "learning_rate": 4.6035294090139145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4132920,
      "step": 7270
    },
    {
      "epoch": 127.63716814159292,
      "grad_norm": 1.1321657439111732e-05,
      "learning_rate": 4.6029987157466226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4136024,
      "step": 7275
    },
    {
      "epoch": 127.72566371681415,
      "grad_norm": 2.4104074327624403e-05,
      "learning_rate": 4.602467698170502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4138840,
      "step": 7280
    },
    {
      "epoch": 127.8141592920354,
      "grad_norm": 2.4852610295056365e-05,
      "learning_rate": 4.601936356367439e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4142152,
      "step": 7285
    },
    {
      "epoch": 127.90265486725664,
      "grad_norm": 6.276774911384564e-06,
      "learning_rate": 4.601404690419377e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4144856,
      "step": 7290
    },
    {
      "epoch": 127.99115044247787,
      "grad_norm": 2.2514284410135588e-06,
      "learning_rate": 4.600872700408303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4148088,
      "step": 7295
    },
    {
      "epoch": 128.07079646017698,
      "grad_norm": 3.130777622573078e-05,
      "learning_rate": 4.600340386416258e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4150680,
      "step": 7300
    },
    {
      "epoch": 128.15929203539824,
      "grad_norm": 6.087739166105166e-06,
      "learning_rate": 4.5998077485253296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4153544,
      "step": 7305
    },
    {
      "epoch": 128.24778761061947,
      "grad_norm": 3.166629539919086e-06,
      "learning_rate": 4.59927478681766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4156136,
      "step": 7310
    },
    {
      "epoch": 128.3362831858407,
      "grad_norm": 3.5289235711388756e-06,
      "learning_rate": 4.5987415013754366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4158792,
      "step": 7315
    },
    {
      "epoch": 128.42477876106196,
      "grad_norm": 2.1271594050631393e-06,
      "learning_rate": 4.598207892280899e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4161672,
      "step": 7320
    },
    {
      "epoch": 128.5132743362832,
      "grad_norm": 2.0499010133789852e-06,
      "learning_rate": 4.597673959616337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4164632,
      "step": 7325
    },
    {
      "epoch": 128.60176991150442,
      "grad_norm": 2.9892347811255604e-05,
      "learning_rate": 4.597139703464089e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4167768,
      "step": 7330
    },
    {
      "epoch": 128.69026548672565,
      "grad_norm": 3.6066703614778817e-06,
      "learning_rate": 4.596605123906545e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4170616,
      "step": 7335
    },
    {
      "epoch": 128.7787610619469,
      "grad_norm": 1.2054641956638079e-05,
      "learning_rate": 4.596070221026143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4173320,
      "step": 7340
    },
    {
      "epoch": 128.86725663716814,
      "grad_norm": 5.714271992474096e-06,
      "learning_rate": 4.595534994905372e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4176216,
      "step": 7345
    },
    {
      "epoch": 128.95575221238937,
      "grad_norm": 2.063269857899286e-05,
      "learning_rate": 4.594999445626771e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4178760,
      "step": 7350
    },
    {
      "epoch": 129.0353982300885,
      "grad_norm": 1.7735463188728318e-05,
      "learning_rate": 4.5944635732729276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4181184,
      "step": 7355
    },
    {
      "epoch": 129.12389380530973,
      "grad_norm": 9.957710972230416e-06,
      "learning_rate": 4.5939273779264804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4184208,
      "step": 7360
    },
    {
      "epoch": 129.21238938053096,
      "grad_norm": 2.259666189274867e-06,
      "learning_rate": 4.593390859670118e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4187680,
      "step": 7365
    },
    {
      "epoch": 129.30088495575222,
      "grad_norm": 1.0842664778465405e-05,
      "learning_rate": 4.5928540185865776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4190400,
      "step": 7370
    },
    {
      "epoch": 129.38938053097345,
      "grad_norm": 2.212994786532363e-06,
      "learning_rate": 4.592316854758648e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4192720,
      "step": 7375
    },
    {
      "epoch": 129.47787610619469,
      "grad_norm": 3.6382268717716215e-06,
      "learning_rate": 4.5917793682691646e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4195584,
      "step": 7380
    },
    {
      "epoch": 129.56637168141592,
      "grad_norm": 2.525178615542245e-06,
      "learning_rate": 4.5912415592010164e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4198048,
      "step": 7385
    },
    {
      "epoch": 129.65486725663717,
      "grad_norm": 3.682812121041934e-06,
      "learning_rate": 4.5907034276371386e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4201184,
      "step": 7390
    },
    {
      "epoch": 129.7433628318584,
      "grad_norm": 6.4791734075697605e-06,
      "learning_rate": 4.5901649736605196e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4203936,
      "step": 7395
    },
    {
      "epoch": 129.83185840707964,
      "grad_norm": 3.2078010008262936e-06,
      "learning_rate": 4.589626197354195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4207120,
      "step": 7400
    },
    {
      "epoch": 129.83185840707964,
      "eval_loss": 0.4079572260379791,
      "eval_runtime": 1.0617,
      "eval_samples_per_second": 23.547,
      "eval_steps_per_second": 12.244,
      "num_input_tokens_seen": 4207120,
      "step": 7400
    },
    {
      "epoch": 129.9203539823009,
      "grad_norm": 3.471783884378965e-06,
      "learning_rate": 4.5890870988012504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4210320,
      "step": 7405
    },
    {
      "epoch": 130.0,
      "grad_norm": 3.533292328938842e-06,
      "learning_rate": 4.5885476780848226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4212296,
      "step": 7410
    },
    {
      "epoch": 130.08849557522123,
      "grad_norm": 3.589886091504013e-06,
      "learning_rate": 4.5880079352880964e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4215336,
      "step": 7415
    },
    {
      "epoch": 130.1769911504425,
      "grad_norm": 3.6154501685814466e-06,
      "learning_rate": 4.5874678704943065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4218280,
      "step": 7420
    },
    {
      "epoch": 130.26548672566372,
      "grad_norm": 2.538707121857442e-06,
      "learning_rate": 4.5869274837867394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4221432,
      "step": 7425
    },
    {
      "epoch": 130.35398230088495,
      "grad_norm": 1.3415623470791616e-05,
      "learning_rate": 4.5863867752487275e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4224120,
      "step": 7430
    },
    {
      "epoch": 130.44247787610618,
      "grad_norm": 2.468918410158949e-06,
      "learning_rate": 4.5858457449636554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4226808,
      "step": 7435
    },
    {
      "epoch": 130.53097345132744,
      "grad_norm": 3.85400790037238e-06,
      "learning_rate": 4.5853043930149574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4229880,
      "step": 7440
    },
    {
      "epoch": 130.61946902654867,
      "grad_norm": 8.651150892546866e-06,
      "learning_rate": 4.584762719486117e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4232296,
      "step": 7445
    },
    {
      "epoch": 130.7079646017699,
      "grad_norm": 2.0022234821226448e-05,
      "learning_rate": 4.584220724460665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4235080,
      "step": 7450
    },
    {
      "epoch": 130.79646017699116,
      "grad_norm": 4.459006049728487e-06,
      "learning_rate": 4.5836784080221865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4238360,
      "step": 7455
    },
    {
      "epoch": 130.8849557522124,
      "grad_norm": 7.468875992344692e-06,
      "learning_rate": 4.583135770254312e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4241000,
      "step": 7460
    },
    {
      "epoch": 130.97345132743362,
      "grad_norm": 5.363603122532368e-06,
      "learning_rate": 4.5825928112407236e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4244472,
      "step": 7465
    },
    {
      "epoch": 131.05309734513276,
      "grad_norm": 4.435590653883992e-06,
      "learning_rate": 4.582049531065152e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4246536,
      "step": 7470
    },
    {
      "epoch": 131.141592920354,
      "grad_norm": 2.079581918223994e-06,
      "learning_rate": 4.5815059298113783e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4249544,
      "step": 7475
    },
    {
      "epoch": 131.23008849557522,
      "grad_norm": 1.183915082947351e-05,
      "learning_rate": 4.580962007563232e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4252808,
      "step": 7480
    },
    {
      "epoch": 131.31858407079645,
      "grad_norm": 4.30996369686909e-05,
      "learning_rate": 4.5804177644045935e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4255528,
      "step": 7485
    },
    {
      "epoch": 131.4070796460177,
      "grad_norm": 2.5538922727719182e-06,
      "learning_rate": 4.579873200419391e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4258472,
      "step": 7490
    },
    {
      "epoch": 131.49557522123894,
      "grad_norm": 1.3528287126973737e-05,
      "learning_rate": 4.5793283156916046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4261464,
      "step": 7495
    },
    {
      "epoch": 131.58407079646017,
      "grad_norm": 5.0257476686965674e-05,
      "learning_rate": 4.578783110305261e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4263784,
      "step": 7500
    },
    {
      "epoch": 131.67256637168143,
      "grad_norm": 2.6427771899761865e-06,
      "learning_rate": 4.578237584344438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4267032,
      "step": 7505
    },
    {
      "epoch": 131.76106194690266,
      "grad_norm": 2.537288764870027e-06,
      "learning_rate": 4.577691737893263e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4269560,
      "step": 7510
    },
    {
      "epoch": 131.8495575221239,
      "grad_norm": 3.13170471599733e-06,
      "learning_rate": 4.577145571035912e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4272664,
      "step": 7515
    },
    {
      "epoch": 131.93805309734512,
      "grad_norm": 4.749397248815512e-06,
      "learning_rate": 4.576599083856611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4275608,
      "step": 7520
    },
    {
      "epoch": 132.01769911504425,
      "grad_norm": 2.9606458156195004e-06,
      "learning_rate": 4.576052276439635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4277864,
      "step": 7525
    },
    {
      "epoch": 132.10619469026548,
      "grad_norm": 3.15346323986887e-06,
      "learning_rate": 4.575505148869308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4280696,
      "step": 7530
    },
    {
      "epoch": 132.1946902654867,
      "grad_norm": 6.448569365602452e-06,
      "learning_rate": 4.574957701230006e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4283400,
      "step": 7535
    },
    {
      "epoch": 132.28318584070797,
      "grad_norm": 1.323356536886422e-05,
      "learning_rate": 4.57440993360615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4286648,
      "step": 7540
    },
    {
      "epoch": 132.3716814159292,
      "grad_norm": 4.290339802537346e-06,
      "learning_rate": 4.5738618460822134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4289512,
      "step": 7545
    },
    {
      "epoch": 132.46017699115043,
      "grad_norm": 1.6892036001081578e-06,
      "learning_rate": 4.573313438742719e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4292472,
      "step": 7550
    },
    {
      "epoch": 132.5486725663717,
      "grad_norm": 1.7339270925731398e-05,
      "learning_rate": 4.5727647116722374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4295480,
      "step": 7555
    },
    {
      "epoch": 132.63716814159292,
      "grad_norm": 2.022774424403906e-06,
      "learning_rate": 4.5722156649553884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4298536,
      "step": 7560
    },
    {
      "epoch": 132.72566371681415,
      "grad_norm": 1.35392974698334e-05,
      "learning_rate": 4.571666298676843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4301528,
      "step": 7565
    },
    {
      "epoch": 132.81415929203538,
      "grad_norm": 2.7262590265308972e-06,
      "learning_rate": 4.571116612921321e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4304312,
      "step": 7570
    },
    {
      "epoch": 132.90265486725664,
      "grad_norm": 5.13371287524933e-06,
      "learning_rate": 4.57056660777359e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4306552,
      "step": 7575
    },
    {
      "epoch": 132.99115044247787,
      "grad_norm": 4.6856798689987045e-06,
      "learning_rate": 4.5700162833184666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4309288,
      "step": 7580
    },
    {
      "epoch": 133.07079646017698,
      "grad_norm": 4.724446625914425e-06,
      "learning_rate": 4.5694656396408195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4312024,
      "step": 7585
    },
    {
      "epoch": 133.15929203539824,
      "grad_norm": 1.8602451746119186e-05,
      "learning_rate": 4.5689146768255646e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4314760,
      "step": 7590
    },
    {
      "epoch": 133.24778761061947,
      "grad_norm": 2.278743295391905e-06,
      "learning_rate": 4.568363394957667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4317400,
      "step": 7595
    },
    {
      "epoch": 133.3362831858407,
      "grad_norm": 1.7582129657967016e-05,
      "learning_rate": 4.567811794122141e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4320568,
      "step": 7600
    },
    {
      "epoch": 133.3362831858407,
      "eval_loss": 0.4073774218559265,
      "eval_runtime": 1.0644,
      "eval_samples_per_second": 23.488,
      "eval_steps_per_second": 12.214,
      "num_input_tokens_seen": 4320568,
      "step": 7600
    },
    {
      "epoch": 133.42477876106196,
      "grad_norm": 3.765938527067192e-05,
      "learning_rate": 4.56725987440405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4322968,
      "step": 7605
    },
    {
      "epoch": 133.5132743362832,
      "grad_norm": 2.799526100716321e-06,
      "learning_rate": 4.566707635888508e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4325864,
      "step": 7610
    },
    {
      "epoch": 133.60176991150442,
      "grad_norm": 2.0925767785229255e-06,
      "learning_rate": 4.566155078660677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4328504,
      "step": 7615
    },
    {
      "epoch": 133.69026548672565,
      "grad_norm": 2.9431796519929776e-06,
      "learning_rate": 4.565602202805768e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4331016,
      "step": 7620
    },
    {
      "epoch": 133.7787610619469,
      "grad_norm": 3.8144153222674504e-06,
      "learning_rate": 4.56504900840904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4333976,
      "step": 7625
    },
    {
      "epoch": 133.86725663716814,
      "grad_norm": 4.888194325758377e-06,
      "learning_rate": 4.564495495555805e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4336936,
      "step": 7630
    },
    {
      "epoch": 133.95575221238937,
      "grad_norm": 8.765111488173716e-06,
      "learning_rate": 4.5639416643314204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4340040,
      "step": 7635
    },
    {
      "epoch": 134.0353982300885,
      "grad_norm": 2.574181962700095e-05,
      "learning_rate": 4.5633875148212946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4342576,
      "step": 7640
    },
    {
      "epoch": 134.12389380530973,
      "grad_norm": 3.2970062875392614e-06,
      "learning_rate": 4.562833047110883e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4345248,
      "step": 7645
    },
    {
      "epoch": 134.21238938053096,
      "grad_norm": 9.584137842466589e-06,
      "learning_rate": 4.5622782612856923e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4349056,
      "step": 7650
    },
    {
      "epoch": 134.30088495575222,
      "grad_norm": 1.0156657481275033e-05,
      "learning_rate": 4.561723157431278e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4351904,
      "step": 7655
    },
    {
      "epoch": 134.38938053097345,
      "grad_norm": 3.4439738101355033e-06,
      "learning_rate": 4.5611677356332435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4354880,
      "step": 7660
    },
    {
      "epoch": 134.47787610619469,
      "grad_norm": 9.414160558662843e-06,
      "learning_rate": 4.560611995977242e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4357456,
      "step": 7665
    },
    {
      "epoch": 134.56637168141592,
      "grad_norm": 3.841255420411471e-06,
      "learning_rate": 4.560055938548975e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4360032,
      "step": 7670
    },
    {
      "epoch": 134.65486725663717,
      "grad_norm": 3.302616732980823e-06,
      "learning_rate": 4.5594995634341944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4362768,
      "step": 7675
    },
    {
      "epoch": 134.7433628318584,
      "grad_norm": 4.301837179809809e-06,
      "learning_rate": 4.5589428707187e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4365536,
      "step": 7680
    },
    {
      "epoch": 134.83185840707964,
      "grad_norm": 3.6718081446451833e-06,
      "learning_rate": 4.55838586048834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4368656,
      "step": 7685
    },
    {
      "epoch": 134.9203539823009,
      "grad_norm": 2.730429059738526e-06,
      "learning_rate": 4.557828532829013e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4371408,
      "step": 7690
    },
    {
      "epoch": 135.0,
      "grad_norm": 0.00011756270396290347,
      "learning_rate": 4.557270887826667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4373936,
      "step": 7695
    },
    {
      "epoch": 135.08849557522123,
      "grad_norm": 2.3783693450241117e-06,
      "learning_rate": 4.556712925567296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4376576,
      "step": 7700
    },
    {
      "epoch": 135.1769911504425,
      "grad_norm": 5.1920401347160805e-06,
      "learning_rate": 4.5561546461369454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4379200,
      "step": 7705
    },
    {
      "epoch": 135.26548672566372,
      "grad_norm": 1.3893626601202413e-05,
      "learning_rate": 4.55559604962171e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4381760,
      "step": 7710
    },
    {
      "epoch": 135.35398230088495,
      "grad_norm": 1.5921269778118585e-06,
      "learning_rate": 4.55503713610773e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4384368,
      "step": 7715
    },
    {
      "epoch": 135.44247787610618,
      "grad_norm": 2.2600779630010948e-06,
      "learning_rate": 4.5544779056812e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4387568,
      "step": 7720
    },
    {
      "epoch": 135.53097345132744,
      "grad_norm": 1.0643651876307558e-05,
      "learning_rate": 4.553918358428358e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4390400,
      "step": 7725
    },
    {
      "epoch": 135.61946902654867,
      "grad_norm": 6.309224318101769e-06,
      "learning_rate": 4.553358494435494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4393760,
      "step": 7730
    },
    {
      "epoch": 135.7079646017699,
      "grad_norm": 5.059957857156405e-06,
      "learning_rate": 4.5527983137889464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4397072,
      "step": 7735
    },
    {
      "epoch": 135.79646017699116,
      "grad_norm": 3.537767724992591e-06,
      "learning_rate": 4.5522378165751015e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4400416,
      "step": 7740
    },
    {
      "epoch": 135.8849557522124,
      "grad_norm": 3.270695742685348e-05,
      "learning_rate": 4.5516770028803954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4403344,
      "step": 7745
    },
    {
      "epoch": 135.97345132743362,
      "grad_norm": 2.7026724183087936e-06,
      "learning_rate": 4.5511158727913116e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4406080,
      "step": 7750
    },
    {
      "epoch": 136.05309734513276,
      "grad_norm": 4.526101292867679e-06,
      "learning_rate": 4.5505544263943856e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4408264,
      "step": 7755
    },
    {
      "epoch": 136.141592920354,
      "grad_norm": 2.952226850538864e-06,
      "learning_rate": 4.549992663776197e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4410808,
      "step": 7760
    },
    {
      "epoch": 136.23008849557522,
      "grad_norm": 1.4057474800210912e-05,
      "learning_rate": 4.5494305850233786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4413320,
      "step": 7765
    },
    {
      "epoch": 136.31858407079645,
      "grad_norm": 1.2827884347643703e-05,
      "learning_rate": 4.5488681902226094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4415896,
      "step": 7770
    },
    {
      "epoch": 136.4070796460177,
      "grad_norm": 2.110542936861748e-06,
      "learning_rate": 4.5483054794606174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4418248,
      "step": 7775
    },
    {
      "epoch": 136.49557522123894,
      "grad_norm": 2.0365091586427297e-06,
      "learning_rate": 4.547742452824179e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4421496,
      "step": 7780
    },
    {
      "epoch": 136.58407079646017,
      "grad_norm": 2.432111841699225e-06,
      "learning_rate": 4.5471791104001215e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4424920,
      "step": 7785
    },
    {
      "epoch": 136.67256637168143,
      "grad_norm": 3.919995833712164e-06,
      "learning_rate": 4.546615452275319e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4428104,
      "step": 7790
    },
    {
      "epoch": 136.76106194690266,
      "grad_norm": 5.2742740081157535e-05,
      "learning_rate": 4.5460514785366944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4431048,
      "step": 7795
    },
    {
      "epoch": 136.8495575221239,
      "grad_norm": 2.2512954274134245e-06,
      "learning_rate": 4.545487189271219e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4434056,
      "step": 7800
    },
    {
      "epoch": 136.8495575221239,
      "eval_loss": 0.41203686594963074,
      "eval_runtime": 1.0572,
      "eval_samples_per_second": 23.647,
      "eval_steps_per_second": 12.297,
      "num_input_tokens_seen": 4434056,
      "step": 7800
    },
    {
      "epoch": 136.93805309734512,
      "grad_norm": 1.786596476449631e-05,
      "learning_rate": 4.544922584565914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4437096,
      "step": 7805
    },
    {
      "epoch": 137.01769911504425,
      "grad_norm": 3.5968316751677776e-06,
      "learning_rate": 4.544357664507848e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4439648,
      "step": 7810
    },
    {
      "epoch": 137.10619469026548,
      "grad_norm": 2.1679879864677787e-06,
      "learning_rate": 4.54379242918414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4442304,
      "step": 7815
    },
    {
      "epoch": 137.1946902654867,
      "grad_norm": 2.119195187333389e-06,
      "learning_rate": 4.543226878681955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4445168,
      "step": 7820
    },
    {
      "epoch": 137.28318584070797,
      "grad_norm": 1.929876088979654e-05,
      "learning_rate": 4.5426610130885087e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4447856,
      "step": 7825
    },
    {
      "epoch": 137.3716814159292,
      "grad_norm": 3.606687869250891e-06,
      "learning_rate": 4.542094832491064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4450352,
      "step": 7830
    },
    {
      "epoch": 137.46017699115043,
      "grad_norm": 4.187557806289988e-06,
      "learning_rate": 4.541528336976934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4452832,
      "step": 7835
    },
    {
      "epoch": 137.5486725663717,
      "grad_norm": 6.300292170635657e-06,
      "learning_rate": 4.540961526633479e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4456032,
      "step": 7840
    },
    {
      "epoch": 137.63716814159292,
      "grad_norm": 2.114472408720758e-06,
      "learning_rate": 4.540394401548108e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4459216,
      "step": 7845
    },
    {
      "epoch": 137.72566371681415,
      "grad_norm": 3.97470421376056e-06,
      "learning_rate": 4.539826961808279e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4462432,
      "step": 7850
    },
    {
      "epoch": 137.81415929203538,
      "grad_norm": 8.601730769441929e-06,
      "learning_rate": 4.5392592075014994e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4465264,
      "step": 7855
    },
    {
      "epoch": 137.90265486725664,
      "grad_norm": 2.0908232727379072e-06,
      "learning_rate": 4.538691138715322e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4467952,
      "step": 7860
    },
    {
      "epoch": 137.99115044247787,
      "grad_norm": 3.681661110022105e-05,
      "learning_rate": 4.5381227555373516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4470720,
      "step": 7865
    },
    {
      "epoch": 138.07079646017698,
      "grad_norm": 2.0508732632151805e-06,
      "learning_rate": 4.537554058055239e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4473336,
      "step": 7870
    },
    {
      "epoch": 138.15929203539824,
      "grad_norm": 3.3082028494391125e-06,
      "learning_rate": 4.5369850463566865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4475688,
      "step": 7875
    },
    {
      "epoch": 138.24778761061947,
      "grad_norm": 4.8592264647595584e-06,
      "learning_rate": 4.5364157205294404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4478488,
      "step": 7880
    },
    {
      "epoch": 138.3362831858407,
      "grad_norm": 4.642312433134066e-06,
      "learning_rate": 4.5358460806612996e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4481448,
      "step": 7885
    },
    {
      "epoch": 138.42477876106196,
      "grad_norm": 6.910349384270376e-06,
      "learning_rate": 4.535276126840109e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4484328,
      "step": 7890
    },
    {
      "epoch": 138.5132743362832,
      "grad_norm": 3.410339559195563e-06,
      "learning_rate": 4.5347058591537626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4487240,
      "step": 7895
    },
    {
      "epoch": 138.60176991150442,
      "grad_norm": 3.979014763899613e-06,
      "learning_rate": 4.534135277690203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4490040,
      "step": 7900
    },
    {
      "epoch": 138.69026548672565,
      "grad_norm": 1.8127454950445099e-06,
      "learning_rate": 4.533564382537421e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4492888,
      "step": 7905
    },
    {
      "epoch": 138.7787610619469,
      "grad_norm": 4.5591426896862686e-05,
      "learning_rate": 4.532993173783456e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4496184,
      "step": 7910
    },
    {
      "epoch": 138.86725663716814,
      "grad_norm": 5.3417311391967814e-06,
      "learning_rate": 4.5324216515163954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4499032,
      "step": 7915
    },
    {
      "epoch": 138.95575221238937,
      "grad_norm": 1.1237087164772674e-05,
      "learning_rate": 4.531849815824375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4501736,
      "step": 7920
    },
    {
      "epoch": 139.0353982300885,
      "grad_norm": 4.211789018881973e-06,
      "learning_rate": 4.5312776667955795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4504168,
      "step": 7925
    },
    {
      "epoch": 139.12389380530973,
      "grad_norm": 2.2328807972371578e-05,
      "learning_rate": 4.5307052045182405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4507304,
      "step": 7930
    },
    {
      "epoch": 139.21238938053096,
      "grad_norm": 4.27528038926539e-06,
      "learning_rate": 4.53013242908064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4509976,
      "step": 7935
    },
    {
      "epoch": 139.30088495575222,
      "grad_norm": 2.8108522656111745e-06,
      "learning_rate": 4.529559340571107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4512648,
      "step": 7940
    },
    {
      "epoch": 139.38938053097345,
      "grad_norm": 3.4033344036288327e-06,
      "learning_rate": 4.528985939078018e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4515400,
      "step": 7945
    },
    {
      "epoch": 139.47787610619469,
      "grad_norm": 1.7650753534326213e-06,
      "learning_rate": 4.5284122246898e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4518744,
      "step": 7950
    },
    {
      "epoch": 139.56637168141592,
      "grad_norm": 4.444790192792425e-06,
      "learning_rate": 4.527838197494926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4521736,
      "step": 7955
    },
    {
      "epoch": 139.65486725663717,
      "grad_norm": 8.977372090157587e-06,
      "learning_rate": 4.527263857581918e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4524632,
      "step": 7960
    },
    {
      "epoch": 139.7433628318584,
      "grad_norm": 2.9389955216174712e-06,
      "learning_rate": 4.526689205039347e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4527400,
      "step": 7965
    },
    {
      "epoch": 139.83185840707964,
      "grad_norm": 3.4931408663396724e-06,
      "learning_rate": 4.5261142399558324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4530216,
      "step": 7970
    },
    {
      "epoch": 139.9203539823009,
      "grad_norm": 2.1297814782883506e-06,
      "learning_rate": 4.525538962420041e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4533128,
      "step": 7975
    },
    {
      "epoch": 140.0,
      "grad_norm": 5.3458284128282685e-06,
      "learning_rate": 4.524963372520685e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4535568,
      "step": 7980
    },
    {
      "epoch": 140.08849557522123,
      "grad_norm": 1.6534692122149863e-06,
      "learning_rate": 4.524387470346531e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4538912,
      "step": 7985
    },
    {
      "epoch": 140.1769911504425,
      "grad_norm": 4.84796191813075e-06,
      "learning_rate": 4.5238112559863885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4541728,
      "step": 7990
    },
    {
      "epoch": 140.26548672566372,
      "grad_norm": 3.829223260254366e-06,
      "learning_rate": 4.5232347295291175e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4545056,
      "step": 7995
    },
    {
      "epoch": 140.35398230088495,
      "grad_norm": 1.6914887964958325e-05,
      "learning_rate": 4.522657891063626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4547840,
      "step": 8000
    },
    {
      "epoch": 140.35398230088495,
      "eval_loss": 0.42564472556114197,
      "eval_runtime": 1.0607,
      "eval_samples_per_second": 23.569,
      "eval_steps_per_second": 12.256,
      "num_input_tokens_seen": 4547840,
      "step": 8000
    },
    {
      "epoch": 140.44247787610618,
      "grad_norm": 4.5895358198322356e-06,
      "learning_rate": 4.52208074067887e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4550544,
      "step": 8005
    },
    {
      "epoch": 140.53097345132744,
      "grad_norm": 1.9826931747957133e-05,
      "learning_rate": 4.5215032784638516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4553536,
      "step": 8010
    },
    {
      "epoch": 140.61946902654867,
      "grad_norm": 6.184925950947218e-06,
      "learning_rate": 4.5209255045076245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4556096,
      "step": 8015
    },
    {
      "epoch": 140.7079646017699,
      "grad_norm": 8.2253354776185e-06,
      "learning_rate": 4.5203474188992875e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4558784,
      "step": 8020
    },
    {
      "epoch": 140.79646017699116,
      "grad_norm": 2.5840412490651943e-05,
      "learning_rate": 4.51976902172799e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4561232,
      "step": 8025
    },
    {
      "epoch": 140.8849557522124,
      "grad_norm": 6.779300747439265e-06,
      "learning_rate": 4.519190313082927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4564544,
      "step": 8030
    },
    {
      "epoch": 140.97345132743362,
      "grad_norm": 5.01085196447093e-06,
      "learning_rate": 4.518611293053343e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4567280,
      "step": 8035
    },
    {
      "epoch": 141.05309734513276,
      "grad_norm": 3.925979399355128e-06,
      "learning_rate": 4.51803196172853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4569480,
      "step": 8040
    },
    {
      "epoch": 141.141592920354,
      "grad_norm": 2.0951925762346946e-05,
      "learning_rate": 4.517452319197828e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4572360,
      "step": 8045
    },
    {
      "epoch": 141.23008849557522,
      "grad_norm": 3.056342393392697e-06,
      "learning_rate": 4.5168723655506265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4575320,
      "step": 8050
    },
    {
      "epoch": 141.31858407079645,
      "grad_norm": 9.85821861831937e-06,
      "learning_rate": 4.51629210087636e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4578104,
      "step": 8055
    },
    {
      "epoch": 141.4070796460177,
      "grad_norm": 4.330214323999826e-06,
      "learning_rate": 4.515711525264513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4581096,
      "step": 8060
    },
    {
      "epoch": 141.49557522123894,
      "grad_norm": 7.6644319051411e-06,
      "learning_rate": 4.5151306388046175e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4584184,
      "step": 8065
    },
    {
      "epoch": 141.58407079646017,
      "grad_norm": 1.9715548660315108e-06,
      "learning_rate": 4.514549441586255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4587320,
      "step": 8070
    },
    {
      "epoch": 141.67256637168143,
      "grad_norm": 7.330171683861408e-06,
      "learning_rate": 4.513967933699051e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4589944,
      "step": 8075
    },
    {
      "epoch": 141.76106194690266,
      "grad_norm": 3.5686007322510704e-06,
      "learning_rate": 4.513386115232684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4592568,
      "step": 8080
    },
    {
      "epoch": 141.8495575221239,
      "grad_norm": 4.7583721425326075e-06,
      "learning_rate": 4.5128039862768745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4595848,
      "step": 8085
    },
    {
      "epoch": 141.93805309734512,
      "grad_norm": 2.0434632460819557e-05,
      "learning_rate": 4.512221546921397e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4598648,
      "step": 8090
    },
    {
      "epoch": 142.01769911504425,
      "grad_norm": 3.316750053272699e-06,
      "learning_rate": 4.5116387972560694e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4601320,
      "step": 8095
    },
    {
      "epoch": 142.10619469026548,
      "grad_norm": 2.2368691134033725e-05,
      "learning_rate": 4.511055737370759e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4604488,
      "step": 8100
    },
    {
      "epoch": 142.1946902654867,
      "grad_norm": 2.5595429633540334e-06,
      "learning_rate": 4.510472367355383e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4607032,
      "step": 8105
    },
    {
      "epoch": 142.28318584070797,
      "grad_norm": 1.5209885759759345e-06,
      "learning_rate": 4.509888687299901e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4609544,
      "step": 8110
    },
    {
      "epoch": 142.3716814159292,
      "grad_norm": 4.145593720750185e-06,
      "learning_rate": 4.5093046972943266e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4612680,
      "step": 8115
    },
    {
      "epoch": 142.46017699115043,
      "grad_norm": 1.6163701729965396e-05,
      "learning_rate": 4.508720397428717e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4615560,
      "step": 8120
    },
    {
      "epoch": 142.5486725663717,
      "grad_norm": 2.861743723769905e-06,
      "learning_rate": 4.508135787793178e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4618184,
      "step": 8125
    },
    {
      "epoch": 142.63716814159292,
      "grad_norm": 1.8975624698214233e-05,
      "learning_rate": 4.5075508684778664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4620632,
      "step": 8130
    },
    {
      "epoch": 142.72566371681415,
      "grad_norm": 5.481399966811296e-06,
      "learning_rate": 4.506965639572982e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4623784,
      "step": 8135
    },
    {
      "epoch": 142.81415929203538,
      "grad_norm": 6.33726449450478e-06,
      "learning_rate": 4.506380101168774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4627128,
      "step": 8140
    },
    {
      "epoch": 142.90265486725664,
      "grad_norm": 3.4399345167912543e-06,
      "learning_rate": 4.505794253355542e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4630616,
      "step": 8145
    },
    {
      "epoch": 142.99115044247787,
      "grad_norm": 3.0541091291524936e-06,
      "learning_rate": 4.5052080962236286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4633144,
      "step": 8150
    },
    {
      "epoch": 143.07079646017698,
      "grad_norm": 5.613892881228821e-06,
      "learning_rate": 4.504621629863428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4635648,
      "step": 8155
    },
    {
      "epoch": 143.15929203539824,
      "grad_norm": 4.5793603931088e-06,
      "learning_rate": 4.504034854365381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4638848,
      "step": 8160
    },
    {
      "epoch": 143.24778761061947,
      "grad_norm": 7.540988463006215e-06,
      "learning_rate": 4.503447769819974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4641744,
      "step": 8165
    },
    {
      "epoch": 143.3362831858407,
      "grad_norm": 4.3375898712838534e-06,
      "learning_rate": 4.502860376317745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4644528,
      "step": 8170
    },
    {
      "epoch": 143.42477876106196,
      "grad_norm": 2.290726115461439e-06,
      "learning_rate": 4.502272673949276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4647616,
      "step": 8175
    },
    {
      "epoch": 143.5132743362832,
      "grad_norm": 2.5693877887533745e-06,
      "learning_rate": 4.501684662805199e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4650352,
      "step": 8180
    },
    {
      "epoch": 143.60176991150442,
      "grad_norm": 2.4864291390258586e-06,
      "learning_rate": 4.5010963429761924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4652992,
      "step": 8185
    },
    {
      "epoch": 143.69026548672565,
      "grad_norm": 1.8598071619635448e-05,
      "learning_rate": 4.500507714552982e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4655984,
      "step": 8190
    },
    {
      "epoch": 143.7787610619469,
      "grad_norm": 1.6271866343231522e-06,
      "learning_rate": 4.499918777626342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4659088,
      "step": 8195
    },
    {
      "epoch": 143.86725663716814,
      "grad_norm": 1.097534732252825e-05,
      "learning_rate": 4.499329532287093e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4662192,
      "step": 8200
    },
    {
      "epoch": 143.86725663716814,
      "eval_loss": 0.4116646647453308,
      "eval_runtime": 1.0629,
      "eval_samples_per_second": 23.52,
      "eval_steps_per_second": 12.231,
      "num_input_tokens_seen": 4662192,
      "step": 8200
    },
    {
      "epoch": 143.95575221238937,
      "grad_norm": 5.405311640060972e-06,
      "learning_rate": 4.4987399786261064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4664672,
      "step": 8205
    },
    {
      "epoch": 144.0353982300885,
      "grad_norm": 9.207245966535993e-06,
      "learning_rate": 4.498150116734297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4666768,
      "step": 8210
    },
    {
      "epoch": 144.12389380530973,
      "grad_norm": 6.065949037292739e-06,
      "learning_rate": 4.4975599467026294e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4669056,
      "step": 8215
    },
    {
      "epoch": 144.21238938053096,
      "grad_norm": 7.752300007268786e-06,
      "learning_rate": 4.496969468622114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4672064,
      "step": 8220
    },
    {
      "epoch": 144.30088495575222,
      "grad_norm": 5.087327735964209e-05,
      "learning_rate": 4.496378682583813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4675344,
      "step": 8225
    },
    {
      "epoch": 144.38938053097345,
      "grad_norm": 1.041206360241631e-05,
      "learning_rate": 4.495787588678829e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4678432,
      "step": 8230
    },
    {
      "epoch": 144.47787610619469,
      "grad_norm": 2.742755168583244e-05,
      "learning_rate": 4.4951961869983196e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4681344,
      "step": 8235
    },
    {
      "epoch": 144.56637168141592,
      "grad_norm": 2.851987119356636e-06,
      "learning_rate": 4.494604477633485e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4684192,
      "step": 8240
    },
    {
      "epoch": 144.65486725663717,
      "grad_norm": 3.094849489571061e-06,
      "learning_rate": 4.4940124606755734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4686688,
      "step": 8245
    },
    {
      "epoch": 144.7433628318584,
      "grad_norm": 2.2139333850645926e-06,
      "learning_rate": 4.493420136215882e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4689776,
      "step": 8250
    },
    {
      "epoch": 144.83185840707964,
      "grad_norm": 6.588014912267681e-06,
      "learning_rate": 4.492827504345756e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4692384,
      "step": 8255
    },
    {
      "epoch": 144.9203539823009,
      "grad_norm": 1.9723847799468786e-06,
      "learning_rate": 4.492234565156584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4695744,
      "step": 8260
    },
    {
      "epoch": 145.0,
      "grad_norm": 3.647387075034203e-06,
      "learning_rate": 4.491641318739807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4698160,
      "step": 8265
    },
    {
      "epoch": 145.08849557522123,
      "grad_norm": 2.5092817850236315e-06,
      "learning_rate": 4.4910477651869096e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4701280,
      "step": 8270
    },
    {
      "epoch": 145.1769911504425,
      "grad_norm": 3.7046268062113086e-06,
      "learning_rate": 4.4904539045894254e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4703744,
      "step": 8275
    },
    {
      "epoch": 145.26548672566372,
      "grad_norm": 5.296067229210166e-06,
      "learning_rate": 4.4898597370389364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4706736,
      "step": 8280
    },
    {
      "epoch": 145.35398230088495,
      "grad_norm": 7.240949798870133e-06,
      "learning_rate": 4.489265262627069e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4709312,
      "step": 8285
    },
    {
      "epoch": 145.44247787610618,
      "grad_norm": 2.397072648818721e-06,
      "learning_rate": 4.488670481445499e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4712288,
      "step": 8290
    },
    {
      "epoch": 145.53097345132744,
      "grad_norm": 1.7801595504352008e-06,
      "learning_rate": 4.488075393585951e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4715216,
      "step": 8295
    },
    {
      "epoch": 145.61946902654867,
      "grad_norm": 1.9458702809060924e-06,
      "learning_rate": 4.487479999140193e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4717840,
      "step": 8300
    },
    {
      "epoch": 145.7079646017699,
      "grad_norm": 2.2687925138598075e-06,
      "learning_rate": 4.4868842982000425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4720560,
      "step": 8305
    },
    {
      "epoch": 145.79646017699116,
      "grad_norm": 1.850227135946625e-06,
      "learning_rate": 4.486288290857365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4723824,
      "step": 8310
    },
    {
      "epoch": 145.8849557522124,
      "grad_norm": 2.2189674382389057e-06,
      "learning_rate": 4.4856919772040715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4726704,
      "step": 8315
    },
    {
      "epoch": 145.97345132743362,
      "grad_norm": 1.0788558938656934e-05,
      "learning_rate": 4.485095357332122e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4729616,
      "step": 8320
    },
    {
      "epoch": 146.05309734513276,
      "grad_norm": 2.578245357653941e-06,
      "learning_rate": 4.484498431333521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4731848,
      "step": 8325
    },
    {
      "epoch": 146.141592920354,
      "grad_norm": 2.972304400827852e-06,
      "learning_rate": 4.4839011993003245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4734840,
      "step": 8330
    },
    {
      "epoch": 146.23008849557522,
      "grad_norm": 1.2972931472177152e-05,
      "learning_rate": 4.4833036613246305e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4737624,
      "step": 8335
    },
    {
      "epoch": 146.31858407079645,
      "grad_norm": 2.699170863706968e-06,
      "learning_rate": 4.482705817498589e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4740712,
      "step": 8340
    },
    {
      "epoch": 146.4070796460177,
      "grad_norm": 3.682885562739102e-06,
      "learning_rate": 4.4821076679143934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4743656,
      "step": 8345
    },
    {
      "epoch": 146.49557522123894,
      "grad_norm": 2.457787559251301e-06,
      "learning_rate": 4.481509212664288e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4746168,
      "step": 8350
    },
    {
      "epoch": 146.58407079646017,
      "grad_norm": 5.3971207307768054e-06,
      "learning_rate": 4.480910451840559e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4748712,
      "step": 8355
    },
    {
      "epoch": 146.67256637168143,
      "grad_norm": 2.1926739464106504e-06,
      "learning_rate": 4.480311385535546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4751848,
      "step": 8360
    },
    {
      "epoch": 146.76106194690266,
      "grad_norm": 2.463199507474201e-06,
      "learning_rate": 4.47971201384163e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4754888,
      "step": 8365
    },
    {
      "epoch": 146.8495575221239,
      "grad_norm": 3.783110514632426e-06,
      "learning_rate": 4.4791123368512446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4757928,
      "step": 8370
    },
    {
      "epoch": 146.93805309734512,
      "grad_norm": 4.340073701314395e-06,
      "learning_rate": 4.478512354656864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4760760,
      "step": 8375
    },
    {
      "epoch": 147.01769911504425,
      "grad_norm": 3.091987537118257e-06,
      "learning_rate": 4.477912067351016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4763120,
      "step": 8380
    },
    {
      "epoch": 147.10619469026548,
      "grad_norm": 2.2610079213336576e-06,
      "learning_rate": 4.477311475026271e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4765792,
      "step": 8385
    },
    {
      "epoch": 147.1946902654867,
      "grad_norm": 2.685099616428488e-06,
      "learning_rate": 4.476710577775248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4768768,
      "step": 8390
    },
    {
      "epoch": 147.28318584070797,
      "grad_norm": 2.823209342750488e-06,
      "learning_rate": 4.476109375690612e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4771664,
      "step": 8395
    },
    {
      "epoch": 147.3716814159292,
      "grad_norm": 4.951292794430628e-06,
      "learning_rate": 4.4755078688650784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4774160,
      "step": 8400
    },
    {
      "epoch": 147.3716814159292,
      "eval_loss": 0.4215090572834015,
      "eval_runtime": 1.0662,
      "eval_samples_per_second": 23.448,
      "eval_steps_per_second": 12.193,
      "num_input_tokens_seen": 4774160,
      "step": 8400
    },
    {
      "epoch": 147.46017699115043,
      "grad_norm": 1.5386489167212858e-06,
      "learning_rate": 4.474906057391406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4776896,
      "step": 8405
    },
    {
      "epoch": 147.5486725663717,
      "grad_norm": 9.992721970775165e-06,
      "learning_rate": 4.4743039413624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4779664,
      "step": 8410
    },
    {
      "epoch": 147.63716814159292,
      "grad_norm": 4.848473054153146e-06,
      "learning_rate": 4.473701520870916e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4782400,
      "step": 8415
    },
    {
      "epoch": 147.72566371681415,
      "grad_norm": 5.746150691265939e-06,
      "learning_rate": 4.4730987960098544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4785120,
      "step": 8420
    },
    {
      "epoch": 147.81415929203538,
      "grad_norm": 2.9139882826711982e-06,
      "learning_rate": 4.4724957668721635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4787952,
      "step": 8425
    },
    {
      "epoch": 147.90265486725664,
      "grad_norm": 5.608481842500623e-06,
      "learning_rate": 4.471892433550836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4790736,
      "step": 8430
    },
    {
      "epoch": 147.99115044247787,
      "grad_norm": 2.738601551754982e-06,
      "learning_rate": 4.471288796138916e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4794000,
      "step": 8435
    },
    {
      "epoch": 148.07079646017698,
      "grad_norm": 2.4779574232525192e-06,
      "learning_rate": 4.470684854729491e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4796384,
      "step": 8440
    },
    {
      "epoch": 148.15929203539824,
      "grad_norm": 1.6739494412831846e-06,
      "learning_rate": 4.4700806094156955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4799536,
      "step": 8445
    },
    {
      "epoch": 148.24778761061947,
      "grad_norm": 1.653581603022758e-05,
      "learning_rate": 4.469476060290713e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4801936,
      "step": 8450
    },
    {
      "epoch": 148.3362831858407,
      "grad_norm": 5.1101351346005686e-06,
      "learning_rate": 4.468871207447772e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4804416,
      "step": 8455
    },
    {
      "epoch": 148.42477876106196,
      "grad_norm": 1.6668664102326147e-06,
      "learning_rate": 4.4682660509801486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4807184,
      "step": 8460
    },
    {
      "epoch": 148.5132743362832,
      "grad_norm": 3.62906916961947e-06,
      "learning_rate": 4.467660590981165e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4809936,
      "step": 8465
    },
    {
      "epoch": 148.60176991150442,
      "grad_norm": 2.826266381816822e-06,
      "learning_rate": 4.467054827544191e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4812720,
      "step": 8470
    },
    {
      "epoch": 148.69026548672565,
      "grad_norm": 1.1949457075388636e-05,
      "learning_rate": 4.4664487607626434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4815968,
      "step": 8475
    },
    {
      "epoch": 148.7787610619469,
      "grad_norm": 2.304011104570236e-06,
      "learning_rate": 4.4658423907299845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4818896,
      "step": 8480
    },
    {
      "epoch": 148.86725663716814,
      "grad_norm": 1.5659556083846837e-05,
      "learning_rate": 4.465235717539725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4821952,
      "step": 8485
    },
    {
      "epoch": 148.95575221238937,
      "grad_norm": 6.651870080531808e-06,
      "learning_rate": 4.464628741285421e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4824848,
      "step": 8490
    },
    {
      "epoch": 149.0353982300885,
      "grad_norm": 1.350530396848626e-06,
      "learning_rate": 4.4640214620606754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4827704,
      "step": 8495
    },
    {
      "epoch": 149.12389380530973,
      "grad_norm": 1.1638264368230011e-05,
      "learning_rate": 4.46341387995914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4830552,
      "step": 8500
    },
    {
      "epoch": 149.21238938053096,
      "grad_norm": 2.7355708880349994e-06,
      "learning_rate": 4.4628059950745106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4833368,
      "step": 8505
    },
    {
      "epoch": 149.30088495575222,
      "grad_norm": 1.860564952949062e-05,
      "learning_rate": 4.4621978075005297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4836584,
      "step": 8510
    },
    {
      "epoch": 149.38938053097345,
      "grad_norm": 2.1417294192360714e-05,
      "learning_rate": 4.461589317330989e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4839592,
      "step": 8515
    },
    {
      "epoch": 149.47787610619469,
      "grad_norm": 7.700564310653135e-06,
      "learning_rate": 4.460980524659724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4842328,
      "step": 8520
    },
    {
      "epoch": 149.56637168141592,
      "grad_norm": 2.6545396849542158e-06,
      "learning_rate": 4.46037142958062e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4844728,
      "step": 8525
    },
    {
      "epoch": 149.65486725663717,
      "grad_norm": 1.6028875506890472e-06,
      "learning_rate": 4.4597620321876046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4847288,
      "step": 8530
    },
    {
      "epoch": 149.7433628318584,
      "grad_norm": 2.7378773665986955e-06,
      "learning_rate": 4.459152332574656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4850424,
      "step": 8535
    },
    {
      "epoch": 149.83185840707964,
      "grad_norm": 2.697747504498693e-06,
      "learning_rate": 4.4585423308357985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4853496,
      "step": 8540
    },
    {
      "epoch": 149.9203539823009,
      "grad_norm": 2.5504862151137786e-06,
      "learning_rate": 4.457932027065102e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4856296,
      "step": 8545
    },
    {
      "epoch": 150.0,
      "grad_norm": 1.0962587566609727e-06,
      "learning_rate": 4.45732142135668e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4859016,
      "step": 8550
    },
    {
      "epoch": 150.08849557522123,
      "grad_norm": 2.509714022380649e-06,
      "learning_rate": 4.4567105138046986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4861608,
      "step": 8555
    },
    {
      "epoch": 150.1769911504425,
      "grad_norm": 1.5174282452790067e-05,
      "learning_rate": 4.456099304503365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4864600,
      "step": 8560
    },
    {
      "epoch": 150.26548672566372,
      "grad_norm": 9.06403784028953e-06,
      "learning_rate": 4.455487793546939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4867384,
      "step": 8565
    },
    {
      "epoch": 150.35398230088495,
      "grad_norm": 2.097520700772293e-05,
      "learning_rate": 4.454875981029719e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4870200,
      "step": 8570
    },
    {
      "epoch": 150.44247787610618,
      "grad_norm": 2.1249124984024093e-06,
      "learning_rate": 4.454263867046057e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4873048,
      "step": 8575
    },
    {
      "epoch": 150.53097345132744,
      "grad_norm": 7.860955520300195e-06,
      "learning_rate": 4.4536514516903484e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4875720,
      "step": 8580
    },
    {
      "epoch": 150.61946902654867,
      "grad_norm": 1.4642835139966337e-06,
      "learning_rate": 4.453038735057034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4878888,
      "step": 8585
    },
    {
      "epoch": 150.7079646017699,
      "grad_norm": 1.808304318728915e-06,
      "learning_rate": 4.4524257172406034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4881928,
      "step": 8590
    },
    {
      "epoch": 150.79646017699116,
      "grad_norm": 4.859053206018871e-06,
      "learning_rate": 4.451812398335592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4884856,
      "step": 8595
    },
    {
      "epoch": 150.8849557522124,
      "grad_norm": 9.087668331630994e-06,
      "learning_rate": 4.4511987784365805e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4887640,
      "step": 8600
    },
    {
      "epoch": 150.8849557522124,
      "eval_loss": 0.4240879714488983,
      "eval_runtime": 1.0637,
      "eval_samples_per_second": 23.502,
      "eval_steps_per_second": 12.221,
      "num_input_tokens_seen": 4887640,
      "step": 8600
    },
    {
      "epoch": 150.97345132743362,
      "grad_norm": 2.0486079392867396e-06,
      "learning_rate": 4.450584857638197e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4890808,
      "step": 8605
    },
    {
      "epoch": 151.05309734513276,
      "grad_norm": 4.470518888410879e-06,
      "learning_rate": 4.449970636035116e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4893112,
      "step": 8610
    },
    {
      "epoch": 151.141592920354,
      "grad_norm": 1.1497131708892994e-06,
      "learning_rate": 4.4493561137220574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4895576,
      "step": 8615
    },
    {
      "epoch": 151.23008849557522,
      "grad_norm": 6.47743945592083e-06,
      "learning_rate": 4.44874129079379e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4898424,
      "step": 8620
    },
    {
      "epoch": 151.31858407079645,
      "grad_norm": 4.0662575884198304e-06,
      "learning_rate": 4.4481261673451255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4901528,
      "step": 8625
    },
    {
      "epoch": 151.4070796460177,
      "grad_norm": 1.4127653230389114e-05,
      "learning_rate": 4.4475107434709245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4904328,
      "step": 8630
    },
    {
      "epoch": 151.49557522123894,
      "grad_norm": 1.9097353742836276e-06,
      "learning_rate": 4.446895019266093e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4907400,
      "step": 8635
    },
    {
      "epoch": 151.58407079646017,
      "grad_norm": 3.3357043776049977e-06,
      "learning_rate": 4.446278994825583e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4910600,
      "step": 8640
    },
    {
      "epoch": 151.67256637168143,
      "grad_norm": 2.1817484139319276e-06,
      "learning_rate": 4.445662670244394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4914008,
      "step": 8645
    },
    {
      "epoch": 151.76106194690266,
      "grad_norm": 2.495061835361412e-06,
      "learning_rate": 4.44504604561757e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4916760,
      "step": 8650
    },
    {
      "epoch": 151.8495575221239,
      "grad_norm": 2.5767794795683585e-06,
      "learning_rate": 4.4444291210402035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4919976,
      "step": 8655
    },
    {
      "epoch": 151.93805309734512,
      "grad_norm": 4.793405878444901e-06,
      "learning_rate": 4.443811896607431e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4922728,
      "step": 8660
    },
    {
      "epoch": 152.01769911504425,
      "grad_norm": 3.591811719161342e-06,
      "learning_rate": 4.443194372414436e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4925128,
      "step": 8665
    },
    {
      "epoch": 152.10619469026548,
      "grad_norm": 9.939459232555237e-06,
      "learning_rate": 4.442576548556449e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4927704,
      "step": 8670
    },
    {
      "epoch": 152.1946902654867,
      "grad_norm": 2.366302851442015e-06,
      "learning_rate": 4.441958425128747e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4930392,
      "step": 8675
    },
    {
      "epoch": 152.28318584070797,
      "grad_norm": 1.8970064274981269e-06,
      "learning_rate": 4.4413400022266515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4933624,
      "step": 8680
    },
    {
      "epoch": 152.3716814159292,
      "grad_norm": 8.257059562311042e-06,
      "learning_rate": 4.4407212799455313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4936360,
      "step": 8685
    },
    {
      "epoch": 152.46017699115043,
      "grad_norm": 6.0881820900249295e-06,
      "learning_rate": 4.4401022583808003e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4939368,
      "step": 8690
    },
    {
      "epoch": 152.5486725663717,
      "grad_norm": 3.6269461816118564e-06,
      "learning_rate": 4.439482937627921e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4942776,
      "step": 8695
    },
    {
      "epoch": 152.63716814159292,
      "grad_norm": 3.0900794172339374e-06,
      "learning_rate": 4.4388633177824004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4945656,
      "step": 8700
    },
    {
      "epoch": 152.72566371681415,
      "grad_norm": 1.460772523387277e-06,
      "learning_rate": 4.4382433989397895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4948312,
      "step": 8705
    },
    {
      "epoch": 152.81415929203538,
      "grad_norm": 4.096951215615263e-06,
      "learning_rate": 4.4376231811956895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4951720,
      "step": 8710
    },
    {
      "epoch": 152.90265486725664,
      "grad_norm": 1.6701754930181778e-06,
      "learning_rate": 4.437002664645745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4954392,
      "step": 8715
    },
    {
      "epoch": 152.99115044247787,
      "grad_norm": 3.2093568734126166e-05,
      "learning_rate": 4.436381849385649e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4957192,
      "step": 8720
    },
    {
      "epoch": 153.07079646017698,
      "grad_norm": 1.854076913332392e-06,
      "learning_rate": 4.435760735511136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4959344,
      "step": 8725
    },
    {
      "epoch": 153.15929203539824,
      "grad_norm": 3.026838385267183e-06,
      "learning_rate": 4.435139323117992e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4963120,
      "step": 8730
    },
    {
      "epoch": 153.24778761061947,
      "grad_norm": 2.188716962336912e-06,
      "learning_rate": 4.434517612302046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4965952,
      "step": 8735
    },
    {
      "epoch": 153.3362831858407,
      "grad_norm": 3.2036179618444294e-05,
      "learning_rate": 4.433895603159174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4968464,
      "step": 8740
    },
    {
      "epoch": 153.42477876106196,
      "grad_norm": 1.6530248103663325e-06,
      "learning_rate": 4.433273295785296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4971920,
      "step": 8745
    },
    {
      "epoch": 153.5132743362832,
      "grad_norm": 2.923463171100593e-06,
      "learning_rate": 4.432650690276382e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4974480,
      "step": 8750
    },
    {
      "epoch": 153.60176991150442,
      "grad_norm": 2.524802994230413e-06,
      "learning_rate": 4.4320277867284435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4977408,
      "step": 8755
    },
    {
      "epoch": 153.69026548672565,
      "grad_norm": 1.6590936411375878e-06,
      "learning_rate": 4.431404585237541e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4980384,
      "step": 8760
    },
    {
      "epoch": 153.7787610619469,
      "grad_norm": 2.928639560195734e-06,
      "learning_rate": 4.43078108589978e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4983376,
      "step": 8765
    },
    {
      "epoch": 153.86725663716814,
      "grad_norm": 7.4029921961482614e-06,
      "learning_rate": 4.4301572888113116e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4986032,
      "step": 8770
    },
    {
      "epoch": 153.95575221238937,
      "grad_norm": 1.5396594790217932e-06,
      "learning_rate": 4.4295331940683337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4988576,
      "step": 8775
    },
    {
      "epoch": 154.0353982300885,
      "grad_norm": 6.01530700805597e-06,
      "learning_rate": 4.428908801767089e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4990864,
      "step": 8780
    },
    {
      "epoch": 154.12389380530973,
      "grad_norm": 1.2875876564066857e-05,
      "learning_rate": 4.428284112003868e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4993824,
      "step": 8785
    },
    {
      "epoch": 154.21238938053096,
      "grad_norm": 4.203006483294303e-06,
      "learning_rate": 4.4276591248750033e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4997072,
      "step": 8790
    },
    {
      "epoch": 154.30088495575222,
      "grad_norm": 2.229471647297032e-06,
      "learning_rate": 4.4270338404768774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 4999696,
      "step": 8795
    },
    {
      "epoch": 154.38938053097345,
      "grad_norm": 2.4698983907001093e-05,
      "learning_rate": 4.426408258905917e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5002864,
      "step": 8800
    },
    {
      "epoch": 154.38938053097345,
      "eval_loss": 0.42251870036125183,
      "eval_runtime": 1.0594,
      "eval_samples_per_second": 23.599,
      "eval_steps_per_second": 12.271,
      "num_input_tokens_seen": 5002864,
      "step": 8800
    },
    {
      "epoch": 154.47787610619469,
      "grad_norm": 2.4004684746614657e-06,
      "learning_rate": 4.425782380258594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5005568,
      "step": 8805
    },
    {
      "epoch": 154.56637168141592,
      "grad_norm": 4.02778823627159e-06,
      "learning_rate": 4.425156204631427e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5008672,
      "step": 8810
    },
    {
      "epoch": 154.65486725663717,
      "grad_norm": 2.8618371743505122e-06,
      "learning_rate": 4.424529732120981e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5011440,
      "step": 8815
    },
    {
      "epoch": 154.7433628318584,
      "grad_norm": 2.3892475837783422e-06,
      "learning_rate": 4.423902962823864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5014576,
      "step": 8820
    },
    {
      "epoch": 154.83185840707964,
      "grad_norm": 8.455752322333865e-06,
      "learning_rate": 4.423275896836733e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5017440,
      "step": 8825
    },
    {
      "epoch": 154.9203539823009,
      "grad_norm": 1.4868960533931386e-05,
      "learning_rate": 4.42264853425629e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5019984,
      "step": 8830
    },
    {
      "epoch": 155.0,
      "grad_norm": 2.058408426819369e-06,
      "learning_rate": 4.4220208751792816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5022264,
      "step": 8835
    },
    {
      "epoch": 155.08849557522123,
      "grad_norm": 1.8329685190110467e-05,
      "learning_rate": 4.421392919702499e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5025208,
      "step": 8840
    },
    {
      "epoch": 155.1769911504425,
      "grad_norm": 1.457069993193727e-05,
      "learning_rate": 4.4207646679227846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5027896,
      "step": 8845
    },
    {
      "epoch": 155.26548672566372,
      "grad_norm": 2.751716010607197e-06,
      "learning_rate": 4.42013611993702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5031048,
      "step": 8850
    },
    {
      "epoch": 155.35398230088495,
      "grad_norm": 2.1705538983951556e-06,
      "learning_rate": 4.419507275842135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5033688,
      "step": 8855
    },
    {
      "epoch": 155.44247787610618,
      "grad_norm": 1.9722990600712365e-06,
      "learning_rate": 4.418878135735106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5036840,
      "step": 8860
    },
    {
      "epoch": 155.53097345132744,
      "grad_norm": 3.163763039992773e-06,
      "learning_rate": 4.418248699712955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5039656,
      "step": 8865
    },
    {
      "epoch": 155.61946902654867,
      "grad_norm": 2.6164766495639924e-06,
      "learning_rate": 4.417618967872748e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5042568,
      "step": 8870
    },
    {
      "epoch": 155.7079646017699,
      "grad_norm": 2.0870202206424437e-06,
      "learning_rate": 4.4169889403115985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5045416,
      "step": 8875
    },
    {
      "epoch": 155.79646017699116,
      "grad_norm": 1.691374109213939e-06,
      "learning_rate": 4.4163586171266627e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5048088,
      "step": 8880
    },
    {
      "epoch": 155.8849557522124,
      "grad_norm": 3.0937690098653547e-06,
      "learning_rate": 4.415727998415147e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5050776,
      "step": 8885
    },
    {
      "epoch": 155.97345132743362,
      "grad_norm": 6.382872925314587e-06,
      "learning_rate": 4.4150970842742985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5053928,
      "step": 8890
    },
    {
      "epoch": 156.05309734513276,
      "grad_norm": 1.1005130318153533e-06,
      "learning_rate": 4.4144658748014134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5056392,
      "step": 8895
    },
    {
      "epoch": 156.141592920354,
      "grad_norm": 2.409193029961898e-06,
      "learning_rate": 4.413834370093831e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5059352,
      "step": 8900
    },
    {
      "epoch": 156.23008849557522,
      "grad_norm": 7.17720376997022e-06,
      "learning_rate": 4.413202570248939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5062344,
      "step": 8905
    },
    {
      "epoch": 156.31858407079645,
      "grad_norm": 2.511694583517965e-05,
      "learning_rate": 4.412570475364167e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5065304,
      "step": 8910
    },
    {
      "epoch": 156.4070796460177,
      "grad_norm": 4.960398655384779e-06,
      "learning_rate": 4.411938085536994e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5067720,
      "step": 8915
    },
    {
      "epoch": 156.49557522123894,
      "grad_norm": 1.295053834837745e-06,
      "learning_rate": 4.41130540086494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5071192,
      "step": 8920
    },
    {
      "epoch": 156.58407079646017,
      "grad_norm": 1.6264853002212476e-06,
      "learning_rate": 4.4106724214455754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5074184,
      "step": 8925
    },
    {
      "epoch": 156.67256637168143,
      "grad_norm": 1.5824704178157845e-06,
      "learning_rate": 4.4100391473765115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5076616,
      "step": 8930
    },
    {
      "epoch": 156.76106194690266,
      "grad_norm": 7.391477083729114e-06,
      "learning_rate": 4.409405578755408e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5078904,
      "step": 8935
    },
    {
      "epoch": 156.8495575221239,
      "grad_norm": 1.0692667274270207e-05,
      "learning_rate": 4.4087717156799705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5081912,
      "step": 8940
    },
    {
      "epoch": 156.93805309734512,
      "grad_norm": 4.28622752224328e-06,
      "learning_rate": 4.408137558247946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5085192,
      "step": 8945
    },
    {
      "epoch": 157.01769911504425,
      "grad_norm": 2.197141384385759e-06,
      "learning_rate": 4.4075031065571306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5087352,
      "step": 8950
    },
    {
      "epoch": 157.10619469026548,
      "grad_norm": 5.476168553286698e-06,
      "learning_rate": 4.406868360705366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5090616,
      "step": 8955
    },
    {
      "epoch": 157.1946902654867,
      "grad_norm": 3.0984699606051436e-06,
      "learning_rate": 4.406233320790536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5093528,
      "step": 8960
    },
    {
      "epoch": 157.28318584070797,
      "grad_norm": 2.074478743452346e-06,
      "learning_rate": 4.4055979869105734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5096264,
      "step": 8965
    },
    {
      "epoch": 157.3716814159292,
      "grad_norm": 5.590770797425648e-06,
      "learning_rate": 4.404962359163454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5099288,
      "step": 8970
    },
    {
      "epoch": 157.46017699115043,
      "grad_norm": 2.451438831485575e-06,
      "learning_rate": 4.404326437647199e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5102136,
      "step": 8975
    },
    {
      "epoch": 157.5486725663717,
      "grad_norm": 2.127542529706261e-06,
      "learning_rate": 4.403690222459877e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5104952,
      "step": 8980
    },
    {
      "epoch": 157.63716814159292,
      "grad_norm": 1.4083464066061424e-06,
      "learning_rate": 4.4030537136995984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5107672,
      "step": 8985
    },
    {
      "epoch": 157.72566371681415,
      "grad_norm": 1.3542476153816096e-06,
      "learning_rate": 4.402416911464523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5110504,
      "step": 8990
    },
    {
      "epoch": 157.81415929203538,
      "grad_norm": 7.821465260349214e-06,
      "learning_rate": 4.4017798158528516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5113192,
      "step": 8995
    },
    {
      "epoch": 157.90265486725664,
      "grad_norm": 2.279006821481744e-06,
      "learning_rate": 4.401142426962834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5116216,
      "step": 9000
    },
    {
      "epoch": 157.90265486725664,
      "eval_loss": 0.4309137165546417,
      "eval_runtime": 1.0628,
      "eval_samples_per_second": 23.523,
      "eval_steps_per_second": 12.232,
      "num_input_tokens_seen": 5116216,
      "step": 9000
    },
    {
      "epoch": 157.99115044247787,
      "grad_norm": 2.292756562383147e-06,
      "learning_rate": 4.400504744892763e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5119064,
      "step": 9005
    },
    {
      "epoch": 158.07079646017698,
      "grad_norm": 4.719136995845474e-06,
      "learning_rate": 4.399866769740975e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5121440,
      "step": 9010
    },
    {
      "epoch": 158.15929203539824,
      "grad_norm": 3.8308726288960315e-06,
      "learning_rate": 4.399228501605859e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5124080,
      "step": 9015
    },
    {
      "epoch": 158.24778761061947,
      "grad_norm": 1.3571859199146274e-05,
      "learning_rate": 4.398589940585839e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5126608,
      "step": 9020
    },
    {
      "epoch": 158.3362831858407,
      "grad_norm": 6.201603810040979e-06,
      "learning_rate": 4.3979510867793917e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5129872,
      "step": 9025
    },
    {
      "epoch": 158.42477876106196,
      "grad_norm": 1.1279290447419044e-05,
      "learning_rate": 4.3973119402850346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5132784,
      "step": 9030
    },
    {
      "epoch": 158.5132743362832,
      "grad_norm": 1.7882052816275973e-06,
      "learning_rate": 4.396672501201334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5135456,
      "step": 9035
    },
    {
      "epoch": 158.60176991150442,
      "grad_norm": 2.6624363727023592e-06,
      "learning_rate": 4.396032769626899e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5138720,
      "step": 9040
    },
    {
      "epoch": 158.69026548672565,
      "grad_norm": 3.852382178592961e-06,
      "learning_rate": 4.395392745660384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5141424,
      "step": 9045
    },
    {
      "epoch": 158.7787610619469,
      "grad_norm": 4.138967597100418e-06,
      "learning_rate": 4.394752429400488e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5144544,
      "step": 9050
    },
    {
      "epoch": 158.86725663716814,
      "grad_norm": 1.2021828297292814e-05,
      "learning_rate": 4.394111820945957e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5147264,
      "step": 9055
    },
    {
      "epoch": 158.95575221238937,
      "grad_norm": 3.0615685773227597e-06,
      "learning_rate": 4.393470920395579e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5150592,
      "step": 9060
    },
    {
      "epoch": 159.0353982300885,
      "grad_norm": 9.276913033318124e-07,
      "learning_rate": 4.392829727848192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5153320,
      "step": 9065
    },
    {
      "epoch": 159.12389380530973,
      "grad_norm": 3.149553322145948e-06,
      "learning_rate": 4.392188243402673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5156296,
      "step": 9070
    },
    {
      "epoch": 159.21238938053096,
      "grad_norm": 1.4226300208974862e-06,
      "learning_rate": 4.391546467157949e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5159224,
      "step": 9075
    },
    {
      "epoch": 159.30088495575222,
      "grad_norm": 1.688409065536689e-05,
      "learning_rate": 4.390904399212988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5162072,
      "step": 9080
    },
    {
      "epoch": 159.38938053097345,
      "grad_norm": 1.933507519424893e-05,
      "learning_rate": 4.390262039666807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5164808,
      "step": 9085
    },
    {
      "epoch": 159.47787610619469,
      "grad_norm": 4.106806045456324e-06,
      "learning_rate": 4.389619388618464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5167720,
      "step": 9090
    },
    {
      "epoch": 159.56637168141592,
      "grad_norm": 2.5808265036175726e-06,
      "learning_rate": 4.3889764461670655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5170504,
      "step": 9095
    },
    {
      "epoch": 159.65486725663717,
      "grad_norm": 3.102349865002907e-06,
      "learning_rate": 4.38833321241176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5173800,
      "step": 9100
    },
    {
      "epoch": 159.7433628318584,
      "grad_norm": 1.6900910395634128e-06,
      "learning_rate": 4.3876896874517434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5176872,
      "step": 9105
    },
    {
      "epoch": 159.83185840707964,
      "grad_norm": 2.925702347056358e-06,
      "learning_rate": 4.3870458713862554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5179384,
      "step": 9110
    },
    {
      "epoch": 159.9203539823009,
      "grad_norm": 2.146964789062622e-06,
      "learning_rate": 4.386401764314579e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5181848,
      "step": 9115
    },
    {
      "epoch": 160.0,
      "grad_norm": 1.856878384387528e-06,
      "learning_rate": 4.385757366336045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5184128,
      "step": 9120
    },
    {
      "epoch": 160.08849557522123,
      "grad_norm": 6.98757185091381e-06,
      "learning_rate": 4.385112677550027e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5187152,
      "step": 9125
    },
    {
      "epoch": 160.1769911504425,
      "grad_norm": 3.3028363759513013e-06,
      "learning_rate": 4.384467698055945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5189888,
      "step": 9130
    },
    {
      "epoch": 160.26548672566372,
      "grad_norm": 1.296956497753854e-06,
      "learning_rate": 4.383822427953261e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5192880,
      "step": 9135
    },
    {
      "epoch": 160.35398230088495,
      "grad_norm": 2.132667759724427e-06,
      "learning_rate": 4.3831768673414864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5195648,
      "step": 9140
    },
    {
      "epoch": 160.44247787610618,
      "grad_norm": 7.070972060319036e-06,
      "learning_rate": 4.382531016320173e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5198352,
      "step": 9145
    },
    {
      "epoch": 160.53097345132744,
      "grad_norm": 1.0359089174016844e-05,
      "learning_rate": 4.3818848749889184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5200992,
      "step": 9150
    },
    {
      "epoch": 160.61946902654867,
      "grad_norm": 3.7415645692817634e-06,
      "learning_rate": 4.381238443447368e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5204016,
      "step": 9155
    },
    {
      "epoch": 160.7079646017699,
      "grad_norm": 1.3368445252126548e-05,
      "learning_rate": 4.380591721795208e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5207072,
      "step": 9160
    },
    {
      "epoch": 160.79646017699116,
      "grad_norm": 4.945915407006396e-06,
      "learning_rate": 4.3799447101321723e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5210256,
      "step": 9165
    },
    {
      "epoch": 160.8849557522124,
      "grad_norm": 3.3981402793870075e-06,
      "learning_rate": 4.379297408558036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5213088,
      "step": 9170
    },
    {
      "epoch": 160.97345132743362,
      "grad_norm": 3.278984650023631e-06,
      "learning_rate": 4.378649817172624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5215904,
      "step": 9175
    },
    {
      "epoch": 161.05309734513276,
      "grad_norm": 3.1980157473299187e-06,
      "learning_rate": 4.378001936075801e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5218216,
      "step": 9180
    },
    {
      "epoch": 161.141592920354,
      "grad_norm": 2.4766843580437126e-06,
      "learning_rate": 4.377353765367479e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5220856,
      "step": 9185
    },
    {
      "epoch": 161.23008849557522,
      "grad_norm": 3.5472876334097236e-06,
      "learning_rate": 4.376705305147614e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5223368,
      "step": 9190
    },
    {
      "epoch": 161.31858407079645,
      "grad_norm": 8.481228178425226e-06,
      "learning_rate": 4.376056555516206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5226040,
      "step": 9195
    },
    {
      "epoch": 161.4070796460177,
      "grad_norm": 7.306916813831776e-06,
      "learning_rate": 4.375407516573302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5229496,
      "step": 9200
    },
    {
      "epoch": 161.4070796460177,
      "eval_loss": 0.42694327235221863,
      "eval_runtime": 1.0699,
      "eval_samples_per_second": 23.367,
      "eval_steps_per_second": 12.151,
      "num_input_tokens_seen": 5229496,
      "step": 9200
    },
    {
      "epoch": 161.49557522123894,
      "grad_norm": 5.150469405634794e-06,
      "learning_rate": 4.3747581884189913e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5232456,
      "step": 9205
    },
    {
      "epoch": 161.58407079646017,
      "grad_norm": 4.430695298651699e-06,
      "learning_rate": 4.374108571153408e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5235352,
      "step": 9210
    },
    {
      "epoch": 161.67256637168143,
      "grad_norm": 1.7469490103394492e-06,
      "learning_rate": 4.3734586648767316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5238584,
      "step": 9215
    },
    {
      "epoch": 161.76106194690266,
      "grad_norm": 2.256847210446722e-06,
      "learning_rate": 4.372808469689186e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5241432,
      "step": 9220
    },
    {
      "epoch": 161.8495575221239,
      "grad_norm": 3.3563301258254796e-06,
      "learning_rate": 4.372157985691039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5243896,
      "step": 9225
    },
    {
      "epoch": 161.93805309734512,
      "grad_norm": 8.33633566799108e-06,
      "learning_rate": 4.371507212982603e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5246424,
      "step": 9230
    },
    {
      "epoch": 162.01769911504425,
      "grad_norm": 1.781997525540646e-05,
      "learning_rate": 4.370856151664236e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5248776,
      "step": 9235
    },
    {
      "epoch": 162.10619469026548,
      "grad_norm": 2.408954060229007e-06,
      "learning_rate": 4.3702048018363404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5251608,
      "step": 9240
    },
    {
      "epoch": 162.1946902654867,
      "grad_norm": 1.296849063692207e-06,
      "learning_rate": 4.369553163599362e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5254232,
      "step": 9245
    },
    {
      "epoch": 162.28318584070797,
      "grad_norm": 9.282231985707767e-06,
      "learning_rate": 4.3689012370537904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5257112,
      "step": 9250
    },
    {
      "epoch": 162.3716814159292,
      "grad_norm": 2.0072716324648354e-06,
      "learning_rate": 4.368249022300164e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5259864,
      "step": 9255
    },
    {
      "epoch": 162.46017699115043,
      "grad_norm": 9.20574620977277e-06,
      "learning_rate": 4.367596519439059e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5263208,
      "step": 9260
    },
    {
      "epoch": 162.5486725663717,
      "grad_norm": 6.810781997046433e-06,
      "learning_rate": 4.366943728571101e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5266040,
      "step": 9265
    },
    {
      "epoch": 162.63716814159292,
      "grad_norm": 1.535957608211902e-06,
      "learning_rate": 4.366290649796959e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5268712,
      "step": 9270
    },
    {
      "epoch": 162.72566371681415,
      "grad_norm": 3.0001183404237963e-06,
      "learning_rate": 4.3656372832173456e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5271544,
      "step": 9275
    },
    {
      "epoch": 162.81415929203538,
      "grad_norm": 1.8787021645039204e-06,
      "learning_rate": 4.364983628933017e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5274616,
      "step": 9280
    },
    {
      "epoch": 162.90265486725664,
      "grad_norm": 1.4984881318014232e-06,
      "learning_rate": 4.364329687044777e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5277656,
      "step": 9285
    },
    {
      "epoch": 162.99115044247787,
      "grad_norm": 2.461054236846394e-06,
      "learning_rate": 4.36367545765347e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5280744,
      "step": 9290
    },
    {
      "epoch": 163.07079646017698,
      "grad_norm": 6.092056082707131e-06,
      "learning_rate": 4.363020940859988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5283384,
      "step": 9295
    },
    {
      "epoch": 163.15929203539824,
      "grad_norm": 3.606962991398177e-06,
      "learning_rate": 4.362366136765263e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5286088,
      "step": 9300
    },
    {
      "epoch": 163.24778761061947,
      "grad_norm": 2.693828719202429e-06,
      "learning_rate": 4.361711045470278e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5289016,
      "step": 9305
    },
    {
      "epoch": 163.3362831858407,
      "grad_norm": 1.844183884713857e-06,
      "learning_rate": 4.3610556670760524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5291928,
      "step": 9310
    },
    {
      "epoch": 163.42477876106196,
      "grad_norm": 1.1106204510724638e-05,
      "learning_rate": 4.360400001683657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5294776,
      "step": 9315
    },
    {
      "epoch": 163.5132743362832,
      "grad_norm": 2.187614654758363e-06,
      "learning_rate": 4.3597440493942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5297224,
      "step": 9320
    },
    {
      "epoch": 163.60176991150442,
      "grad_norm": 2.589324822110939e-06,
      "learning_rate": 4.3590878103088405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5300056,
      "step": 9325
    },
    {
      "epoch": 163.69026548672565,
      "grad_norm": 1.6011680372685078e-06,
      "learning_rate": 4.358431284528779e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5303032,
      "step": 9330
    },
    {
      "epoch": 163.7787610619469,
      "grad_norm": 1.2543734555947594e-05,
      "learning_rate": 4.357774472155257e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5306136,
      "step": 9335
    },
    {
      "epoch": 163.86725663716814,
      "grad_norm": 7.658924005227163e-06,
      "learning_rate": 4.3571173732895664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5309416,
      "step": 9340
    },
    {
      "epoch": 163.95575221238937,
      "grad_norm": 1.9364986656000838e-05,
      "learning_rate": 4.356459988033039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5312040,
      "step": 9345
    },
    {
      "epoch": 164.0353982300885,
      "grad_norm": 1.013328892440768e-05,
      "learning_rate": 4.355802316487051e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5314136,
      "step": 9350
    },
    {
      "epoch": 164.12389380530973,
      "grad_norm": 1.025783831209992e-06,
      "learning_rate": 4.355144358753025e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5317080,
      "step": 9355
    },
    {
      "epoch": 164.21238938053096,
      "grad_norm": 1.3625360679725418e-06,
      "learning_rate": 4.354486114932425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5320120,
      "step": 9360
    },
    {
      "epoch": 164.30088495575222,
      "grad_norm": 4.608997642208124e-06,
      "learning_rate": 4.353827585126762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5323048,
      "step": 9365
    },
    {
      "epoch": 164.38938053097345,
      "grad_norm": 1.436566799384309e-05,
      "learning_rate": 4.353168769437588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5325864,
      "step": 9370
    },
    {
      "epoch": 164.47787610619469,
      "grad_norm": 2.060701945083565e-06,
      "learning_rate": 4.3525096679665014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5328824,
      "step": 9375
    },
    {
      "epoch": 164.56637168141592,
      "grad_norm": 1.5040996004245244e-06,
      "learning_rate": 4.351850280815144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5331480,
      "step": 9380
    },
    {
      "epoch": 164.65486725663717,
      "grad_norm": 9.033035894390196e-06,
      "learning_rate": 4.3511906080852014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5334504,
      "step": 9385
    },
    {
      "epoch": 164.7433628318584,
      "grad_norm": 3.1725426197226625e-06,
      "learning_rate": 4.350530649878404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5337752,
      "step": 9390
    },
    {
      "epoch": 164.83185840707964,
      "grad_norm": 6.400703114195494e-06,
      "learning_rate": 4.3498704062965246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5340600,
      "step": 9395
    },
    {
      "epoch": 164.9203539823009,
      "grad_norm": 2.104101895383792e-06,
      "learning_rate": 4.3492098774413815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5343528,
      "step": 9400
    },
    {
      "epoch": 164.9203539823009,
      "eval_loss": 0.42724984884262085,
      "eval_runtime": 1.0636,
      "eval_samples_per_second": 23.505,
      "eval_steps_per_second": 12.222,
      "num_input_tokens_seen": 5343528,
      "step": 9400
    },
    {
      "epoch": 165.0,
      "grad_norm": 6.102975476096617e-06,
      "learning_rate": 4.3485490634148375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5345792,
      "step": 9405
    },
    {
      "epoch": 165.08849557522123,
      "grad_norm": 8.449666893284302e-06,
      "learning_rate": 4.347887964318797e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5349024,
      "step": 9410
    },
    {
      "epoch": 165.1769911504425,
      "grad_norm": 2.256424977531424e-06,
      "learning_rate": 4.34722658025521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5351648,
      "step": 9415
    },
    {
      "epoch": 165.26548672566372,
      "grad_norm": 1.8084901967085898e-06,
      "learning_rate": 4.346564911326071e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5354752,
      "step": 9420
    },
    {
      "epoch": 165.35398230088495,
      "grad_norm": 8.003285074664745e-06,
      "learning_rate": 4.345902957633418e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5357888,
      "step": 9425
    },
    {
      "epoch": 165.44247787610618,
      "grad_norm": 2.2447097762778867e-06,
      "learning_rate": 4.345240719279331e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5360512,
      "step": 9430
    },
    {
      "epoch": 165.53097345132744,
      "grad_norm": 1.3332212347449968e-06,
      "learning_rate": 4.3445781963659374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5363216,
      "step": 9435
    },
    {
      "epoch": 165.61946902654867,
      "grad_norm": 3.1075792321644258e-06,
      "learning_rate": 4.3439153889954045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5366000,
      "step": 9440
    },
    {
      "epoch": 165.7079646017699,
      "grad_norm": 2.4854812181729358e-06,
      "learning_rate": 4.343252297269946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5368528,
      "step": 9445
    },
    {
      "epoch": 165.79646017699116,
      "grad_norm": 7.499101229768712e-06,
      "learning_rate": 4.342588921291821e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5371136,
      "step": 9450
    },
    {
      "epoch": 165.8849557522124,
      "grad_norm": 1.0934186320810113e-06,
      "learning_rate": 4.341925261163328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5374288,
      "step": 9455
    },
    {
      "epoch": 165.97345132743362,
      "grad_norm": 2.18805939766753e-06,
      "learning_rate": 4.341261316986813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5377472,
      "step": 9460
    },
    {
      "epoch": 166.05309734513276,
      "grad_norm": 2.0948684777977178e-06,
      "learning_rate": 4.340597088864664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5379560,
      "step": 9465
    },
    {
      "epoch": 166.141592920354,
      "grad_norm": 1.6342686421921826e-06,
      "learning_rate": 4.339932576899313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5382488,
      "step": 9470
    },
    {
      "epoch": 166.23008849557522,
      "grad_norm": 1.662821318859642e-06,
      "learning_rate": 4.3392677811932375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5386232,
      "step": 9475
    },
    {
      "epoch": 166.31858407079645,
      "grad_norm": 2.2407484721043147e-06,
      "learning_rate": 4.338602701848956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5389112,
      "step": 9480
    },
    {
      "epoch": 166.4070796460177,
      "grad_norm": 1.6111478089442244e-06,
      "learning_rate": 4.337937338969033e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5391752,
      "step": 9485
    },
    {
      "epoch": 166.49557522123894,
      "grad_norm": 5.053921086073387e-06,
      "learning_rate": 4.337271692656075e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5394056,
      "step": 9490
    },
    {
      "epoch": 166.58407079646017,
      "grad_norm": 2.6233831249555806e-06,
      "learning_rate": 4.336605763012733e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5396792,
      "step": 9495
    },
    {
      "epoch": 166.67256637168143,
      "grad_norm": 7.06834225638886e-06,
      "learning_rate": 4.3359395501417026e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5399960,
      "step": 9500
    },
    {
      "epoch": 166.76106194690266,
      "grad_norm": 3.1218935419019544e-06,
      "learning_rate": 4.335273054145722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5403048,
      "step": 9505
    },
    {
      "epoch": 166.8495575221239,
      "grad_norm": 2.095124045808916e-06,
      "learning_rate": 4.334606275127572e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5406024,
      "step": 9510
    },
    {
      "epoch": 166.93805309734512,
      "grad_norm": 2.3311963559535798e-06,
      "learning_rate": 4.33393921319008e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5408744,
      "step": 9515
    },
    {
      "epoch": 167.01769911504425,
      "grad_norm": 2.3252684968610993e-06,
      "learning_rate": 4.3332718684361146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5410896,
      "step": 9520
    },
    {
      "epoch": 167.10619469026548,
      "grad_norm": 1.0359402040194254e-05,
      "learning_rate": 4.332604240968588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5413488,
      "step": 9525
    },
    {
      "epoch": 167.1946902654867,
      "grad_norm": 4.854623057326535e-06,
      "learning_rate": 4.331936330890459e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5416224,
      "step": 9530
    },
    {
      "epoch": 167.28318584070797,
      "grad_norm": 1.4969108406148735e-06,
      "learning_rate": 4.331268138304725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5418960,
      "step": 9535
    },
    {
      "epoch": 167.3716814159292,
      "grad_norm": 3.0373912522918545e-06,
      "learning_rate": 4.330599663314431e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5421488,
      "step": 9540
    },
    {
      "epoch": 167.46017699115043,
      "grad_norm": 5.1476813496265095e-06,
      "learning_rate": 4.329930906022665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5424528,
      "step": 9545
    },
    {
      "epoch": 167.5486725663717,
      "grad_norm": 6.425644642149564e-06,
      "learning_rate": 4.3292618665325564e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5427264,
      "step": 9550
    },
    {
      "epoch": 167.63716814159292,
      "grad_norm": 4.368782356323209e-06,
      "learning_rate": 4.3285925449472796e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5430304,
      "step": 9555
    },
    {
      "epoch": 167.72566371681415,
      "grad_norm": 2.6648604034562595e-06,
      "learning_rate": 4.327922941370054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5433264,
      "step": 9560
    },
    {
      "epoch": 167.81415929203538,
      "grad_norm": 2.3774778128426988e-06,
      "learning_rate": 4.3272530559041384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5436464,
      "step": 9565
    },
    {
      "epoch": 167.90265486725664,
      "grad_norm": 3.3400692700524814e-06,
      "learning_rate": 4.32658288865284e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5439456,
      "step": 9570
    },
    {
      "epoch": 167.99115044247787,
      "grad_norm": 1.795764092094032e-06,
      "learning_rate": 4.325912439719505e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5442544,
      "step": 9575
    },
    {
      "epoch": 168.07079646017698,
      "grad_norm": 1.3707443713428802e-06,
      "learning_rate": 4.3252417092075266e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5444544,
      "step": 9580
    },
    {
      "epoch": 168.15929203539824,
      "grad_norm": 1.863778379629366e-06,
      "learning_rate": 4.3245706972203385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5447296,
      "step": 9585
    },
    {
      "epoch": 168.24778761061947,
      "grad_norm": 1.2874247659055982e-06,
      "learning_rate": 4.323899403861421e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5449888,
      "step": 9590
    },
    {
      "epoch": 168.3362831858407,
      "grad_norm": 1.3406134939941694e-06,
      "learning_rate": 4.3232278292342935e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5452416,
      "step": 9595
    },
    {
      "epoch": 168.42477876106196,
      "grad_norm": 3.4219137887703255e-05,
      "learning_rate": 4.322555973442524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5455520,
      "step": 9600
    },
    {
      "epoch": 168.42477876106196,
      "eval_loss": 0.42814141511917114,
      "eval_runtime": 1.0569,
      "eval_samples_per_second": 23.653,
      "eval_steps_per_second": 12.3,
      "num_input_tokens_seen": 5455520,
      "step": 9600
    },
    {
      "epoch": 168.5132743362832,
      "grad_norm": 1.238009076587332e-06,
      "learning_rate": 4.3218838365897184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5457984,
      "step": 9605
    },
    {
      "epoch": 168.60176991150442,
      "grad_norm": 1.0020120498666074e-05,
      "learning_rate": 4.3212114187795306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5460992,
      "step": 9610
    },
    {
      "epoch": 168.69026548672565,
      "grad_norm": 1.7638693634580704e-06,
      "learning_rate": 4.320538720115656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5464464,
      "step": 9615
    },
    {
      "epoch": 168.7787610619469,
      "grad_norm": 1.0705817885536817e-06,
      "learning_rate": 4.319865740701831e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5467280,
      "step": 9620
    },
    {
      "epoch": 168.86725663716814,
      "grad_norm": 1.7930743752003764e-06,
      "learning_rate": 4.3191924806418396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5470064,
      "step": 9625
    },
    {
      "epoch": 168.95575221238937,
      "grad_norm": 6.493344699265435e-06,
      "learning_rate": 4.318518940039507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5473632,
      "step": 9630
    },
    {
      "epoch": 169.0353982300885,
      "grad_norm": 2.222862804046599e-06,
      "learning_rate": 4.3178451189987e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5476040,
      "step": 9635
    },
    {
      "epoch": 169.12389380530973,
      "grad_norm": 1.926948925756733e-06,
      "learning_rate": 4.3171710176233315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5479496,
      "step": 9640
    },
    {
      "epoch": 169.21238938053096,
      "grad_norm": 2.4305711576744216e-06,
      "learning_rate": 4.316496636017355e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5482344,
      "step": 9645
    },
    {
      "epoch": 169.30088495575222,
      "grad_norm": 1.7073750768759055e-06,
      "learning_rate": 4.315821974284771e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5485160,
      "step": 9650
    },
    {
      "epoch": 169.38938053097345,
      "grad_norm": 1.8549501419329317e-06,
      "learning_rate": 4.315147032529619e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5488504,
      "step": 9655
    },
    {
      "epoch": 169.47787610619469,
      "grad_norm": 2.479001295796479e-06,
      "learning_rate": 4.3144718108559845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5491368,
      "step": 9660
    },
    {
      "epoch": 169.56637168141592,
      "grad_norm": 1.9162698663421907e-06,
      "learning_rate": 4.3137963093679945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5494088,
      "step": 9665
    },
    {
      "epoch": 169.65486725663717,
      "grad_norm": 1.1088643532275455e-06,
      "learning_rate": 4.31312052816982e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5496600,
      "step": 9670
    },
    {
      "epoch": 169.7433628318584,
      "grad_norm": 1.4562145224772394e-05,
      "learning_rate": 4.312444467365675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5499688,
      "step": 9675
    },
    {
      "epoch": 169.83185840707964,
      "grad_norm": 5.4055794862506445e-06,
      "learning_rate": 4.311768127059816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5502408,
      "step": 9680
    },
    {
      "epoch": 169.9203539823009,
      "grad_norm": 6.0948468671995215e-06,
      "learning_rate": 4.3110915073565444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5504648,
      "step": 9685
    },
    {
      "epoch": 170.0,
      "grad_norm": 3.1334516279457603e-06,
      "learning_rate": 4.310414608360203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5507520,
      "step": 9690
    },
    {
      "epoch": 170.08849557522123,
      "grad_norm": 1.9193066691514105e-06,
      "learning_rate": 4.309737430175177e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5510368,
      "step": 9695
    },
    {
      "epoch": 170.1769911504425,
      "grad_norm": 6.233439307834487e-06,
      "learning_rate": 4.309059972905897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5513312,
      "step": 9700
    },
    {
      "epoch": 170.26548672566372,
      "grad_norm": 1.7346100094073336e-06,
      "learning_rate": 4.308382236656836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5516144,
      "step": 9705
    },
    {
      "epoch": 170.35398230088495,
      "grad_norm": 3.1134159144130535e-06,
      "learning_rate": 4.307704221532507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5519584,
      "step": 9710
    },
    {
      "epoch": 170.44247787610618,
      "grad_norm": 1.3131419791534427e-06,
      "learning_rate": 4.307025927637471e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5522768,
      "step": 9715
    },
    {
      "epoch": 170.53097345132744,
      "grad_norm": 2.3142079044191632e-06,
      "learning_rate": 4.306347355076328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5525888,
      "step": 9720
    },
    {
      "epoch": 170.61946902654867,
      "grad_norm": 2.242807795482804e-06,
      "learning_rate": 4.305668503953724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5528976,
      "step": 9725
    },
    {
      "epoch": 170.7079646017699,
      "grad_norm": 3.7331858493416803e-06,
      "learning_rate": 4.3049893743743436e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5531600,
      "step": 9730
    },
    {
      "epoch": 170.79646017699116,
      "grad_norm": 1.0599651432130486e-05,
      "learning_rate": 4.304309966442919e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5534096,
      "step": 9735
    },
    {
      "epoch": 170.8849557522124,
      "grad_norm": 2.0006177692266647e-06,
      "learning_rate": 4.303630280264224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5537040,
      "step": 9740
    },
    {
      "epoch": 170.97345132743362,
      "grad_norm": 9.422651601198595e-06,
      "learning_rate": 4.302950315943074e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5539408,
      "step": 9745
    },
    {
      "epoch": 171.05309734513276,
      "grad_norm": 1.5622840692230966e-06,
      "learning_rate": 4.3022700735843275e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5541880,
      "step": 9750
    },
    {
      "epoch": 171.141592920354,
      "grad_norm": 1.314371502303402e-06,
      "learning_rate": 4.301589553292887e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5544680,
      "step": 9755
    },
    {
      "epoch": 171.23008849557522,
      "grad_norm": 1.3421572475635912e-06,
      "learning_rate": 4.300908755173697e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5547368,
      "step": 9760
    },
    {
      "epoch": 171.31858407079645,
      "grad_norm": 1.9385940959182335e-06,
      "learning_rate": 4.300227679331745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5550840,
      "step": 9765
    },
    {
      "epoch": 171.4070796460177,
      "grad_norm": 1.6863286873558536e-06,
      "learning_rate": 4.299546325872063e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5553928,
      "step": 9770
    },
    {
      "epoch": 171.49557522123894,
      "grad_norm": 2.8059628220944433e-06,
      "learning_rate": 4.2988646948997225e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5557528,
      "step": 9775
    },
    {
      "epoch": 171.58407079646017,
      "grad_norm": 5.738619165640557e-06,
      "learning_rate": 4.29818278651984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5560232,
      "step": 9780
    },
    {
      "epoch": 171.67256637168143,
      "grad_norm": 1.0701314749894664e-06,
      "learning_rate": 4.297500600837574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5563320,
      "step": 9785
    },
    {
      "epoch": 171.76106194690266,
      "grad_norm": 2.3565041828987887e-06,
      "learning_rate": 4.2968181379581276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5566072,
      "step": 9790
    },
    {
      "epoch": 171.8495575221239,
      "grad_norm": 2.74193234872655e-06,
      "learning_rate": 4.296135397986743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5568776,
      "step": 9795
    },
    {
      "epoch": 171.93805309734512,
      "grad_norm": 1.9106551008007955e-06,
      "learning_rate": 4.295452381028709e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5571144,
      "step": 9800
    },
    {
      "epoch": 171.93805309734512,
      "eval_loss": 0.42372962832450867,
      "eval_runtime": 1.0599,
      "eval_samples_per_second": 23.587,
      "eval_steps_per_second": 12.265,
      "num_input_tokens_seen": 5571144,
      "step": 9800
    },
    {
      "epoch": 172.01769911504425,
      "grad_norm": 7.1668582677375525e-06,
      "learning_rate": 4.294769087189354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5573536,
      "step": 9805
    },
    {
      "epoch": 172.10619469026548,
      "grad_norm": 2.147648046957329e-06,
      "learning_rate": 4.294085516574052e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5577200,
      "step": 9810
    },
    {
      "epoch": 172.1946902654867,
      "grad_norm": 1.7529790738990414e-06,
      "learning_rate": 4.2934016692882176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5579760,
      "step": 9815
    },
    {
      "epoch": 172.28318584070797,
      "grad_norm": 6.461349130404415e-06,
      "learning_rate": 4.292717545437308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5582272,
      "step": 9820
    },
    {
      "epoch": 172.3716814159292,
      "grad_norm": 2.4918797407735838e-06,
      "learning_rate": 4.292033145126825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5584960,
      "step": 9825
    },
    {
      "epoch": 172.46017699115043,
      "grad_norm": 1.4384090718522202e-06,
      "learning_rate": 4.29134846846231e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5587952,
      "step": 9830
    },
    {
      "epoch": 172.5486725663717,
      "grad_norm": 1.736196395540901e-06,
      "learning_rate": 4.29066351554935e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5590816,
      "step": 9835
    },
    {
      "epoch": 172.63716814159292,
      "grad_norm": 1.6788688981250743e-06,
      "learning_rate": 4.289978286493574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5593600,
      "step": 9840
    },
    {
      "epoch": 172.72566371681415,
      "grad_norm": 2.1183757326070918e-06,
      "learning_rate": 4.28929278140065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5596416,
      "step": 9845
    },
    {
      "epoch": 172.81415929203538,
      "grad_norm": 5.26711073689512e-06,
      "learning_rate": 4.288607000376295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5599328,
      "step": 9850
    },
    {
      "epoch": 172.90265486725664,
      "grad_norm": 1.6527450270586996e-06,
      "learning_rate": 4.2879209435262624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5602256,
      "step": 9855
    },
    {
      "epoch": 172.99115044247787,
      "grad_norm": 4.424077360454248e-06,
      "learning_rate": 4.287234610956353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5605344,
      "step": 9860
    },
    {
      "epoch": 173.07079646017698,
      "grad_norm": 1.4931038094800897e-06,
      "learning_rate": 4.2865480027724056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5608000,
      "step": 9865
    },
    {
      "epoch": 173.15929203539824,
      "grad_norm": 2.1814862520841416e-06,
      "learning_rate": 4.285861119080306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5610912,
      "step": 9870
    },
    {
      "epoch": 173.24778761061947,
      "grad_norm": 3.8238194974837825e-06,
      "learning_rate": 4.2851739599859784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5613472,
      "step": 9875
    },
    {
      "epoch": 173.3362831858407,
      "grad_norm": 2.260689598188037e-06,
      "learning_rate": 4.2844865255953934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5616848,
      "step": 9880
    },
    {
      "epoch": 173.42477876106196,
      "grad_norm": 9.078176844923291e-06,
      "learning_rate": 4.2837988160145605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5619792,
      "step": 9885
    },
    {
      "epoch": 173.5132743362832,
      "grad_norm": 2.689131179067772e-06,
      "learning_rate": 4.2831108313495336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5622464,
      "step": 9890
    },
    {
      "epoch": 173.60176991150442,
      "grad_norm": 4.5663655328098685e-06,
      "learning_rate": 4.282422571706408e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5625072,
      "step": 9895
    },
    {
      "epoch": 173.69026548672565,
      "grad_norm": 1.0075514182972256e-06,
      "learning_rate": 4.281734037191323e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5627920,
      "step": 9900
    },
    {
      "epoch": 173.7787610619469,
      "grad_norm": 1.8198689986093086e-06,
      "learning_rate": 4.281045227910459e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5631152,
      "step": 9905
    },
    {
      "epoch": 173.86725663716814,
      "grad_norm": 2.4531739200028824e-06,
      "learning_rate": 4.280356143970038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5633968,
      "step": 9910
    },
    {
      "epoch": 173.95575221238937,
      "grad_norm": 2.5493231987638865e-06,
      "learning_rate": 4.279666785476327e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5636880,
      "step": 9915
    },
    {
      "epoch": 174.0353982300885,
      "grad_norm": 2.120228145940928e-06,
      "learning_rate": 4.2789771525356325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5638976,
      "step": 9920
    },
    {
      "epoch": 174.12389380530973,
      "grad_norm": 1.7639662246438093e-06,
      "learning_rate": 4.2782872452543056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5641904,
      "step": 9925
    },
    {
      "epoch": 174.21238938053096,
      "grad_norm": 1.6036204897318385e-06,
      "learning_rate": 4.2775970637387376e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5644704,
      "step": 9930
    },
    {
      "epoch": 174.30088495575222,
      "grad_norm": 5.487676389748231e-06,
      "learning_rate": 4.276906608095363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5647760,
      "step": 9935
    },
    {
      "epoch": 174.38938053097345,
      "grad_norm": 9.114671229326632e-06,
      "learning_rate": 4.276215878430661e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5650896,
      "step": 9940
    },
    {
      "epoch": 174.47787610619469,
      "grad_norm": 2.2919764433027012e-06,
      "learning_rate": 4.275524874851149e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5653824,
      "step": 9945
    },
    {
      "epoch": 174.56637168141592,
      "grad_norm": 2.7282906103209825e-06,
      "learning_rate": 4.274833597463388e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5656496,
      "step": 9950
    },
    {
      "epoch": 174.65486725663717,
      "grad_norm": 4.5575216063298285e-06,
      "learning_rate": 4.2741420463739824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5659728,
      "step": 9955
    },
    {
      "epoch": 174.7433628318584,
      "grad_norm": 1.4826016467850422e-06,
      "learning_rate": 4.273450221689578e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5662688,
      "step": 9960
    },
    {
      "epoch": 174.83185840707964,
      "grad_norm": 2.919992084571277e-06,
      "learning_rate": 4.272758123516863e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5665344,
      "step": 9965
    },
    {
      "epoch": 174.9203539823009,
      "grad_norm": 1.6801461697468767e-06,
      "learning_rate": 4.272065751962567e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5668368,
      "step": 9970
    },
    {
      "epoch": 175.0,
      "grad_norm": 2.8566742003022227e-06,
      "learning_rate": 4.271373107133464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5670672,
      "step": 9975
    },
    {
      "epoch": 175.08849557522123,
      "grad_norm": 4.2588167161738966e-06,
      "learning_rate": 4.270680189136366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5673520,
      "step": 9980
    },
    {
      "epoch": 175.1769911504425,
      "grad_norm": 2.820751660692622e-06,
      "learning_rate": 4.269986998078132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5676432,
      "step": 9985
    },
    {
      "epoch": 175.26548672566372,
      "grad_norm": 1.9972662812506314e-06,
      "learning_rate": 4.2692935340656595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5679040,
      "step": 9990
    },
    {
      "epoch": 175.35398230088495,
      "grad_norm": 4.711402652901597e-06,
      "learning_rate": 4.26859979720589e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5682224,
      "step": 9995
    },
    {
      "epoch": 175.44247787610618,
      "grad_norm": 2.0511595266725635e-06,
      "learning_rate": 4.267905787605806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5684752,
      "step": 10000
    },
    {
      "epoch": 175.44247787610618,
      "eval_loss": 0.44014638662338257,
      "eval_runtime": 1.0647,
      "eval_samples_per_second": 23.482,
      "eval_steps_per_second": 12.211,
      "num_input_tokens_seen": 5684752,
      "step": 10000
    },
    {
      "epoch": 175.53097345132744,
      "grad_norm": 3.939776433981024e-05,
      "learning_rate": 4.267211505372433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5687408,
      "step": 10005
    },
    {
      "epoch": 175.61946902654867,
      "grad_norm": 5.557350050366949e-06,
      "learning_rate": 4.266516950612837e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5690576,
      "step": 10010
    },
    {
      "epoch": 175.7079646017699,
      "grad_norm": 4.48536002295441e-06,
      "learning_rate": 4.265822123434128e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5693824,
      "step": 10015
    },
    {
      "epoch": 175.79646017699116,
      "grad_norm": 2.1234525320323883e-06,
      "learning_rate": 4.265127023943457e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5696208,
      "step": 10020
    },
    {
      "epoch": 175.8849557522124,
      "grad_norm": 1.8163577806262765e-06,
      "learning_rate": 4.2644316522480176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5699024,
      "step": 10025
    },
    {
      "epoch": 175.97345132743362,
      "grad_norm": 5.643008080369327e-06,
      "learning_rate": 4.263736008455044e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5702400,
      "step": 10030
    },
    {
      "epoch": 176.05309734513276,
      "grad_norm": 1.6696578768460313e-06,
      "learning_rate": 4.2630400926718125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5705040,
      "step": 10035
    },
    {
      "epoch": 176.141592920354,
      "grad_norm": 1.6570719481023843e-06,
      "learning_rate": 4.262343905005644e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5707824,
      "step": 10040
    },
    {
      "epoch": 176.23008849557522,
      "grad_norm": 3.306720145701547e-06,
      "learning_rate": 4.261647445563897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5710928,
      "step": 10045
    },
    {
      "epoch": 176.31858407079645,
      "grad_norm": 2.525854597479338e-06,
      "learning_rate": 4.260950714453976e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5713984,
      "step": 10050
    },
    {
      "epoch": 176.4070796460177,
      "grad_norm": 7.129699952201918e-06,
      "learning_rate": 4.2602537117833266e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5716688,
      "step": 10055
    },
    {
      "epoch": 176.49557522123894,
      "grad_norm": 4.589416676026303e-06,
      "learning_rate": 4.259556437659433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5719504,
      "step": 10060
    },
    {
      "epoch": 176.58407079646017,
      "grad_norm": 6.443097845476586e-06,
      "learning_rate": 4.258858892189825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5723040,
      "step": 10065
    },
    {
      "epoch": 176.67256637168143,
      "grad_norm": 1.578706132931984e-06,
      "learning_rate": 4.2581610754820725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5726336,
      "step": 10070
    },
    {
      "epoch": 176.76106194690266,
      "grad_norm": 1.1655288290057797e-05,
      "learning_rate": 4.2574629876437876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5728752,
      "step": 10075
    },
    {
      "epoch": 176.8495575221239,
      "grad_norm": 1.654663492445252e-06,
      "learning_rate": 4.256764628782625e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5731760,
      "step": 10080
    },
    {
      "epoch": 176.93805309734512,
      "grad_norm": 2.2151623397803633e-06,
      "learning_rate": 4.256065999006279e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5734512,
      "step": 10085
    },
    {
      "epoch": 177.01769911504425,
      "grad_norm": 1.3690262221643934e-06,
      "learning_rate": 4.2553670984224885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5736544,
      "step": 10090
    },
    {
      "epoch": 177.10619469026548,
      "grad_norm": 7.897138857515529e-06,
      "learning_rate": 4.254667927139032e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5739264,
      "step": 10095
    },
    {
      "epoch": 177.1946902654867,
      "grad_norm": 1.1835558325401507e-05,
      "learning_rate": 4.2539684852637295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5741616,
      "step": 10100
    },
    {
      "epoch": 177.28318584070797,
      "grad_norm": 2.0794714146177284e-06,
      "learning_rate": 4.253268772904446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5744560,
      "step": 10105
    },
    {
      "epoch": 177.3716814159292,
      "grad_norm": 4.453429482964566e-06,
      "learning_rate": 4.252568790169085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5747488,
      "step": 10110
    },
    {
      "epoch": 177.46017699115043,
      "grad_norm": 1.4717711565026548e-06,
      "learning_rate": 4.251868537165592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5750416,
      "step": 10115
    },
    {
      "epoch": 177.5486725663717,
      "grad_norm": 1.5402054032165324e-06,
      "learning_rate": 4.251168014001955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5753440,
      "step": 10120
    },
    {
      "epoch": 177.63716814159292,
      "grad_norm": 3.5575310448621167e-06,
      "learning_rate": 4.250467220786204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5756464,
      "step": 10125
    },
    {
      "epoch": 177.72566371681415,
      "grad_norm": 1.4183509620124823e-06,
      "learning_rate": 4.249766157626409e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5759648,
      "step": 10130
    },
    {
      "epoch": 177.81415929203538,
      "grad_norm": 4.8772139962238725e-06,
      "learning_rate": 4.249064824630684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5762496,
      "step": 10135
    },
    {
      "epoch": 177.90265486725664,
      "grad_norm": 2.6304792299924884e-06,
      "learning_rate": 4.248363221907183e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5765104,
      "step": 10140
    },
    {
      "epoch": 177.99115044247787,
      "grad_norm": 4.033987806906225e-06,
      "learning_rate": 4.2476613495641026e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5768032,
      "step": 10145
    },
    {
      "epoch": 178.07079646017698,
      "grad_norm": 1.80823121809226e-06,
      "learning_rate": 4.246959207709679e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5770576,
      "step": 10150
    },
    {
      "epoch": 178.15929203539824,
      "grad_norm": 2.0053496427863138e-06,
      "learning_rate": 4.246256796452192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5773232,
      "step": 10155
    },
    {
      "epoch": 178.24778761061947,
      "grad_norm": 9.842343615673599e-07,
      "learning_rate": 4.245554115899962e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5776272,
      "step": 10160
    },
    {
      "epoch": 178.3362831858407,
      "grad_norm": 1.4613796111007105e-06,
      "learning_rate": 4.2448511661613514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5779056,
      "step": 10165
    },
    {
      "epoch": 178.42477876106196,
      "grad_norm": 9.682889867690392e-06,
      "learning_rate": 4.2441479473447635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5782176,
      "step": 10170
    },
    {
      "epoch": 178.5132743362832,
      "grad_norm": 3.431733148318017e-06,
      "learning_rate": 4.243444459558644e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5784656,
      "step": 10175
    },
    {
      "epoch": 178.60176991150442,
      "grad_norm": 2.0742011201946298e-06,
      "learning_rate": 4.24274070291148e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5787488,
      "step": 10180
    },
    {
      "epoch": 178.69026548672565,
      "grad_norm": 1.9441820313659264e-06,
      "learning_rate": 4.242036677511798e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5790464,
      "step": 10185
    },
    {
      "epoch": 178.7787610619469,
      "grad_norm": 2.193458158217254e-06,
      "learning_rate": 4.241332383468169e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5793568,
      "step": 10190
    },
    {
      "epoch": 178.86725663716814,
      "grad_norm": 2.258972017443739e-06,
      "learning_rate": 4.2406278208892034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5796240,
      "step": 10195
    },
    {
      "epoch": 178.95575221238937,
      "grad_norm": 4.051086762046907e-06,
      "learning_rate": 4.2399229898835536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5799088,
      "step": 10200
    },
    {
      "epoch": 178.95575221238937,
      "eval_loss": 0.42910388112068176,
      "eval_runtime": 1.057,
      "eval_samples_per_second": 23.652,
      "eval_steps_per_second": 12.299,
      "num_input_tokens_seen": 5799088,
      "step": 10200
    },
    {
      "epoch": 179.0353982300885,
      "grad_norm": 1.0431304872327019e-05,
      "learning_rate": 4.239217890559914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5801504,
      "step": 10205
    },
    {
      "epoch": 179.12389380530973,
      "grad_norm": 2.9980960789544042e-06,
      "learning_rate": 4.238512523027019e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5804720,
      "step": 10210
    },
    {
      "epoch": 179.21238938053096,
      "grad_norm": 5.224533197178971e-06,
      "learning_rate": 4.237806887393645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5807760,
      "step": 10215
    },
    {
      "epoch": 179.30088495575222,
      "grad_norm": 1.8847190403903369e-06,
      "learning_rate": 4.237100983768611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5810400,
      "step": 10220
    },
    {
      "epoch": 179.38938053097345,
      "grad_norm": 7.428548997268081e-06,
      "learning_rate": 4.2363948122607756e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5813040,
      "step": 10225
    },
    {
      "epoch": 179.47787610619469,
      "grad_norm": 1.4784786799282301e-06,
      "learning_rate": 4.235688372979039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5815760,
      "step": 10230
    },
    {
      "epoch": 179.56637168141592,
      "grad_norm": 4.2462925193831325e-06,
      "learning_rate": 4.234981666032343e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5818896,
      "step": 10235
    },
    {
      "epoch": 179.65486725663717,
      "grad_norm": 6.163040779938456e-06,
      "learning_rate": 4.2342746915296704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5821712,
      "step": 10240
    },
    {
      "epoch": 179.7433628318584,
      "grad_norm": 3.6757999168912647e-06,
      "learning_rate": 4.233567449580047e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5824880,
      "step": 10245
    },
    {
      "epoch": 179.83185840707964,
      "grad_norm": 4.778797119797673e-06,
      "learning_rate": 4.232859940292537e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5827120,
      "step": 10250
    },
    {
      "epoch": 179.9203539823009,
      "grad_norm": 2.598864057290484e-06,
      "learning_rate": 4.232152163776248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5829744,
      "step": 10255
    },
    {
      "epoch": 180.0,
      "grad_norm": 1.9495650121825747e-05,
      "learning_rate": 4.231444120140328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5832608,
      "step": 10260
    },
    {
      "epoch": 180.08849557522123,
      "grad_norm": 2.1409932742244564e-06,
      "learning_rate": 4.230735809493967e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5835344,
      "step": 10265
    },
    {
      "epoch": 180.1769911504425,
      "grad_norm": 2.978809106934932e-06,
      "learning_rate": 4.2300272319463926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5838000,
      "step": 10270
    },
    {
      "epoch": 180.26548672566372,
      "grad_norm": 4.3926106627623085e-06,
      "learning_rate": 4.2293183876068786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5841456,
      "step": 10275
    },
    {
      "epoch": 180.35398230088495,
      "grad_norm": 2.9338775675569195e-06,
      "learning_rate": 4.228609276584737e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5844736,
      "step": 10280
    },
    {
      "epoch": 180.44247787610618,
      "grad_norm": 2.2828298824606463e-06,
      "learning_rate": 4.227899898989323e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5847664,
      "step": 10285
    },
    {
      "epoch": 180.53097345132744,
      "grad_norm": 2.9529321636800887e-06,
      "learning_rate": 4.2271902549300293e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5850640,
      "step": 10290
    },
    {
      "epoch": 180.61946902654867,
      "grad_norm": 2.4259861675091088e-06,
      "learning_rate": 4.226480344516294e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5853648,
      "step": 10295
    },
    {
      "epoch": 180.7079646017699,
      "grad_norm": 1.0994024250976508e-06,
      "learning_rate": 4.2257701678575925e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5856320,
      "step": 10300
    },
    {
      "epoch": 180.79646017699116,
      "grad_norm": 5.339685230865143e-06,
      "learning_rate": 4.225059725063444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5859488,
      "step": 10305
    },
    {
      "epoch": 180.8849557522124,
      "grad_norm": 1.1823436807389953e-06,
      "learning_rate": 4.2243490162434074e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5862000,
      "step": 10310
    },
    {
      "epoch": 180.97345132743362,
      "grad_norm": 4.232905666867737e-06,
      "learning_rate": 4.223638041507083e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5864784,
      "step": 10315
    },
    {
      "epoch": 181.05309734513276,
      "grad_norm": 1.266256390408671e-06,
      "learning_rate": 4.2229268009641124e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5867224,
      "step": 10320
    },
    {
      "epoch": 181.141592920354,
      "grad_norm": 1.7954732811631402e-06,
      "learning_rate": 4.222215294724177e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5870120,
      "step": 10325
    },
    {
      "epoch": 181.23008849557522,
      "grad_norm": 1.4795912193221739e-06,
      "learning_rate": 4.2215035228970005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5873176,
      "step": 10330
    },
    {
      "epoch": 181.31858407079645,
      "grad_norm": 1.907670366563252e-06,
      "learning_rate": 4.2207914855923464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5875640,
      "step": 10335
    },
    {
      "epoch": 181.4070796460177,
      "grad_norm": 1.350651700704475e-06,
      "learning_rate": 4.220079182920021e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5878280,
      "step": 10340
    },
    {
      "epoch": 181.49557522123894,
      "grad_norm": 3.680364216052112e-06,
      "learning_rate": 4.2193666149898705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5881016,
      "step": 10345
    },
    {
      "epoch": 181.58407079646017,
      "grad_norm": 1.4970255506341346e-06,
      "learning_rate": 4.21865378191178e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5884168,
      "step": 10350
    },
    {
      "epoch": 181.67256637168143,
      "grad_norm": 1.1562982535906485e-06,
      "learning_rate": 4.217940683795678e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5886712,
      "step": 10355
    },
    {
      "epoch": 181.76106194690266,
      "grad_norm": 3.856760031339945e-06,
      "learning_rate": 4.217227320751534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5889544,
      "step": 10360
    },
    {
      "epoch": 181.8495575221239,
      "grad_norm": 4.517028628470143e-06,
      "learning_rate": 4.216513692889358e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5892392,
      "step": 10365
    },
    {
      "epoch": 181.93805309734512,
      "grad_norm": 2.224741365353111e-06,
      "learning_rate": 4.215799800319199e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5895608,
      "step": 10370
    },
    {
      "epoch": 182.01769911504425,
      "grad_norm": 2.9211005312390625e-06,
      "learning_rate": 4.2150856431511485e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5898128,
      "step": 10375
    },
    {
      "epoch": 182.10619469026548,
      "grad_norm": 2.621075509523507e-06,
      "learning_rate": 4.214371221495339e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5900848,
      "step": 10380
    },
    {
      "epoch": 182.1946902654867,
      "grad_norm": 1.9624969809228787e-06,
      "learning_rate": 4.213656535461942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5903488,
      "step": 10385
    },
    {
      "epoch": 182.28318584070797,
      "grad_norm": 4.7220682972692885e-06,
      "learning_rate": 4.2129415851611734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5906112,
      "step": 10390
    },
    {
      "epoch": 182.3716814159292,
      "grad_norm": 1.140801850851858e-05,
      "learning_rate": 4.2122263707032855e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5909056,
      "step": 10395
    },
    {
      "epoch": 182.46017699115043,
      "grad_norm": 6.1185623962956015e-06,
      "learning_rate": 4.211510892198574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5911888,
      "step": 10400
    },
    {
      "epoch": 182.46017699115043,
      "eval_loss": 0.4353715777397156,
      "eval_runtime": 1.0589,
      "eval_samples_per_second": 23.61,
      "eval_steps_per_second": 12.277,
      "num_input_tokens_seen": 5911888,
      "step": 10400
    },
    {
      "epoch": 182.5486725663717,
      "grad_norm": 2.0242389382474357e-06,
      "learning_rate": 4.210795149757375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5914992,
      "step": 10405
    },
    {
      "epoch": 182.63716814159292,
      "grad_norm": 2.4022829165915027e-06,
      "learning_rate": 4.210079143490065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5918080,
      "step": 10410
    },
    {
      "epoch": 182.72566371681415,
      "grad_norm": 3.2249567993858363e-06,
      "learning_rate": 4.2093628735070604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5920704,
      "step": 10415
    },
    {
      "epoch": 182.81415929203538,
      "grad_norm": 6.660524377366528e-06,
      "learning_rate": 4.208646339918819e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5923680,
      "step": 10420
    },
    {
      "epoch": 182.90265486725664,
      "grad_norm": 1.8846207012757077e-06,
      "learning_rate": 4.2079295428358414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5926752,
      "step": 10425
    },
    {
      "epoch": 182.99115044247787,
      "grad_norm": 1.9147234979755012e-06,
      "learning_rate": 4.207212482368664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5929520,
      "step": 10430
    },
    {
      "epoch": 183.07079646017698,
      "grad_norm": 1.9381045603950042e-06,
      "learning_rate": 4.206495158627867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5932064,
      "step": 10435
    },
    {
      "epoch": 183.15929203539824,
      "grad_norm": 1.835828015828156e-06,
      "learning_rate": 4.205777571724073e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5935312,
      "step": 10440
    },
    {
      "epoch": 183.24778761061947,
      "grad_norm": 2.7299963676341577e-06,
      "learning_rate": 4.20505972176794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5938096,
      "step": 10445
    },
    {
      "epoch": 183.3362831858407,
      "grad_norm": 4.679648554883897e-06,
      "learning_rate": 4.204341608870171e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5941376,
      "step": 10450
    },
    {
      "epoch": 183.42477876106196,
      "grad_norm": 2.05777269002283e-06,
      "learning_rate": 4.203623233141508e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5944368,
      "step": 10455
    },
    {
      "epoch": 183.5132743362832,
      "grad_norm": 1.626346715966065e-06,
      "learning_rate": 4.2029045946927334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5946816,
      "step": 10460
    },
    {
      "epoch": 183.60176991150442,
      "grad_norm": 1.883194954643841e-06,
      "learning_rate": 4.20218569363467e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5949456,
      "step": 10465
    },
    {
      "epoch": 183.69026548672565,
      "grad_norm": 1.5963879604896647e-06,
      "learning_rate": 4.2014665300781834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5952544,
      "step": 10470
    },
    {
      "epoch": 183.7787610619469,
      "grad_norm": 5.577493539021816e-06,
      "learning_rate": 4.200747104134174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5955488,
      "step": 10475
    },
    {
      "epoch": 183.86725663716814,
      "grad_norm": 2.7586072519625304e-06,
      "learning_rate": 4.200027415913588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5958112,
      "step": 10480
    },
    {
      "epoch": 183.95575221238937,
      "grad_norm": 2.5222286694770446e-06,
      "learning_rate": 4.1993074655274126e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5960688,
      "step": 10485
    },
    {
      "epoch": 184.0353982300885,
      "grad_norm": 1.6342951312253717e-06,
      "learning_rate": 4.198587253086669e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5963120,
      "step": 10490
    },
    {
      "epoch": 184.12389380530973,
      "grad_norm": 8.371893272851594e-06,
      "learning_rate": 4.197866778702426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5965872,
      "step": 10495
    },
    {
      "epoch": 184.21238938053096,
      "grad_norm": 1.4581586356143816e-06,
      "learning_rate": 4.197146042485789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5968512,
      "step": 10500
    },
    {
      "epoch": 184.30088495575222,
      "grad_norm": 5.986219548503868e-06,
      "learning_rate": 4.1964250445479046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5972032,
      "step": 10505
    },
    {
      "epoch": 184.38938053097345,
      "grad_norm": 2.4807347926980583e-06,
      "learning_rate": 4.19570378499996e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5975248,
      "step": 10510
    },
    {
      "epoch": 184.47787610619469,
      "grad_norm": 2.5953404474421404e-06,
      "learning_rate": 4.194982263953182e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5977968,
      "step": 10515
    },
    {
      "epoch": 184.56637168141592,
      "grad_norm": 1.9423614503466524e-06,
      "learning_rate": 4.194260481518838e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5980656,
      "step": 10520
    },
    {
      "epoch": 184.65486725663717,
      "grad_norm": 1.4793336049478967e-06,
      "learning_rate": 4.1935384378082366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5983440,
      "step": 10525
    },
    {
      "epoch": 184.7433628318584,
      "grad_norm": 1.7568140719959047e-06,
      "learning_rate": 4.1928161329327267e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5986064,
      "step": 10530
    },
    {
      "epoch": 184.83185840707964,
      "grad_norm": 1.4382113704414223e-06,
      "learning_rate": 4.1920935670036945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5988640,
      "step": 10535
    },
    {
      "epoch": 184.9203539823009,
      "grad_norm": 6.351531283144141e-06,
      "learning_rate": 4.1913707401325705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5991168,
      "step": 10540
    },
    {
      "epoch": 185.0,
      "grad_norm": 2.9414129585347837e-06,
      "learning_rate": 4.1906476524308235e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5993944,
      "step": 10545
    },
    {
      "epoch": 185.08849557522123,
      "grad_norm": 1.3073431546217762e-05,
      "learning_rate": 4.189924304009962e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 5996888,
      "step": 10550
    },
    {
      "epoch": 185.1769911504425,
      "grad_norm": 3.2654049846314592e-06,
      "learning_rate": 4.189200694981537e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6000120,
      "step": 10555
    },
    {
      "epoch": 185.26548672566372,
      "grad_norm": 2.3826999040466035e-06,
      "learning_rate": 4.188476825457136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6003112,
      "step": 10560
    },
    {
      "epoch": 185.35398230088495,
      "grad_norm": 3.592369921534555e-06,
      "learning_rate": 4.18775269554839e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6005704,
      "step": 10565
    },
    {
      "epoch": 185.44247787610618,
      "grad_norm": 7.056572485453216e-06,
      "learning_rate": 4.187028305366969e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6008600,
      "step": 10570
    },
    {
      "epoch": 185.53097345132744,
      "grad_norm": 2.7321841571392724e-06,
      "learning_rate": 4.1863036550245824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6011480,
      "step": 10575
    },
    {
      "epoch": 185.61946902654867,
      "grad_norm": 1.0483406185812782e-05,
      "learning_rate": 4.1855787446329806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6014408,
      "step": 10580
    },
    {
      "epoch": 185.7079646017699,
      "grad_norm": 2.678063992789248e-06,
      "learning_rate": 4.184853574303955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6017176,
      "step": 10585
    },
    {
      "epoch": 185.79646017699116,
      "grad_norm": 1.9376957425265573e-06,
      "learning_rate": 4.184128144149334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6019752,
      "step": 10590
    },
    {
      "epoch": 185.8849557522124,
      "grad_norm": 1.7111480019593728e-06,
      "learning_rate": 4.1834024542809896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6022696,
      "step": 10595
    },
    {
      "epoch": 185.97345132743362,
      "grad_norm": 1.818793975871813e-06,
      "learning_rate": 4.1826765048108315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6025544,
      "step": 10600
    },
    {
      "epoch": 185.97345132743362,
      "eval_loss": 0.44329676032066345,
      "eval_runtime": 1.0585,
      "eval_samples_per_second": 23.618,
      "eval_steps_per_second": 12.281,
      "num_input_tokens_seen": 6025544,
      "step": 10600
    },
    {
      "epoch": 186.05309734513276,
      "grad_norm": 1.2356189245110727e-06,
      "learning_rate": 4.181950295850811e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6027568,
      "step": 10605
    },
    {
      "epoch": 186.141592920354,
      "grad_norm": 1.0117388228536583e-05,
      "learning_rate": 4.181223827512918e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6030480,
      "step": 10610
    },
    {
      "epoch": 186.23008849557522,
      "grad_norm": 4.61838862975128e-06,
      "learning_rate": 4.180497099909183e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6033408,
      "step": 10615
    },
    {
      "epoch": 186.31858407079645,
      "grad_norm": 1.7276835251323064e-06,
      "learning_rate": 4.179770113151677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6036192,
      "step": 10620
    },
    {
      "epoch": 186.4070796460177,
      "grad_norm": 4.119057393836556e-06,
      "learning_rate": 4.179042867352511e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6038880,
      "step": 10625
    },
    {
      "epoch": 186.49557522123894,
      "grad_norm": 4.940196504321648e-06,
      "learning_rate": 4.1783153626238334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6041856,
      "step": 10630
    },
    {
      "epoch": 186.58407079646017,
      "grad_norm": 4.4226226236787625e-06,
      "learning_rate": 4.177587599077836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6044752,
      "step": 10635
    },
    {
      "epoch": 186.67256637168143,
      "grad_norm": 1.7620621974856476e-06,
      "learning_rate": 4.1768595768267494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6047632,
      "step": 10640
    },
    {
      "epoch": 186.76106194690266,
      "grad_norm": 1.339170808023482e-06,
      "learning_rate": 4.176131295982843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6050464,
      "step": 10645
    },
    {
      "epoch": 186.8495575221239,
      "grad_norm": 1.613898007235548e-06,
      "learning_rate": 4.1754027566584276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6053968,
      "step": 10650
    },
    {
      "epoch": 186.93805309734512,
      "grad_norm": 2.451193040542421e-06,
      "learning_rate": 4.174673958965852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6056832,
      "step": 10655
    },
    {
      "epoch": 187.01769911504425,
      "grad_norm": 3.1165575364866527e-06,
      "learning_rate": 4.173944903017507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6059416,
      "step": 10660
    },
    {
      "epoch": 187.10619469026548,
      "grad_norm": 1.742085260048043e-06,
      "learning_rate": 4.173215588925822e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6061800,
      "step": 10665
    },
    {
      "epoch": 187.1946902654867,
      "grad_norm": 1.6008160628189216e-06,
      "learning_rate": 4.172486016803266e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6065224,
      "step": 10670
    },
    {
      "epoch": 187.28318584070797,
      "grad_norm": 1.251563276127854e-06,
      "learning_rate": 4.171756186762349e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6068200,
      "step": 10675
    },
    {
      "epoch": 187.3716814159292,
      "grad_norm": 2.0848731310252333e-06,
      "learning_rate": 4.171026098915619e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6071112,
      "step": 10680
    },
    {
      "epoch": 187.46017699115043,
      "grad_norm": 6.38324399915291e-06,
      "learning_rate": 4.170295753375665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6074088,
      "step": 10685
    },
    {
      "epoch": 187.5486725663717,
      "grad_norm": 1.328559847024735e-06,
      "learning_rate": 4.169565150255117e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6076520,
      "step": 10690
    },
    {
      "epoch": 187.63716814159292,
      "grad_norm": 1.4199448514773394e-06,
      "learning_rate": 4.16883428966664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6079496,
      "step": 10695
    },
    {
      "epoch": 187.72566371681415,
      "grad_norm": 1.3672106433659792e-06,
      "learning_rate": 4.168103171722944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6082168,
      "step": 10700
    },
    {
      "epoch": 187.81415929203538,
      "grad_norm": 1.7731819070831989e-06,
      "learning_rate": 4.167371796536777e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6085144,
      "step": 10705
    },
    {
      "epoch": 187.90265486725664,
      "grad_norm": 1.4124938161330647e-06,
      "learning_rate": 4.166640164220924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6088328,
      "step": 10710
    },
    {
      "epoch": 187.99115044247787,
      "grad_norm": 1.7067867474906961e-06,
      "learning_rate": 4.1659082748882144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6090984,
      "step": 10715
    },
    {
      "epoch": 188.07079646017698,
      "grad_norm": 1.7242176681975252e-06,
      "learning_rate": 4.1651761286515135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6093456,
      "step": 10720
    },
    {
      "epoch": 188.15929203539824,
      "grad_norm": 1.7015612456816598e-06,
      "learning_rate": 4.164443725623728e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6096960,
      "step": 10725
    },
    {
      "epoch": 188.24778761061947,
      "grad_norm": 2.0042443793499842e-06,
      "learning_rate": 4.163711065917802e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6099568,
      "step": 10730
    },
    {
      "epoch": 188.3362831858407,
      "grad_norm": 4.121633537579328e-06,
      "learning_rate": 4.1629781496467234e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6102288,
      "step": 10735
    },
    {
      "epoch": 188.42477876106196,
      "grad_norm": 1.4463503248407505e-06,
      "learning_rate": 4.1622449769235164e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6105888,
      "step": 10740
    },
    {
      "epoch": 188.5132743362832,
      "grad_norm": 1.9571114080463303e-06,
      "learning_rate": 4.161511547861243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6108912,
      "step": 10745
    },
    {
      "epoch": 188.60176991150442,
      "grad_norm": 2.456811898809974e-06,
      "learning_rate": 4.1607778625730104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6111760,
      "step": 10750
    },
    {
      "epoch": 188.69026548672565,
      "grad_norm": 2.8883941922686063e-06,
      "learning_rate": 4.160043921171961e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6114768,
      "step": 10755
    },
    {
      "epoch": 188.7787610619469,
      "grad_norm": 1.8753571566776372e-05,
      "learning_rate": 4.159309723771276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6116992,
      "step": 10760
    },
    {
      "epoch": 188.86725663716814,
      "grad_norm": 3.6367457596497843e-06,
      "learning_rate": 4.158575270484181e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6119472,
      "step": 10765
    },
    {
      "epoch": 188.95575221238937,
      "grad_norm": 1.5916286884021247e-06,
      "learning_rate": 4.157840561423936e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6122544,
      "step": 10770
    },
    {
      "epoch": 189.0353982300885,
      "grad_norm": 1.662134764046641e-06,
      "learning_rate": 4.1571055967038416e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6125072,
      "step": 10775
    },
    {
      "epoch": 189.12389380530973,
      "grad_norm": 8.433838047494646e-06,
      "learning_rate": 4.156370376437241e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6128112,
      "step": 10780
    },
    {
      "epoch": 189.21238938053096,
      "grad_norm": 3.323101282148855e-06,
      "learning_rate": 4.155634900737513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6130800,
      "step": 10785
    },
    {
      "epoch": 189.30088495575222,
      "grad_norm": 1.5991845430107787e-06,
      "learning_rate": 4.1548991697180764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6133456,
      "step": 10790
    },
    {
      "epoch": 189.38938053097345,
      "grad_norm": 1.075057525667944e-06,
      "learning_rate": 4.1541631834923914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6136384,
      "step": 10795
    },
    {
      "epoch": 189.47787610619469,
      "grad_norm": 1.712256676000834e-06,
      "learning_rate": 4.153426942173956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6139264,
      "step": 10800
    },
    {
      "epoch": 189.47787610619469,
      "eval_loss": 0.4492727816104889,
      "eval_runtime": 1.0618,
      "eval_samples_per_second": 23.545,
      "eval_steps_per_second": 12.244,
      "num_input_tokens_seen": 6139264,
      "step": 10800
    },
    {
      "epoch": 189.56637168141592,
      "grad_norm": 1.5333433793784934e-06,
      "learning_rate": 4.152690445876308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6142000,
      "step": 10805
    },
    {
      "epoch": 189.65486725663717,
      "grad_norm": 3.472319349384634e-06,
      "learning_rate": 4.1519536947130245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6144832,
      "step": 10810
    },
    {
      "epoch": 189.7433628318584,
      "grad_norm": 1.3199127124607912e-06,
      "learning_rate": 4.151216688797722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6147456,
      "step": 10815
    },
    {
      "epoch": 189.83185840707964,
      "grad_norm": 7.185224603745155e-06,
      "learning_rate": 4.150479428244054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6150800,
      "step": 10820
    },
    {
      "epoch": 189.9203539823009,
      "grad_norm": 6.773881978006102e-06,
      "learning_rate": 4.1497419131657176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6154256,
      "step": 10825
    },
    {
      "epoch": 190.0,
      "grad_norm": 1.991072167584207e-06,
      "learning_rate": 4.149004143676447e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6156512,
      "step": 10830
    },
    {
      "epoch": 190.08849557522123,
      "grad_norm": 4.569707925838884e-06,
      "learning_rate": 4.148266119890015e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6159904,
      "step": 10835
    },
    {
      "epoch": 190.1769911504425,
      "grad_norm": 2.2703857212036382e-06,
      "learning_rate": 4.1475278419202324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6162944,
      "step": 10840
    },
    {
      "epoch": 190.26548672566372,
      "grad_norm": 3.2156665383809013e-06,
      "learning_rate": 4.146789309880953e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6165680,
      "step": 10845
    },
    {
      "epoch": 190.35398230088495,
      "grad_norm": 2.978380280183046e-06,
      "learning_rate": 4.146050523886068e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6168448,
      "step": 10850
    },
    {
      "epoch": 190.44247787610618,
      "grad_norm": 1.6054901834650082e-06,
      "learning_rate": 4.1453114840495055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6171136,
      "step": 10855
    },
    {
      "epoch": 190.53097345132744,
      "grad_norm": 3.0898418117430992e-06,
      "learning_rate": 4.1445721904852364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6173744,
      "step": 10860
    },
    {
      "epoch": 190.61946902654867,
      "grad_norm": 2.3191080344986403e-06,
      "learning_rate": 4.143832643307269e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6176416,
      "step": 10865
    },
    {
      "epoch": 190.7079646017699,
      "grad_norm": 6.1326732065936085e-06,
      "learning_rate": 4.1430928426296503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6179328,
      "step": 10870
    },
    {
      "epoch": 190.79646017699116,
      "grad_norm": 6.10708093518042e-06,
      "learning_rate": 4.142352788566466e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6182416,
      "step": 10875
    },
    {
      "epoch": 190.8849557522124,
      "grad_norm": 1.1494248610688373e-06,
      "learning_rate": 4.1416124812318424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6185072,
      "step": 10880
    },
    {
      "epoch": 190.97345132743362,
      "grad_norm": 1.5841400227145641e-06,
      "learning_rate": 4.1408719207399453e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6187904,
      "step": 10885
    },
    {
      "epoch": 191.05309734513276,
      "grad_norm": 3.490092012725654e-06,
      "learning_rate": 4.140131107204978e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6190096,
      "step": 10890
    },
    {
      "epoch": 191.141592920354,
      "grad_norm": 1.7139809642685577e-06,
      "learning_rate": 4.139390040741182e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6193072,
      "step": 10895
    },
    {
      "epoch": 191.23008849557522,
      "grad_norm": 2.5815081698965514e-06,
      "learning_rate": 4.1386487214628396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6195648,
      "step": 10900
    },
    {
      "epoch": 191.31858407079645,
      "grad_norm": 1.392501758346043e-06,
      "learning_rate": 4.137907149484272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6198896,
      "step": 10905
    },
    {
      "epoch": 191.4070796460177,
      "grad_norm": 1.6673460550009622e-06,
      "learning_rate": 4.137165324919839e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6201488,
      "step": 10910
    },
    {
      "epoch": 191.49557522123894,
      "grad_norm": 3.730104708665749e-06,
      "learning_rate": 4.136423247883939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6204720,
      "step": 10915
    },
    {
      "epoch": 191.58407079646017,
      "grad_norm": 1.1783268973886152e-06,
      "learning_rate": 4.135680918491009e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6207168,
      "step": 10920
    },
    {
      "epoch": 191.67256637168143,
      "grad_norm": 1.0876269698201213e-06,
      "learning_rate": 4.1349383368555265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6210000,
      "step": 10925
    },
    {
      "epoch": 191.76106194690266,
      "grad_norm": 1.8445704199621105e-06,
      "learning_rate": 4.1341955030920065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6212704,
      "step": 10930
    },
    {
      "epoch": 191.8495575221239,
      "grad_norm": 2.7751650577556575e-06,
      "learning_rate": 4.1334524173150036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6215680,
      "step": 10935
    },
    {
      "epoch": 191.93805309734512,
      "grad_norm": 2.3768470782670192e-06,
      "learning_rate": 4.13270907963911e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6218800,
      "step": 10940
    },
    {
      "epoch": 192.01769911504425,
      "grad_norm": 1.4102595287113218e-06,
      "learning_rate": 4.131965490178959e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6221184,
      "step": 10945
    },
    {
      "epoch": 192.10619469026548,
      "grad_norm": 8.303636604978237e-07,
      "learning_rate": 4.131221649049222e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6223936,
      "step": 10950
    },
    {
      "epoch": 192.1946902654867,
      "grad_norm": 1.7402563798896153e-06,
      "learning_rate": 4.130477556364606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6227248,
      "step": 10955
    },
    {
      "epoch": 192.28318584070797,
      "grad_norm": 3.8561861401831266e-06,
      "learning_rate": 4.129733212239861e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6229952,
      "step": 10960
    },
    {
      "epoch": 192.3716814159292,
      "grad_norm": 1.1083131994382711e-06,
      "learning_rate": 4.128988616789774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6232816,
      "step": 10965
    },
    {
      "epoch": 192.46017699115043,
      "grad_norm": 7.519735845562536e-06,
      "learning_rate": 4.1282437701291724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6235552,
      "step": 10970
    },
    {
      "epoch": 192.5486725663717,
      "grad_norm": 6.769204446754884e-06,
      "learning_rate": 4.1274986723729184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6238512,
      "step": 10975
    },
    {
      "epoch": 192.63716814159292,
      "grad_norm": 1.979848548216978e-06,
      "learning_rate": 4.126753323635917e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6241136,
      "step": 10980
    },
    {
      "epoch": 192.72566371681415,
      "grad_norm": 1.7711453210722539e-06,
      "learning_rate": 4.12600772403311e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6244112,
      "step": 10985
    },
    {
      "epoch": 192.81415929203538,
      "grad_norm": 1.1161658903802163e-06,
      "learning_rate": 4.125261873679479e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6247248,
      "step": 10990
    },
    {
      "epoch": 192.90265486725664,
      "grad_norm": 2.883022943933611e-06,
      "learning_rate": 4.124515772690042e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6250080,
      "step": 10995
    },
    {
      "epoch": 192.99115044247787,
      "grad_norm": 2.143379560948233e-06,
      "learning_rate": 4.123769421179858e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6252832,
      "step": 11000
    },
    {
      "epoch": 192.99115044247787,
      "eval_loss": 0.4487813711166382,
      "eval_runtime": 1.0613,
      "eval_samples_per_second": 23.557,
      "eval_steps_per_second": 12.25,
      "num_input_tokens_seen": 6252832,
      "step": 11000
    },
    {
      "epoch": 193.07079646017698,
      "grad_norm": 3.5223665690864436e-06,
      "learning_rate": 4.1230228192640236e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6254912,
      "step": 11005
    },
    {
      "epoch": 193.15929203539824,
      "grad_norm": 1.4233747833714006e-06,
      "learning_rate": 4.122275967057675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6257696,
      "step": 11010
    },
    {
      "epoch": 193.24778761061947,
      "grad_norm": 1.5849044530114043e-06,
      "learning_rate": 4.1215288646759846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6260304,
      "step": 11015
    },
    {
      "epoch": 193.3362831858407,
      "grad_norm": 4.902552518615266e-06,
      "learning_rate": 4.120781512234166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6263312,
      "step": 11020
    },
    {
      "epoch": 193.42477876106196,
      "grad_norm": 1.576393515279051e-06,
      "learning_rate": 4.120033909847471e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6266928,
      "step": 11025
    },
    {
      "epoch": 193.5132743362832,
      "grad_norm": 1.5283452512449003e-06,
      "learning_rate": 4.119286057631187e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6269808,
      "step": 11030
    },
    {
      "epoch": 193.60176991150442,
      "grad_norm": 9.12763482574519e-07,
      "learning_rate": 4.118537955700646e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6272512,
      "step": 11035
    },
    {
      "epoch": 193.69026548672565,
      "grad_norm": 1.4090054492044146e-06,
      "learning_rate": 4.11778960417121e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6275904,
      "step": 11040
    },
    {
      "epoch": 193.7787610619469,
      "grad_norm": 4.234141670167446e-06,
      "learning_rate": 4.117041003158288e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6279312,
      "step": 11045
    },
    {
      "epoch": 193.86725663716814,
      "grad_norm": 3.218345682398649e-06,
      "learning_rate": 4.1162921527773215e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6281760,
      "step": 11050
    },
    {
      "epoch": 193.95575221238937,
      "grad_norm": 3.02677767649584e-06,
      "learning_rate": 4.115543053143794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6284656,
      "step": 11055
    },
    {
      "epoch": 194.0353982300885,
      "grad_norm": 3.1767353902978357e-06,
      "learning_rate": 4.114793704373226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6286984,
      "step": 11060
    },
    {
      "epoch": 194.12389380530973,
      "grad_norm": 4.06848357670242e-06,
      "learning_rate": 4.114044106581175e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6289928,
      "step": 11065
    },
    {
      "epoch": 194.21238938053096,
      "grad_norm": 1.6945765537457191e-06,
      "learning_rate": 4.11329425988324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6292840,
      "step": 11070
    },
    {
      "epoch": 194.30088495575222,
      "grad_norm": 4.126075964450138e-06,
      "learning_rate": 4.112544164395056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6295624,
      "step": 11075
    },
    {
      "epoch": 194.38938053097345,
      "grad_norm": 1.315528379564057e-06,
      "learning_rate": 4.111793820232297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6298104,
      "step": 11080
    },
    {
      "epoch": 194.47787610619469,
      "grad_norm": 1.232660451933043e-06,
      "learning_rate": 4.1110432275106767e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6300776,
      "step": 11085
    },
    {
      "epoch": 194.56637168141592,
      "grad_norm": 3.047029622393893e-06,
      "learning_rate": 4.110292386345944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6303560,
      "step": 11090
    },
    {
      "epoch": 194.65486725663717,
      "grad_norm": 1.6733882830521907e-06,
      "learning_rate": 4.109541296853891e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6306280,
      "step": 11095
    },
    {
      "epoch": 194.7433628318584,
      "grad_norm": 1.3864524817108759e-06,
      "learning_rate": 4.108789959150341e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6309400,
      "step": 11100
    },
    {
      "epoch": 194.83185840707964,
      "grad_norm": 7.777183782309294e-06,
      "learning_rate": 4.108038373351163e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6312760,
      "step": 11105
    },
    {
      "epoch": 194.9203539823009,
      "grad_norm": 5.719358341593761e-06,
      "learning_rate": 4.10728653957226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6315624,
      "step": 11110
    },
    {
      "epoch": 195.0,
      "grad_norm": 7.005206043686485e-06,
      "learning_rate": 4.106534457929575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6318352,
      "step": 11115
    },
    {
      "epoch": 195.08849557522123,
      "grad_norm": 2.927961986642913e-06,
      "learning_rate": 4.105782128539086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6321360,
      "step": 11120
    },
    {
      "epoch": 195.1769911504425,
      "grad_norm": 1.5751919590911712e-06,
      "learning_rate": 4.1050295515168144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6324256,
      "step": 11125
    },
    {
      "epoch": 195.26548672566372,
      "grad_norm": 2.6030127173726214e-06,
      "learning_rate": 4.1042767269788155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6326896,
      "step": 11130
    },
    {
      "epoch": 195.35398230088495,
      "grad_norm": 1.5773840686961194e-06,
      "learning_rate": 4.103523655041185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6329600,
      "step": 11135
    },
    {
      "epoch": 195.44247787610618,
      "grad_norm": 1.1843351330753649e-06,
      "learning_rate": 4.102770335820055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6332768,
      "step": 11140
    },
    {
      "epoch": 195.53097345132744,
      "grad_norm": 1.8321769630347262e-06,
      "learning_rate": 4.1020167694315984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6335536,
      "step": 11145
    },
    {
      "epoch": 195.61946902654867,
      "grad_norm": 1.064110847437405e-06,
      "learning_rate": 4.101262955992023e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6338064,
      "step": 11150
    },
    {
      "epoch": 195.7079646017699,
      "grad_norm": 4.5527190195571166e-06,
      "learning_rate": 4.100508895617578e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6341056,
      "step": 11155
    },
    {
      "epoch": 195.79646017699116,
      "grad_norm": 5.396259894041577e-06,
      "learning_rate": 4.099754588424547e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6343856,
      "step": 11160
    },
    {
      "epoch": 195.8849557522124,
      "grad_norm": 5.522248557099374e-06,
      "learning_rate": 4.0990000345292546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6346928,
      "step": 11165
    },
    {
      "epoch": 195.97345132743362,
      "grad_norm": 1.787505880201934e-06,
      "learning_rate": 4.098245234048064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6349888,
      "step": 11170
    },
    {
      "epoch": 196.05309734513276,
      "grad_norm": 1.9734352463274263e-06,
      "learning_rate": 4.0974901870973726e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6352344,
      "step": 11175
    },
    {
      "epoch": 196.141592920354,
      "grad_norm": 2.647403107403079e-06,
      "learning_rate": 4.096734893793619e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6355608,
      "step": 11180
    },
    {
      "epoch": 196.23008849557522,
      "grad_norm": 2.2179910956765525e-06,
      "learning_rate": 4.095979354253279e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6358552,
      "step": 11185
    },
    {
      "epoch": 196.31858407079645,
      "grad_norm": 1.7641846170590725e-06,
      "learning_rate": 4.0952235685928656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6360840,
      "step": 11190
    },
    {
      "epoch": 196.4070796460177,
      "grad_norm": 4.752382210426731e-06,
      "learning_rate": 4.094467536928932e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6363720,
      "step": 11195
    },
    {
      "epoch": 196.49557522123894,
      "grad_norm": 1.1637982879619813e-06,
      "learning_rate": 4.093711259378067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6366440,
      "step": 11200
    },
    {
      "epoch": 196.49557522123894,
      "eval_loss": 0.4484032988548279,
      "eval_runtime": 1.0655,
      "eval_samples_per_second": 23.463,
      "eval_steps_per_second": 12.201,
      "num_input_tokens_seen": 6366440,
      "step": 11200
    },
    {
      "epoch": 196.58407079646017,
      "grad_norm": 3.975635991082527e-06,
      "learning_rate": 4.092954736056897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6369384,
      "step": 11205
    },
    {
      "epoch": 196.67256637168143,
      "grad_norm": 4.4288067329034675e-06,
      "learning_rate": 4.09219796708209e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6372072,
      "step": 11210
    },
    {
      "epoch": 196.76106194690266,
      "grad_norm": 1.6312645811922266e-06,
      "learning_rate": 4.0914409525703464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6374968,
      "step": 11215
    },
    {
      "epoch": 196.8495575221239,
      "grad_norm": 3.0302740015031304e-06,
      "learning_rate": 4.090683692638408e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6377592,
      "step": 11220
    },
    {
      "epoch": 196.93805309734512,
      "grad_norm": 3.2685825317457784e-06,
      "learning_rate": 4.089926187403056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6380888,
      "step": 11225
    },
    {
      "epoch": 197.01769911504425,
      "grad_norm": 2.1016901428083656e-06,
      "learning_rate": 4.0891684369811044e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6383104,
      "step": 11230
    },
    {
      "epoch": 197.10619469026548,
      "grad_norm": 9.848600939221797e-07,
      "learning_rate": 4.0884104414894107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6386400,
      "step": 11235
    },
    {
      "epoch": 197.1946902654867,
      "grad_norm": 8.801518561085686e-06,
      "learning_rate": 4.087652201044864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6389392,
      "step": 11240
    },
    {
      "epoch": 197.28318584070797,
      "grad_norm": 1.3620302752315183e-06,
      "learning_rate": 4.086893715764397e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6391968,
      "step": 11245
    },
    {
      "epoch": 197.3716814159292,
      "grad_norm": 1.49135337323969e-06,
      "learning_rate": 4.086134985764977e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6394720,
      "step": 11250
    },
    {
      "epoch": 197.46017699115043,
      "grad_norm": 1.841872403929301e-06,
      "learning_rate": 4.0853760111636085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6397808,
      "step": 11255
    },
    {
      "epoch": 197.5486725663717,
      "grad_norm": 2.2260639980231645e-06,
      "learning_rate": 4.084616792077337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6400448,
      "step": 11260
    },
    {
      "epoch": 197.63716814159292,
      "grad_norm": 3.516890501487069e-06,
      "learning_rate": 4.083857328623243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6403136,
      "step": 11265
    },
    {
      "epoch": 197.72566371681415,
      "grad_norm": 2.119704959113733e-06,
      "learning_rate": 4.083097620918444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6406288,
      "step": 11270
    },
    {
      "epoch": 197.81415929203538,
      "grad_norm": 5.075856734038098e-06,
      "learning_rate": 4.082337669080097e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6408896,
      "step": 11275
    },
    {
      "epoch": 197.90265486725664,
      "grad_norm": 1.8026538555204752e-06,
      "learning_rate": 4.081577473225398e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6411728,
      "step": 11280
    },
    {
      "epoch": 197.99115044247787,
      "grad_norm": 2.748168753896607e-06,
      "learning_rate": 4.080817033471577e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6414592,
      "step": 11285
    },
    {
      "epoch": 198.07079646017698,
      "grad_norm": 1.3288075706441305e-06,
      "learning_rate": 4.080056349935903e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6416920,
      "step": 11290
    },
    {
      "epoch": 198.15929203539824,
      "grad_norm": 6.4855335040192585e-06,
      "learning_rate": 4.079295422735684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6419336,
      "step": 11295
    },
    {
      "epoch": 198.24778761061947,
      "grad_norm": 9.868879260466201e-07,
      "learning_rate": 4.078534251988264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6422376,
      "step": 11300
    },
    {
      "epoch": 198.3362831858407,
      "grad_norm": 1.7687702893454116e-06,
      "learning_rate": 4.077772837811025e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6425320,
      "step": 11305
    },
    {
      "epoch": 198.42477876106196,
      "grad_norm": 1.9336832792760106e-06,
      "learning_rate": 4.0770111803213874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6428056,
      "step": 11310
    },
    {
      "epoch": 198.5132743362832,
      "grad_norm": 8.857588795763149e-07,
      "learning_rate": 4.076249279636807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6431480,
      "step": 11315
    },
    {
      "epoch": 198.60176991150442,
      "grad_norm": 1.439816060155863e-06,
      "learning_rate": 4.075487135874781e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6433976,
      "step": 11320
    },
    {
      "epoch": 198.69026548672565,
      "grad_norm": 1.9613009953900473e-06,
      "learning_rate": 4.074724749152837e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6437080,
      "step": 11325
    },
    {
      "epoch": 198.7787610619469,
      "grad_norm": 2.0938537090842146e-06,
      "learning_rate": 4.07396211958855e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6439896,
      "step": 11330
    },
    {
      "epoch": 198.86725663716814,
      "grad_norm": 1.215833094647678e-06,
      "learning_rate": 4.073199247299523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6442888,
      "step": 11335
    },
    {
      "epoch": 198.95575221238937,
      "grad_norm": 1.9831365989375627e-06,
      "learning_rate": 4.072436132403403e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6445448,
      "step": 11340
    },
    {
      "epoch": 199.0353982300885,
      "grad_norm": 1.7361619484290713e-06,
      "learning_rate": 4.0716727750178704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6447712,
      "step": 11345
    },
    {
      "epoch": 199.12389380530973,
      "grad_norm": 2.2379119855031604e-06,
      "learning_rate": 4.0709091752606455e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6450336,
      "step": 11350
    },
    {
      "epoch": 199.21238938053096,
      "grad_norm": 1.8025192503046128e-06,
      "learning_rate": 4.070145333249484e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6453584,
      "step": 11355
    },
    {
      "epoch": 199.30088495575222,
      "grad_norm": 1.2198232752780314e-06,
      "learning_rate": 4.069381249102181e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6456752,
      "step": 11360
    },
    {
      "epoch": 199.38938053097345,
      "grad_norm": 3.328129196233931e-06,
      "learning_rate": 4.0686169229365665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6459536,
      "step": 11365
    },
    {
      "epoch": 199.47787610619469,
      "grad_norm": 1.3087467323202873e-06,
      "learning_rate": 4.067852354870511e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6462160,
      "step": 11370
    },
    {
      "epoch": 199.56637168141592,
      "grad_norm": 1.2134634062022087e-06,
      "learning_rate": 4.067087545021919e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6465072,
      "step": 11375
    },
    {
      "epoch": 199.65486725663717,
      "grad_norm": 1.3340500117919873e-06,
      "learning_rate": 4.066322493508734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6467920,
      "step": 11380
    },
    {
      "epoch": 199.7433628318584,
      "grad_norm": 1.3176453421692713e-06,
      "learning_rate": 4.065557200448937e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6470416,
      "step": 11385
    },
    {
      "epoch": 199.83185840707964,
      "grad_norm": 7.912633464002283e-07,
      "learning_rate": 4.064791665960546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6473280,
      "step": 11390
    },
    {
      "epoch": 199.9203539823009,
      "grad_norm": 4.581635494105285e-06,
      "learning_rate": 4.064025890161615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6476512,
      "step": 11395
    },
    {
      "epoch": 200.0,
      "grad_norm": 1.635540570532612e-06,
      "learning_rate": 4.0632598731702373e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6478776,
      "step": 11400
    },
    {
      "epoch": 200.0,
      "eval_loss": 0.4492186903953552,
      "eval_runtime": 1.0713,
      "eval_samples_per_second": 23.336,
      "eval_steps_per_second": 12.135,
      "num_input_tokens_seen": 6478776,
      "step": 11400
    },
    {
      "epoch": 200.08849557522123,
      "grad_norm": 1.4944949953132891e-06,
      "learning_rate": 4.0624936151045426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6481576,
      "step": 11405
    },
    {
      "epoch": 200.1769911504425,
      "grad_norm": 2.194245553255314e-06,
      "learning_rate": 4.061727116082696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6484520,
      "step": 11410
    },
    {
      "epoch": 200.26548672566372,
      "grad_norm": 1.2640377917705337e-06,
      "learning_rate": 4.060960376222903e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6487592,
      "step": 11415
    },
    {
      "epoch": 200.35398230088495,
      "grad_norm": 1.4120295190878096e-06,
      "learning_rate": 4.0601933956434034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6490152,
      "step": 11420
    },
    {
      "epoch": 200.44247787610618,
      "grad_norm": 7.749733867967734e-07,
      "learning_rate": 4.059426174462476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6493016,
      "step": 11425
    },
    {
      "epoch": 200.53097345132744,
      "grad_norm": 5.128458269609837e-06,
      "learning_rate": 4.058658712798435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6496104,
      "step": 11430
    },
    {
      "epoch": 200.61946902654867,
      "grad_norm": 1.0324563390895491e-06,
      "learning_rate": 4.0578910107696336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6498632,
      "step": 11435
    },
    {
      "epoch": 200.7079646017699,
      "grad_norm": 1.2275489780222415e-06,
      "learning_rate": 4.05712306849446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6501512,
      "step": 11440
    },
    {
      "epoch": 200.79646017699116,
      "grad_norm": 1.85925841833523e-06,
      "learning_rate": 4.0563548860913415e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6504200,
      "step": 11445
    },
    {
      "epoch": 200.8849557522124,
      "grad_norm": 1.3921229538027546e-06,
      "learning_rate": 4.0555864636787414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6507496,
      "step": 11450
    },
    {
      "epoch": 200.97345132743362,
      "grad_norm": 1.6528758806089172e-06,
      "learning_rate": 4.054817801375159e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6510568,
      "step": 11455
    },
    {
      "epoch": 201.05309734513276,
      "grad_norm": 1.1322931641188916e-06,
      "learning_rate": 4.054048899299134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6512936,
      "step": 11460
    },
    {
      "epoch": 201.141592920354,
      "grad_norm": 1.6798330761957914e-06,
      "learning_rate": 4.0532797575692385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6515416,
      "step": 11465
    },
    {
      "epoch": 201.23008849557522,
      "grad_norm": 8.838471785566071e-07,
      "learning_rate": 4.052510376304085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6518744,
      "step": 11470
    },
    {
      "epoch": 201.31858407079645,
      "grad_norm": 1.5627765606041066e-06,
      "learning_rate": 4.051740755622321e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6521080,
      "step": 11475
    },
    {
      "epoch": 201.4070796460177,
      "grad_norm": 1.0038344271379174e-06,
      "learning_rate": 4.050970895642632e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6523720,
      "step": 11480
    },
    {
      "epoch": 201.49557522123894,
      "grad_norm": 3.0391868222068297e-06,
      "learning_rate": 4.050200796483741e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6526760,
      "step": 11485
    },
    {
      "epoch": 201.58407079646017,
      "grad_norm": 9.717964530864265e-07,
      "learning_rate": 4.049430458264405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6529992,
      "step": 11490
    },
    {
      "epoch": 201.67256637168143,
      "grad_norm": 3.2339696645067306e-06,
      "learning_rate": 4.048659881103422e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6532584,
      "step": 11495
    },
    {
      "epoch": 201.76106194690266,
      "grad_norm": 1.8119761762136477e-06,
      "learning_rate": 4.0478890651196235e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6535768,
      "step": 11500
    },
    {
      "epoch": 201.8495575221239,
      "grad_norm": 5.715405677619856e-06,
      "learning_rate": 4.047118010431879e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6538728,
      "step": 11505
    },
    {
      "epoch": 201.93805309734512,
      "grad_norm": 1.3411756754067028e-06,
      "learning_rate": 4.046346717159094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6541320,
      "step": 11510
    },
    {
      "epoch": 202.01769911504425,
      "grad_norm": 1.3435428627417423e-06,
      "learning_rate": 4.045575185420214e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6543928,
      "step": 11515
    },
    {
      "epoch": 202.10619469026548,
      "grad_norm": 1.8220889614894986e-06,
      "learning_rate": 4.0448034153342165e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6546680,
      "step": 11520
    },
    {
      "epoch": 202.1946902654867,
      "grad_norm": 1.3852849178874749e-06,
      "learning_rate": 4.0440314070201194e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6549400,
      "step": 11525
    },
    {
      "epoch": 202.28318584070797,
      "grad_norm": 1.1832344171125442e-06,
      "learning_rate": 4.043259160596976e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6552088,
      "step": 11530
    },
    {
      "epoch": 202.3716814159292,
      "grad_norm": 1.7744266642694129e-06,
      "learning_rate": 4.0424866761838767e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6555048,
      "step": 11535
    },
    {
      "epoch": 202.46017699115043,
      "grad_norm": 1.0186043255089317e-05,
      "learning_rate": 4.041713953899948e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6558280,
      "step": 11540
    },
    {
      "epoch": 202.5486725663717,
      "grad_norm": 2.502475126675563e-06,
      "learning_rate": 4.0409409938643515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6561240,
      "step": 11545
    },
    {
      "epoch": 202.63716814159292,
      "grad_norm": 4.705413630290423e-06,
      "learning_rate": 4.0401677961962904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6564040,
      "step": 11550
    },
    {
      "epoch": 202.72566371681415,
      "grad_norm": 4.151103439653525e-06,
      "learning_rate": 4.039394361015001e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6566536,
      "step": 11555
    },
    {
      "epoch": 202.81415929203538,
      "grad_norm": 2.5388039830431808e-06,
      "learning_rate": 4.038620688439755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6569368,
      "step": 11560
    },
    {
      "epoch": 202.90265486725664,
      "grad_norm": 1.6703896790204453e-06,
      "learning_rate": 4.037846778589862e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6572024,
      "step": 11565
    },
    {
      "epoch": 202.99115044247787,
      "grad_norm": 3.2930383895291016e-06,
      "learning_rate": 4.0370726315846715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6574904,
      "step": 11570
    },
    {
      "epoch": 203.07079646017698,
      "grad_norm": 1.1435306532803224e-06,
      "learning_rate": 4.036298247543565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6577192,
      "step": 11575
    },
    {
      "epoch": 203.15929203539824,
      "grad_norm": 1.5049697594804456e-06,
      "learning_rate": 4.035523626585962e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6580680,
      "step": 11580
    },
    {
      "epoch": 203.24778761061947,
      "grad_norm": 1.2396469628583873e-06,
      "learning_rate": 4.0347487688313194e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6583640,
      "step": 11585
    },
    {
      "epoch": 203.3362831858407,
      "grad_norm": 1.1291936061752494e-06,
      "learning_rate": 4.0339736743991296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6586648,
      "step": 11590
    },
    {
      "epoch": 203.42477876106196,
      "grad_norm": 2.1078531062812544e-06,
      "learning_rate": 4.0331983434089227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6589480,
      "step": 11595
    },
    {
      "epoch": 203.5132743362832,
      "grad_norm": 4.17250294049154e-06,
      "learning_rate": 4.032422775980264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6592280,
      "step": 11600
    },
    {
      "epoch": 203.5132743362832,
      "eval_loss": 0.45208242535591125,
      "eval_runtime": 1.0684,
      "eval_samples_per_second": 23.399,
      "eval_steps_per_second": 12.167,
      "num_input_tokens_seen": 6592280,
      "step": 11600
    },
    {
      "epoch": 203.60176991150442,
      "grad_norm": 1.0816828535098466e-06,
      "learning_rate": 4.031646972232754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6594760,
      "step": 11605
    },
    {
      "epoch": 203.69026548672565,
      "grad_norm": 1.3139002703610458e-06,
      "learning_rate": 4.0308709322860344e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6598040,
      "step": 11610
    },
    {
      "epoch": 203.7787610619469,
      "grad_norm": 3.7435129343066365e-06,
      "learning_rate": 4.0300946562597784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6600792,
      "step": 11615
    },
    {
      "epoch": 203.86725663716814,
      "grad_norm": 7.916132744867355e-06,
      "learning_rate": 4.029318144273698e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6604136,
      "step": 11620
    },
    {
      "epoch": 203.95575221238937,
      "grad_norm": 2.3505451736127725e-06,
      "learning_rate": 4.0285413964475415e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6606392,
      "step": 11625
    },
    {
      "epoch": 204.0353982300885,
      "grad_norm": 1.4706243973705568e-06,
      "learning_rate": 4.0277644129010927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6608488,
      "step": 11630
    },
    {
      "epoch": 204.12389380530973,
      "grad_norm": 1.4524167681884137e-06,
      "learning_rate": 4.0269871937541724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6611128,
      "step": 11635
    },
    {
      "epoch": 204.21238938053096,
      "grad_norm": 1.9808317119895946e-06,
      "learning_rate": 4.026209739126637e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6613832,
      "step": 11640
    },
    {
      "epoch": 204.30088495575222,
      "grad_norm": 3.843083050014684e-06,
      "learning_rate": 4.025432049138381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6617144,
      "step": 11645
    },
    {
      "epoch": 204.38938053097345,
      "grad_norm": 1.2041155059705488e-06,
      "learning_rate": 4.0246541239093325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6619848,
      "step": 11650
    },
    {
      "epoch": 204.47787610619469,
      "grad_norm": 1.542932750453474e-06,
      "learning_rate": 4.023875963559459e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6622664,
      "step": 11655
    },
    {
      "epoch": 204.56637168141592,
      "grad_norm": 8.765926509113342e-07,
      "learning_rate": 4.023097568208761e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6625672,
      "step": 11660
    },
    {
      "epoch": 204.65486725663717,
      "grad_norm": 1.1482914032967528e-06,
      "learning_rate": 4.022318937977277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6628312,
      "step": 11665
    },
    {
      "epoch": 204.7433628318584,
      "grad_norm": 2.46404033532599e-06,
      "learning_rate": 4.021540072985084e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6631224,
      "step": 11670
    },
    {
      "epoch": 204.83185840707964,
      "grad_norm": 2.6876627998717595e-06,
      "learning_rate": 4.020760973352289e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6634568,
      "step": 11675
    },
    {
      "epoch": 204.9203539823009,
      "grad_norm": 1.3302602610565373e-06,
      "learning_rate": 4.019981639199042e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6637528,
      "step": 11680
    },
    {
      "epoch": 205.0,
      "grad_norm": 1.290190425606852e-06,
      "learning_rate": 4.0192020706455245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6639624,
      "step": 11685
    },
    {
      "epoch": 205.08849557522123,
      "grad_norm": 4.242806880938588e-06,
      "learning_rate": 4.018422267811956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6642280,
      "step": 11690
    },
    {
      "epoch": 205.1769911504425,
      "grad_norm": 3.1893296181806363e-06,
      "learning_rate": 4.017642230818592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6645784,
      "step": 11695
    },
    {
      "epoch": 205.26548672566372,
      "grad_norm": 1.3341017393031507e-06,
      "learning_rate": 4.0168619597857246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6648008,
      "step": 11700
    },
    {
      "epoch": 205.35398230088495,
      "grad_norm": 1.7925315205502557e-06,
      "learning_rate": 4.016081454833681e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6650904,
      "step": 11705
    },
    {
      "epoch": 205.44247787610618,
      "grad_norm": 1.1367701517883688e-06,
      "learning_rate": 4.0153007160828245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6653848,
      "step": 11710
    },
    {
      "epoch": 205.53097345132744,
      "grad_norm": 1.6368003343814053e-06,
      "learning_rate": 4.0145197436535555e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6656632,
      "step": 11715
    },
    {
      "epoch": 205.61946902654867,
      "grad_norm": 2.377887767579523e-06,
      "learning_rate": 4.0137385376663095e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6659528,
      "step": 11720
    },
    {
      "epoch": 205.7079646017699,
      "grad_norm": 1.1348606676619966e-06,
      "learning_rate": 4.012957098241558e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6663000,
      "step": 11725
    },
    {
      "epoch": 205.79646017699116,
      "grad_norm": 1.7203528841491789e-06,
      "learning_rate": 4.0121754254998076e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6665848,
      "step": 11730
    },
    {
      "epoch": 205.8849557522124,
      "grad_norm": 3.4101976780220866e-06,
      "learning_rate": 4.011393519561606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6668472,
      "step": 11735
    },
    {
      "epoch": 205.97345132743362,
      "grad_norm": 1.1124528782602283e-06,
      "learning_rate": 4.010611380547529e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6671320,
      "step": 11740
    },
    {
      "epoch": 206.05309734513276,
      "grad_norm": 1.4864128843328217e-06,
      "learning_rate": 4.009829008578192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6673536,
      "step": 11745
    },
    {
      "epoch": 206.141592920354,
      "grad_norm": 8.927169119488099e-07,
      "learning_rate": 4.00904640377425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6676256,
      "step": 11750
    },
    {
      "epoch": 206.23008849557522,
      "grad_norm": 1.7881246776596527e-06,
      "learning_rate": 4.0082635662563886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6679232,
      "step": 11755
    },
    {
      "epoch": 206.31858407079645,
      "grad_norm": 2.742400056376937e-06,
      "learning_rate": 4.007480496145331e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6682224,
      "step": 11760
    },
    {
      "epoch": 206.4070796460177,
      "grad_norm": 1.0480717946848017e-06,
      "learning_rate": 4.006697193561837e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6684880,
      "step": 11765
    },
    {
      "epoch": 206.49557522123894,
      "grad_norm": 1.2266851854292327e-06,
      "learning_rate": 4.005913658626701e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6688224,
      "step": 11770
    },
    {
      "epoch": 206.58407079646017,
      "grad_norm": 9.515209740129649e-07,
      "learning_rate": 4.005129891460754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6691040,
      "step": 11775
    },
    {
      "epoch": 206.67256637168143,
      "grad_norm": 1.3624086250274559e-06,
      "learning_rate": 4.004345892184864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6694016,
      "step": 11780
    },
    {
      "epoch": 206.76106194690266,
      "grad_norm": 2.159259111067513e-06,
      "learning_rate": 4.003561660919932e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6696736,
      "step": 11785
    },
    {
      "epoch": 206.8495575221239,
      "grad_norm": 2.1821335849381285e-06,
      "learning_rate": 4.002777197786897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6700016,
      "step": 11790
    },
    {
      "epoch": 206.93805309734512,
      "grad_norm": 4.501533567236038e-06,
      "learning_rate": 4.0019925029067326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6702592,
      "step": 11795
    },
    {
      "epoch": 207.01769911504425,
      "grad_norm": 5.403725936048431e-06,
      "learning_rate": 4.0012075764004495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6704968,
      "step": 11800
    },
    {
      "epoch": 207.01769911504425,
      "eval_loss": 0.45573821663856506,
      "eval_runtime": 1.061,
      "eval_samples_per_second": 23.563,
      "eval_steps_per_second": 12.253,
      "num_input_tokens_seen": 6704968,
      "step": 11800
    },
    {
      "epoch": 207.10619469026548,
      "grad_norm": 1.4801373708905885e-06,
      "learning_rate": 4.000422418389094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6708008,
      "step": 11805
    },
    {
      "epoch": 207.1946902654867,
      "grad_norm": 2.281687329741544e-06,
      "learning_rate": 3.999637028993744e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6711368,
      "step": 11810
    },
    {
      "epoch": 207.28318584070797,
      "grad_norm": 1.2288161315154866e-06,
      "learning_rate": 3.99885140833552e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6714280,
      "step": 11815
    },
    {
      "epoch": 207.3716814159292,
      "grad_norm": 1.3265181451060926e-06,
      "learning_rate": 3.998065556535572e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6716840,
      "step": 11820
    },
    {
      "epoch": 207.46017699115043,
      "grad_norm": 7.118061375877005e-07,
      "learning_rate": 3.9972794737150895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6719288,
      "step": 11825
    },
    {
      "epoch": 207.5486725663717,
      "grad_norm": 1.1143188203277532e-06,
      "learning_rate": 3.996493159995297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6722488,
      "step": 11830
    },
    {
      "epoch": 207.63716814159292,
      "grad_norm": 1.6522320720468997e-06,
      "learning_rate": 3.995706615497453e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6725464,
      "step": 11835
    },
    {
      "epoch": 207.72566371681415,
      "grad_norm": 6.541509378621413e-07,
      "learning_rate": 3.994919840342852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6728376,
      "step": 11840
    },
    {
      "epoch": 207.81415929203538,
      "grad_norm": 2.3746567876514746e-06,
      "learning_rate": 3.994132834652825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6730856,
      "step": 11845
    },
    {
      "epoch": 207.90265486725664,
      "grad_norm": 7.808196755831887e-07,
      "learning_rate": 3.99334559854874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6733608,
      "step": 11850
    },
    {
      "epoch": 207.99115044247787,
      "grad_norm": 1.420289322595636e-06,
      "learning_rate": 3.9925581321519955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6736904,
      "step": 11855
    },
    {
      "epoch": 208.07079646017698,
      "grad_norm": 2.0964048417226877e-06,
      "learning_rate": 3.991770435584031e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6739352,
      "step": 11860
    },
    {
      "epoch": 208.15929203539824,
      "grad_norm": 1.6336825865437277e-06,
      "learning_rate": 3.990982508966319e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6742152,
      "step": 11865
    },
    {
      "epoch": 208.24778761061947,
      "grad_norm": 4.157649073022185e-06,
      "learning_rate": 3.990194352420367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6745224,
      "step": 11870
    },
    {
      "epoch": 208.3362831858407,
      "grad_norm": 4.4499947762233205e-06,
      "learning_rate": 3.9894059660677184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6748200,
      "step": 11875
    },
    {
      "epoch": 208.42477876106196,
      "grad_norm": 1.3185066336518503e-06,
      "learning_rate": 3.9886173500299526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6750888,
      "step": 11880
    },
    {
      "epoch": 208.5132743362832,
      "grad_norm": 2.1684961666323943e-06,
      "learning_rate": 3.987828504428685e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6753848,
      "step": 11885
    },
    {
      "epoch": 208.60176991150442,
      "grad_norm": 2.062406565528363e-06,
      "learning_rate": 3.987039429385565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6756264,
      "step": 11890
    },
    {
      "epoch": 208.69026548672565,
      "grad_norm": 9.83513928076718e-07,
      "learning_rate": 3.986250125022277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6759224,
      "step": 11895
    },
    {
      "epoch": 208.7787610619469,
      "grad_norm": 1.098767370422138e-06,
      "learning_rate": 3.985460591460544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6761736,
      "step": 11900
    },
    {
      "epoch": 208.86725663716814,
      "grad_norm": 5.508076810656348e-06,
      "learning_rate": 3.984670828822118e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6764824,
      "step": 11905
    },
    {
      "epoch": 208.95575221238937,
      "grad_norm": 1.4557039094142965e-06,
      "learning_rate": 3.983880837228794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6768040,
      "step": 11910
    },
    {
      "epoch": 209.0353982300885,
      "grad_norm": 7.770983074806281e-07,
      "learning_rate": 3.983090616802396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6770584,
      "step": 11915
    },
    {
      "epoch": 209.12389380530973,
      "grad_norm": 1.0965549108732375e-06,
      "learning_rate": 3.982300167664788e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6773256,
      "step": 11920
    },
    {
      "epoch": 209.21238938053096,
      "grad_norm": 1.3267798522065277e-06,
      "learning_rate": 3.981509489937868e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6775960,
      "step": 11925
    },
    {
      "epoch": 209.30088495575222,
      "grad_norm": 9.842293593465001e-07,
      "learning_rate": 3.9807185837435643e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6778840,
      "step": 11930
    },
    {
      "epoch": 209.38938053097345,
      "grad_norm": 8.432967888438725e-07,
      "learning_rate": 3.9799274492038484e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6781656,
      "step": 11935
    },
    {
      "epoch": 209.47787610619469,
      "grad_norm": 1.1795575574069517e-06,
      "learning_rate": 3.979136086440722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6784824,
      "step": 11940
    },
    {
      "epoch": 209.56637168141592,
      "grad_norm": 1.3722878975386266e-06,
      "learning_rate": 3.9783444955762226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6787592,
      "step": 11945
    },
    {
      "epoch": 209.65486725663717,
      "grad_norm": 1.1584772892092587e-06,
      "learning_rate": 3.977552676732424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6790600,
      "step": 11950
    },
    {
      "epoch": 209.7433628318584,
      "grad_norm": 1.343548319709953e-06,
      "learning_rate": 3.976760630031435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6793096,
      "step": 11955
    },
    {
      "epoch": 209.83185840707964,
      "grad_norm": 2.554633510953863e-06,
      "learning_rate": 3.975968355595398e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6796776,
      "step": 11960
    },
    {
      "epoch": 209.9203539823009,
      "grad_norm": 2.3672891984460875e-06,
      "learning_rate": 3.9751758535464935e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6799464,
      "step": 11965
    },
    {
      "epoch": 210.0,
      "grad_norm": 1.0322339676349657e-06,
      "learning_rate": 3.9743831240069326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6802096,
      "step": 11970
    },
    {
      "epoch": 210.08849557522123,
      "grad_norm": 2.9130628718121443e-06,
      "learning_rate": 3.9735901670989675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6804736,
      "step": 11975
    },
    {
      "epoch": 210.1769911504425,
      "grad_norm": 1.3783306940240436e-06,
      "learning_rate": 3.97279698294488e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6807408,
      "step": 11980
    },
    {
      "epoch": 210.26548672566372,
      "grad_norm": 9.778965477380552e-07,
      "learning_rate": 3.9720035716669876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6810368,
      "step": 11985
    },
    {
      "epoch": 210.35398230088495,
      "grad_norm": 1.2672337561525637e-06,
      "learning_rate": 3.9712099333876474e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6813776,
      "step": 11990
    },
    {
      "epoch": 210.44247787610618,
      "grad_norm": 9.283787107960961e-07,
      "learning_rate": 3.9704160682292475e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6816432,
      "step": 11995
    },
    {
      "epoch": 210.53097345132744,
      "grad_norm": 1.7942353451871895e-06,
      "learning_rate": 3.9696219763142106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6819568,
      "step": 12000
    },
    {
      "epoch": 210.53097345132744,
      "eval_loss": 0.44632595777511597,
      "eval_runtime": 1.0662,
      "eval_samples_per_second": 23.447,
      "eval_steps_per_second": 12.193,
      "num_input_tokens_seen": 6819568,
      "step": 12000
    },
    {
      "epoch": 210.61946902654867,
      "grad_norm": 1.0262928071824717e-06,
      "learning_rate": 3.968827657764997e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6822048,
      "step": 12005
    },
    {
      "epoch": 210.7079646017699,
      "grad_norm": 2.1237383407424204e-06,
      "learning_rate": 3.9680331127041e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6824960,
      "step": 12010
    },
    {
      "epoch": 210.79646017699116,
      "grad_norm": 2.5427873424632708e-06,
      "learning_rate": 3.9672383412540495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6827856,
      "step": 12015
    },
    {
      "epoch": 210.8849557522124,
      "grad_norm": 1.9791305021499284e-06,
      "learning_rate": 3.966443343537407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6831024,
      "step": 12020
    },
    {
      "epoch": 210.97345132743362,
      "grad_norm": 9.458694876229856e-07,
      "learning_rate": 3.965648119676772e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6833952,
      "step": 12025
    },
    {
      "epoch": 211.05309734513276,
      "grad_norm": 9.881410960588255e-07,
      "learning_rate": 3.96485266979478e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6836016,
      "step": 12030
    },
    {
      "epoch": 211.141592920354,
      "grad_norm": 4.176853508397471e-06,
      "learning_rate": 3.9640569940140974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6838512,
      "step": 12035
    },
    {
      "epoch": 211.23008849557522,
      "grad_norm": 9.962379863281967e-07,
      "learning_rate": 3.963261092457428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6841424,
      "step": 12040
    },
    {
      "epoch": 211.31858407079645,
      "grad_norm": 1.0267485777148977e-06,
      "learning_rate": 3.962464965247509e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6844480,
      "step": 12045
    },
    {
      "epoch": 211.4070796460177,
      "grad_norm": 1.454361722608155e-06,
      "learning_rate": 3.9616686125071135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6847456,
      "step": 12050
    },
    {
      "epoch": 211.49557522123894,
      "grad_norm": 6.583928779946291e-07,
      "learning_rate": 3.9608720343590506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6850112,
      "step": 12055
    },
    {
      "epoch": 211.58407079646017,
      "grad_norm": 1.148387696048303e-06,
      "learning_rate": 3.960075230926161e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6853136,
      "step": 12060
    },
    {
      "epoch": 211.67256637168143,
      "grad_norm": 1.0771645975182764e-06,
      "learning_rate": 3.959278202331322e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6856240,
      "step": 12065
    },
    {
      "epoch": 211.76106194690266,
      "grad_norm": 1.361692284262972e-06,
      "learning_rate": 3.958480948697446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6859168,
      "step": 12070
    },
    {
      "epoch": 211.8495575221239,
      "grad_norm": 7.881403689680155e-07,
      "learning_rate": 3.95768347014748e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6862240,
      "step": 12075
    },
    {
      "epoch": 211.93805309734512,
      "grad_norm": 7.299021831386199e-07,
      "learning_rate": 3.956885766804404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6865344,
      "step": 12080
    },
    {
      "epoch": 212.01769911504425,
      "grad_norm": 2.6830819024326047e-06,
      "learning_rate": 3.956087838791235e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6868176,
      "step": 12085
    },
    {
      "epoch": 212.10619469026548,
      "grad_norm": 2.7354658413969446e-06,
      "learning_rate": 3.955289686231022e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6871296,
      "step": 12090
    },
    {
      "epoch": 212.1946902654867,
      "grad_norm": 8.387081038563338e-07,
      "learning_rate": 3.9544913092468504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6873888,
      "step": 12095
    },
    {
      "epoch": 212.28318584070797,
      "grad_norm": 1.304673219237884e-06,
      "learning_rate": 3.9536927079618425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6876848,
      "step": 12100
    },
    {
      "epoch": 212.3716814159292,
      "grad_norm": 5.424222763394937e-06,
      "learning_rate": 3.9528938824991494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6880000,
      "step": 12105
    },
    {
      "epoch": 212.46017699115043,
      "grad_norm": 2.219047019025311e-06,
      "learning_rate": 3.952094832981962e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6882928,
      "step": 12110
    },
    {
      "epoch": 212.5486725663717,
      "grad_norm": 6.984154765632411e-07,
      "learning_rate": 3.951295559533503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6886208,
      "step": 12115
    },
    {
      "epoch": 212.63716814159292,
      "grad_norm": 1.256527525583806e-06,
      "learning_rate": 3.95049606227703e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6888736,
      "step": 12120
    },
    {
      "epoch": 212.72566371681415,
      "grad_norm": 9.264368827643921e-07,
      "learning_rate": 3.949696341335838e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6891696,
      "step": 12125
    },
    {
      "epoch": 212.81415929203538,
      "grad_norm": 1.3556930298364023e-06,
      "learning_rate": 3.9488963968332503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6894224,
      "step": 12130
    },
    {
      "epoch": 212.90265486725664,
      "grad_norm": 8.806981668385561e-07,
      "learning_rate": 3.948096228892631e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6896720,
      "step": 12135
    },
    {
      "epoch": 212.99115044247787,
      "grad_norm": 9.968828180717537e-07,
      "learning_rate": 3.947295837637375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6899408,
      "step": 12140
    },
    {
      "epoch": 213.07079646017698,
      "grad_norm": 7.171113907133986e-07,
      "learning_rate": 3.9464952231909135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6902216,
      "step": 12145
    },
    {
      "epoch": 213.15929203539824,
      "grad_norm": 1.3957567261968507e-06,
      "learning_rate": 3.945694385676711e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6905544,
      "step": 12150
    },
    {
      "epoch": 213.24778761061947,
      "grad_norm": 3.1772774491400924e-06,
      "learning_rate": 3.944893325218265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6907976,
      "step": 12155
    },
    {
      "epoch": 213.3362831858407,
      "grad_norm": 9.770178621693049e-07,
      "learning_rate": 3.944092041939112e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6911176,
      "step": 12160
    },
    {
      "epoch": 213.42477876106196,
      "grad_norm": 8.789657499619352e-07,
      "learning_rate": 3.943290535962818e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6913848,
      "step": 12165
    },
    {
      "epoch": 213.5132743362832,
      "grad_norm": 1.853947878771578e-06,
      "learning_rate": 3.942488807412985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6916952,
      "step": 12170
    },
    {
      "epoch": 213.60176991150442,
      "grad_norm": 5.1409324441920035e-06,
      "learning_rate": 3.941686856413251e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6919560,
      "step": 12175
    },
    {
      "epoch": 213.69026548672565,
      "grad_norm": 3.00024339594529e-06,
      "learning_rate": 3.9408846830872874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6922472,
      "step": 12180
    },
    {
      "epoch": 213.7787610619469,
      "grad_norm": 4.268864358891733e-06,
      "learning_rate": 3.940082287558798e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6925528,
      "step": 12185
    },
    {
      "epoch": 213.86725663716814,
      "grad_norm": 1.1274654525550432e-06,
      "learning_rate": 3.939279669951522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6928072,
      "step": 12190
    },
    {
      "epoch": 213.95575221238937,
      "grad_norm": 5.259584668237949e-06,
      "learning_rate": 3.938476830389234e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6930616,
      "step": 12195
    },
    {
      "epoch": 214.0353982300885,
      "grad_norm": 3.2683508379705017e-06,
      "learning_rate": 3.937673768995742e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6933264,
      "step": 12200
    },
    {
      "epoch": 214.0353982300885,
      "eval_loss": 0.4518740773200989,
      "eval_runtime": 1.0617,
      "eval_samples_per_second": 23.548,
      "eval_steps_per_second": 12.245,
      "num_input_tokens_seen": 6933264,
      "step": 12200
    },
    {
      "epoch": 214.12389380530973,
      "grad_norm": 1.430286488357524e-06,
      "learning_rate": 3.936870485894888e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6935952,
      "step": 12205
    },
    {
      "epoch": 214.21238938053096,
      "grad_norm": 2.9499399261112558e-06,
      "learning_rate": 3.9360669812105475e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6938960,
      "step": 12210
    },
    {
      "epoch": 214.30088495575222,
      "grad_norm": 1.0727806056820555e-06,
      "learning_rate": 3.9352632550666325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6941936,
      "step": 12215
    },
    {
      "epoch": 214.38938053097345,
      "grad_norm": 1.0940425454464275e-06,
      "learning_rate": 3.9344593075870866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6944496,
      "step": 12220
    },
    {
      "epoch": 214.47787610619469,
      "grad_norm": 1.125289259107376e-06,
      "learning_rate": 3.933655138895889e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6947264,
      "step": 12225
    },
    {
      "epoch": 214.56637168141592,
      "grad_norm": 1.747732767398702e-06,
      "learning_rate": 3.932850749117053e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6949744,
      "step": 12230
    },
    {
      "epoch": 214.65486725663717,
      "grad_norm": 1.0637041896188748e-06,
      "learning_rate": 3.932046138374624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6952432,
      "step": 12235
    },
    {
      "epoch": 214.7433628318584,
      "grad_norm": 2.702301799217821e-06,
      "learning_rate": 3.9312413067926854e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6955008,
      "step": 12240
    },
    {
      "epoch": 214.83185840707964,
      "grad_norm": 6.611600383621408e-06,
      "learning_rate": 3.9304362544953506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6957856,
      "step": 12245
    },
    {
      "epoch": 214.9203539823009,
      "grad_norm": 1.7500472040410386e-06,
      "learning_rate": 3.929630981606769e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6960784,
      "step": 12250
    },
    {
      "epoch": 215.0,
      "grad_norm": 6.633783868892351e-07,
      "learning_rate": 3.928825488251124e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6963416,
      "step": 12255
    },
    {
      "epoch": 215.08849557522123,
      "grad_norm": 1.4932096519260085e-06,
      "learning_rate": 3.9280197745526344e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6966216,
      "step": 12260
    },
    {
      "epoch": 215.1769911504425,
      "grad_norm": 5.558392217608343e-07,
      "learning_rate": 3.9272138406355495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6968904,
      "step": 12265
    },
    {
      "epoch": 215.26548672566372,
      "grad_norm": 1.5916380107228179e-06,
      "learning_rate": 3.926407686624154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6971272,
      "step": 12270
    },
    {
      "epoch": 215.35398230088495,
      "grad_norm": 2.351545163037372e-06,
      "learning_rate": 3.9256013126427684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6973976,
      "step": 12275
    },
    {
      "epoch": 215.44247787610618,
      "grad_norm": 2.0530806068563834e-06,
      "learning_rate": 3.9247947188157455e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6977736,
      "step": 12280
    },
    {
      "epoch": 215.53097345132744,
      "grad_norm": 1.1107030104540172e-06,
      "learning_rate": 3.9239879052674715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6980936,
      "step": 12285
    },
    {
      "epoch": 215.61946902654867,
      "grad_norm": 1.1306409533062833e-06,
      "learning_rate": 3.9231808721223673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6984104,
      "step": 12290
    },
    {
      "epoch": 215.7079646017699,
      "grad_norm": 2.294732666996424e-06,
      "learning_rate": 3.9223736195048886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6986696,
      "step": 12295
    },
    {
      "epoch": 215.79646017699116,
      "grad_norm": 1.2089446954632876e-06,
      "learning_rate": 3.921566147539523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6989560,
      "step": 12300
    },
    {
      "epoch": 215.8849557522124,
      "grad_norm": 7.352697366513894e-07,
      "learning_rate": 3.920758456350792e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6992344,
      "step": 12305
    },
    {
      "epoch": 215.97345132743362,
      "grad_norm": 1.6055603282438824e-06,
      "learning_rate": 3.919950546063253e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6995064,
      "step": 12310
    },
    {
      "epoch": 216.05309734513276,
      "grad_norm": 1.1441898095654324e-06,
      "learning_rate": 3.919142416801496e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 6997408,
      "step": 12315
    },
    {
      "epoch": 216.141592920354,
      "grad_norm": 9.905132856147247e-07,
      "learning_rate": 3.918334068690144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7000240,
      "step": 12320
    },
    {
      "epoch": 216.23008849557522,
      "grad_norm": 2.0332468011474703e-06,
      "learning_rate": 3.917525501853855e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7002960,
      "step": 12325
    },
    {
      "epoch": 216.31858407079645,
      "grad_norm": 2.104484565279563e-06,
      "learning_rate": 3.916716716417319e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7005824,
      "step": 12330
    },
    {
      "epoch": 216.4070796460177,
      "grad_norm": 2.507207000235212e-06,
      "learning_rate": 3.915907712505263e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7008576,
      "step": 12335
    },
    {
      "epoch": 216.49557522123894,
      "grad_norm": 1.6537200053790002e-06,
      "learning_rate": 3.915098490242444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7011344,
      "step": 12340
    },
    {
      "epoch": 216.58407079646017,
      "grad_norm": 1.1817453469120665e-06,
      "learning_rate": 3.914289049753654e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7014416,
      "step": 12345
    },
    {
      "epoch": 216.67256637168143,
      "grad_norm": 1.0228841347270645e-06,
      "learning_rate": 3.913479391163719e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7017344,
      "step": 12350
    },
    {
      "epoch": 216.76106194690266,
      "grad_norm": 1.088293970497034e-06,
      "learning_rate": 3.9126695145975e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7020400,
      "step": 12355
    },
    {
      "epoch": 216.8495575221239,
      "grad_norm": 1.7280644897255115e-06,
      "learning_rate": 3.911859420179889e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7023536,
      "step": 12360
    },
    {
      "epoch": 216.93805309734512,
      "grad_norm": 1.428527525604295e-06,
      "learning_rate": 3.911049108035813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7026192,
      "step": 12365
    },
    {
      "epoch": 217.01769911504425,
      "grad_norm": 1.4658941154266358e-06,
      "learning_rate": 3.910238578290232e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7028648,
      "step": 12370
    },
    {
      "epoch": 217.10619469026548,
      "grad_norm": 8.885523925528105e-07,
      "learning_rate": 3.90942783106814e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7031272,
      "step": 12375
    },
    {
      "epoch": 217.1946902654867,
      "grad_norm": 8.560526794099133e-07,
      "learning_rate": 3.908616866494564e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7033928,
      "step": 12380
    },
    {
      "epoch": 217.28318584070797,
      "grad_norm": 1.508458012722258e-06,
      "learning_rate": 3.907805684694566e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7036984,
      "step": 12385
    },
    {
      "epoch": 217.3716814159292,
      "grad_norm": 7.799605441505264e-07,
      "learning_rate": 3.90699428579324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7039480,
      "step": 12390
    },
    {
      "epoch": 217.46017699115043,
      "grad_norm": 1.4197958080330864e-06,
      "learning_rate": 3.906182669915713e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7042536,
      "step": 12395
    },
    {
      "epoch": 217.5486725663717,
      "grad_norm": 2.1500695766007993e-06,
      "learning_rate": 3.9053708371871476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7045688,
      "step": 12400
    },
    {
      "epoch": 217.5486725663717,
      "eval_loss": 0.45370200276374817,
      "eval_runtime": 1.0614,
      "eval_samples_per_second": 23.555,
      "eval_steps_per_second": 12.248,
      "num_input_tokens_seen": 7045688,
      "step": 12400
    },
    {
      "epoch": 217.63716814159292,
      "grad_norm": 1.1526954040164128e-06,
      "learning_rate": 3.904558787732738e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7048344,
      "step": 12405
    },
    {
      "epoch": 217.72566371681415,
      "grad_norm": 8.278893233182316e-07,
      "learning_rate": 3.9037465216777135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7051480,
      "step": 12410
    },
    {
      "epoch": 217.81415929203538,
      "grad_norm": 9.580463711245102e-07,
      "learning_rate": 3.902934039147334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7054264,
      "step": 12415
    },
    {
      "epoch": 217.90265486725664,
      "grad_norm": 1.200405336021504e-06,
      "learning_rate": 3.902121340266894e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7057272,
      "step": 12420
    },
    {
      "epoch": 217.99115044247787,
      "grad_norm": 1.0213324230790022e-06,
      "learning_rate": 3.9013084251617246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7060104,
      "step": 12425
    },
    {
      "epoch": 218.07079646017698,
      "grad_norm": 9.759406793818926e-07,
      "learning_rate": 3.9004952939571865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7062728,
      "step": 12430
    },
    {
      "epoch": 218.15929203539824,
      "grad_norm": 1.7537091707708896e-06,
      "learning_rate": 3.899681946778673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7065416,
      "step": 12435
    },
    {
      "epoch": 218.24778761061947,
      "grad_norm": 3.301629703855724e-06,
      "learning_rate": 3.898868383751615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7068200,
      "step": 12440
    },
    {
      "epoch": 218.3362831858407,
      "grad_norm": 8.903971320251003e-07,
      "learning_rate": 3.8980546050014724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7070984,
      "step": 12445
    },
    {
      "epoch": 218.42477876106196,
      "grad_norm": 1.3643176544064772e-06,
      "learning_rate": 3.897240610653741e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7074312,
      "step": 12450
    },
    {
      "epoch": 218.5132743362832,
      "grad_norm": 1.3625125347971334e-06,
      "learning_rate": 3.896426400833948e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7077048,
      "step": 12455
    },
    {
      "epoch": 218.60176991150442,
      "grad_norm": 1.0555901326370076e-06,
      "learning_rate": 3.895611975667656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7079784,
      "step": 12460
    },
    {
      "epoch": 218.69026548672565,
      "grad_norm": 1.0474443570274161e-06,
      "learning_rate": 3.8947973352804584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7083080,
      "step": 12465
    },
    {
      "epoch": 218.7787610619469,
      "grad_norm": 1.547467036289163e-06,
      "learning_rate": 3.893982479797984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7085848,
      "step": 12470
    },
    {
      "epoch": 218.86725663716814,
      "grad_norm": 4.529832040134352e-06,
      "learning_rate": 3.8931674093458926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7088760,
      "step": 12475
    },
    {
      "epoch": 218.95575221238937,
      "grad_norm": 1.016720261759474e-06,
      "learning_rate": 3.89235212404988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7091816,
      "step": 12480
    },
    {
      "epoch": 219.0353982300885,
      "grad_norm": 1.1974473181908252e-06,
      "learning_rate": 3.891536624035672e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7094424,
      "step": 12485
    },
    {
      "epoch": 219.12389380530973,
      "grad_norm": 4.6298778215714265e-06,
      "learning_rate": 3.8907209094290295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7097208,
      "step": 12490
    },
    {
      "epoch": 219.21238938053096,
      "grad_norm": 3.823111455858452e-06,
      "learning_rate": 3.8899049803557466e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7100344,
      "step": 12495
    },
    {
      "epoch": 219.30088495575222,
      "grad_norm": 8.619890650152229e-07,
      "learning_rate": 3.889088836941648e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7103320,
      "step": 12500
    },
    {
      "epoch": 219.38938053097345,
      "grad_norm": 1.5343159702752018e-06,
      "learning_rate": 3.8882724793125946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7106216,
      "step": 12505
    },
    {
      "epoch": 219.47787610619469,
      "grad_norm": 1.58143825501611e-06,
      "learning_rate": 3.8874559075944794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7109080,
      "step": 12510
    },
    {
      "epoch": 219.56637168141592,
      "grad_norm": 9.58984060162038e-07,
      "learning_rate": 3.886639121913227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7112168,
      "step": 12515
    },
    {
      "epoch": 219.65486725663717,
      "grad_norm": 2.595761316115386e-06,
      "learning_rate": 3.885822122394797e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7115144,
      "step": 12520
    },
    {
      "epoch": 219.7433628318584,
      "grad_norm": 3.2961715987767093e-06,
      "learning_rate": 3.8850049091651794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7117848,
      "step": 12525
    },
    {
      "epoch": 219.83185840707964,
      "grad_norm": 9.939969913830282e-07,
      "learning_rate": 3.8841874823504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7120664,
      "step": 12530
    },
    {
      "epoch": 219.9203539823009,
      "grad_norm": 1.2228553032400669e-06,
      "learning_rate": 3.8833698420765157e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7123208,
      "step": 12535
    },
    {
      "epoch": 220.0,
      "grad_norm": 1.4033572597327293e-06,
      "learning_rate": 3.882551988469618e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7125552,
      "step": 12540
    },
    {
      "epoch": 220.08849557522123,
      "grad_norm": 2.0061347640876193e-06,
      "learning_rate": 3.881733921655829e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7128336,
      "step": 12545
    },
    {
      "epoch": 220.1769911504425,
      "grad_norm": 3.720133236129186e-06,
      "learning_rate": 3.8809156417613054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7131072,
      "step": 12550
    },
    {
      "epoch": 220.26548672566372,
      "grad_norm": 1.9272324607300106e-06,
      "learning_rate": 3.8800971489122364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7133808,
      "step": 12555
    },
    {
      "epoch": 220.35398230088495,
      "grad_norm": 7.656121852051001e-07,
      "learning_rate": 3.8792784432348434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7137520,
      "step": 12560
    },
    {
      "epoch": 220.44247787610618,
      "grad_norm": 1.053833557307371e-06,
      "learning_rate": 3.878459524855381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7140544,
      "step": 12565
    },
    {
      "epoch": 220.53097345132744,
      "grad_norm": 1.3299652437126497e-06,
      "learning_rate": 3.8776403939001384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7143216,
      "step": 12570
    },
    {
      "epoch": 220.61946902654867,
      "grad_norm": 8.481366421619896e-06,
      "learning_rate": 3.876821050495433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7146096,
      "step": 12575
    },
    {
      "epoch": 220.7079646017699,
      "grad_norm": 2.49453250944498e-06,
      "learning_rate": 3.87600149476762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7149184,
      "step": 12580
    },
    {
      "epoch": 220.79646017699116,
      "grad_norm": 1.2164961162852705e-06,
      "learning_rate": 3.8751817268430843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7152144,
      "step": 12585
    },
    {
      "epoch": 220.8849557522124,
      "grad_norm": 5.95782751133811e-07,
      "learning_rate": 3.8743617468482464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7154704,
      "step": 12590
    },
    {
      "epoch": 220.97345132743362,
      "grad_norm": 1.5662653822801076e-06,
      "learning_rate": 3.8735415549095535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7157568,
      "step": 12595
    },
    {
      "epoch": 221.05309734513276,
      "grad_norm": 1.7904222886500065e-06,
      "learning_rate": 3.8727211511534934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7159888,
      "step": 12600
    },
    {
      "epoch": 221.05309734513276,
      "eval_loss": 0.4610222280025482,
      "eval_runtime": 1.0719,
      "eval_samples_per_second": 23.322,
      "eval_steps_per_second": 12.128,
      "num_input_tokens_seen": 7159888,
      "step": 12600
    },
    {
      "epoch": 221.141592920354,
      "grad_norm": 2.29057536671462e-06,
      "learning_rate": 3.8719005357065804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7162768,
      "step": 12605
    },
    {
      "epoch": 221.23008849557522,
      "grad_norm": 1.0342531595597393e-06,
      "learning_rate": 3.8710797086953645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7165776,
      "step": 12610
    },
    {
      "epoch": 221.31858407079645,
      "grad_norm": 1.3541480257117655e-06,
      "learning_rate": 3.870258670246427e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7168688,
      "step": 12615
    },
    {
      "epoch": 221.4070796460177,
      "grad_norm": 4.701563739217818e-06,
      "learning_rate": 3.869437420486384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7171600,
      "step": 12620
    },
    {
      "epoch": 221.49557522123894,
      "grad_norm": 5.23448591138731e-07,
      "learning_rate": 3.8686159595418805e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7174656,
      "step": 12625
    },
    {
      "epoch": 221.58407079646017,
      "grad_norm": 1.3183486089474172e-06,
      "learning_rate": 3.867794287539597e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7177648,
      "step": 12630
    },
    {
      "epoch": 221.67256637168143,
      "grad_norm": 2.4538064735679654e-06,
      "learning_rate": 3.866972404606245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7180208,
      "step": 12635
    },
    {
      "epoch": 221.76106194690266,
      "grad_norm": 1.9462545424175914e-06,
      "learning_rate": 3.866150310868571e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7183504,
      "step": 12640
    },
    {
      "epoch": 221.8495575221239,
      "grad_norm": 1.7869017483462812e-06,
      "learning_rate": 3.8653280064533506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7186512,
      "step": 12645
    },
    {
      "epoch": 221.93805309734512,
      "grad_norm": 1.6260926258837571e-06,
      "learning_rate": 3.864505491487394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7188912,
      "step": 12650
    },
    {
      "epoch": 222.01769911504425,
      "grad_norm": 7.16063709660375e-07,
      "learning_rate": 3.8636827660975414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7191504,
      "step": 12655
    },
    {
      "epoch": 222.10619469026548,
      "grad_norm": 1.0469960898262798e-06,
      "learning_rate": 3.862859830410671e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7194384,
      "step": 12660
    },
    {
      "epoch": 222.1946902654867,
      "grad_norm": 1.698362325441849e-06,
      "learning_rate": 3.862036684553688e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7196960,
      "step": 12665
    },
    {
      "epoch": 222.28318584070797,
      "grad_norm": 1.8143529132430558e-06,
      "learning_rate": 3.8612133286535314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7200176,
      "step": 12670
    },
    {
      "epoch": 222.3716814159292,
      "grad_norm": 1.1051171213694033e-06,
      "learning_rate": 3.860389762837173e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7203392,
      "step": 12675
    },
    {
      "epoch": 222.46017699115043,
      "grad_norm": 3.5600867249740986e-06,
      "learning_rate": 3.859565987231618e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7206432,
      "step": 12680
    },
    {
      "epoch": 222.5486725663717,
      "grad_norm": 6.716098823744687e-07,
      "learning_rate": 3.858742001963902e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7209296,
      "step": 12685
    },
    {
      "epoch": 222.63716814159292,
      "grad_norm": 2.5655958779680077e-06,
      "learning_rate": 3.857917807161094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7212336,
      "step": 12690
    },
    {
      "epoch": 222.72566371681415,
      "grad_norm": 8.671567002238589e-07,
      "learning_rate": 3.857093402950296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7214960,
      "step": 12695
    },
    {
      "epoch": 222.81415929203538,
      "grad_norm": 8.862593290359655e-07,
      "learning_rate": 3.8562687894586414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7217520,
      "step": 12700
    },
    {
      "epoch": 222.90265486725664,
      "grad_norm": 1.0948700719382032e-06,
      "learning_rate": 3.8554439668132946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7220528,
      "step": 12705
    },
    {
      "epoch": 222.99115044247787,
      "grad_norm": 2.7106361812911928e-06,
      "learning_rate": 3.854618935141455e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7223104,
      "step": 12710
    },
    {
      "epoch": 223.07079646017698,
      "grad_norm": 2.0341892650321824e-06,
      "learning_rate": 3.8537936945703525e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7225568,
      "step": 12715
    },
    {
      "epoch": 223.15929203539824,
      "grad_norm": 9.763657544681337e-07,
      "learning_rate": 3.852968245227249e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7228464,
      "step": 12720
    },
    {
      "epoch": 223.24778761061947,
      "grad_norm": 1.207564650940185e-06,
      "learning_rate": 3.85214258723944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7231024,
      "step": 12725
    },
    {
      "epoch": 223.3362831858407,
      "grad_norm": 1.8472103420208441e-06,
      "learning_rate": 3.8513167207342524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7233744,
      "step": 12730
    },
    {
      "epoch": 223.42477876106196,
      "grad_norm": 2.2244128103920957e-06,
      "learning_rate": 3.850490645839044e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7237136,
      "step": 12735
    },
    {
      "epoch": 223.5132743362832,
      "grad_norm": 3.980673682235647e-06,
      "learning_rate": 3.849664362681207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7240288,
      "step": 12740
    },
    {
      "epoch": 223.60176991150442,
      "grad_norm": 1.1198350193808437e-06,
      "learning_rate": 3.848837871388165e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7243232,
      "step": 12745
    },
    {
      "epoch": 223.69026548672565,
      "grad_norm": 1.8816941746990778e-06,
      "learning_rate": 3.848011172087371e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7246096,
      "step": 12750
    },
    {
      "epoch": 223.7787610619469,
      "grad_norm": 6.951423756618169e-07,
      "learning_rate": 3.847184264906315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7249152,
      "step": 12755
    },
    {
      "epoch": 223.86725663716814,
      "grad_norm": 7.893976885497977e-07,
      "learning_rate": 3.846357149972516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7252032,
      "step": 12760
    },
    {
      "epoch": 223.95575221238937,
      "grad_norm": 1.8909857999460655e-06,
      "learning_rate": 3.8455298274135246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7254384,
      "step": 12765
    },
    {
      "epoch": 224.0353982300885,
      "grad_norm": 2.492783323759795e-06,
      "learning_rate": 3.8447022973569254e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7256824,
      "step": 12770
    },
    {
      "epoch": 224.12389380530973,
      "grad_norm": 1.174142653326271e-06,
      "learning_rate": 3.843874559930332e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7260104,
      "step": 12775
    },
    {
      "epoch": 224.21238938053096,
      "grad_norm": 1.109757704398362e-06,
      "learning_rate": 3.843046615261394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7263144,
      "step": 12780
    },
    {
      "epoch": 224.30088495575222,
      "grad_norm": 2.890962150559062e-06,
      "learning_rate": 3.842218463477791e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7266136,
      "step": 12785
    },
    {
      "epoch": 224.38938053097345,
      "grad_norm": 2.5918720893969294e-06,
      "learning_rate": 3.841390104707233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7268968,
      "step": 12790
    },
    {
      "epoch": 224.47787610619469,
      "grad_norm": 3.49877996086434e-06,
      "learning_rate": 3.8405615390774643e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7271496,
      "step": 12795
    },
    {
      "epoch": 224.56637168141592,
      "grad_norm": 1.3889462024962995e-06,
      "learning_rate": 3.839732766716259e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7274296,
      "step": 12800
    },
    {
      "epoch": 224.56637168141592,
      "eval_loss": 0.4564047157764435,
      "eval_runtime": 1.0609,
      "eval_samples_per_second": 23.564,
      "eval_steps_per_second": 12.253,
      "num_input_tokens_seen": 7274296,
      "step": 12800
    },
    {
      "epoch": 224.65486725663717,
      "grad_norm": 1.3140873988959356e-06,
      "learning_rate": 3.838903787751425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7277256,
      "step": 12805
    },
    {
      "epoch": 224.7433628318584,
      "grad_norm": 5.921288561694382e-07,
      "learning_rate": 3.838074602310802e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7280152,
      "step": 12810
    },
    {
      "epoch": 224.83185840707964,
      "grad_norm": 2.0414408936630934e-06,
      "learning_rate": 3.837245210522258e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7282760,
      "step": 12815
    },
    {
      "epoch": 224.9203539823009,
      "grad_norm": 1.327536097051052e-06,
      "learning_rate": 3.8364156125136996e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7285448,
      "step": 12820
    },
    {
      "epoch": 225.0,
      "grad_norm": 2.3342781787505373e-06,
      "learning_rate": 3.835585808413059e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7288280,
      "step": 12825
    },
    {
      "epoch": 225.08849557522123,
      "grad_norm": 1.2625471299543278e-06,
      "learning_rate": 3.8347557983483024e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7291208,
      "step": 12830
    },
    {
      "epoch": 225.1769911504425,
      "grad_norm": 1.4672167480966891e-06,
      "learning_rate": 3.833925582447428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7293928,
      "step": 12835
    },
    {
      "epoch": 225.26548672566372,
      "grad_norm": 1.6703537539797253e-06,
      "learning_rate": 3.8330951608384656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7296728,
      "step": 12840
    },
    {
      "epoch": 225.35398230088495,
      "grad_norm": 1.7407169252692256e-06,
      "learning_rate": 3.832264533649477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7299848,
      "step": 12845
    },
    {
      "epoch": 225.44247787610618,
      "grad_norm": 2.278122565257945e-06,
      "learning_rate": 3.8314337010085555e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7303112,
      "step": 12850
    },
    {
      "epoch": 225.53097345132744,
      "grad_norm": 1.5312277810153319e-06,
      "learning_rate": 3.830602663043824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7305960,
      "step": 12855
    },
    {
      "epoch": 225.61946902654867,
      "grad_norm": 1.95024495042162e-06,
      "learning_rate": 3.8297714198834414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7308984,
      "step": 12860
    },
    {
      "epoch": 225.7079646017699,
      "grad_norm": 1.4022398318047635e-06,
      "learning_rate": 3.828939971655595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7311832,
      "step": 12865
    },
    {
      "epoch": 225.79646017699116,
      "grad_norm": 1.1146964880026644e-06,
      "learning_rate": 3.828108318488505e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7314552,
      "step": 12870
    },
    {
      "epoch": 225.8849557522124,
      "grad_norm": 1.1496920251374831e-06,
      "learning_rate": 3.8272764605104216e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7317736,
      "step": 12875
    },
    {
      "epoch": 225.97345132743362,
      "grad_norm": 7.203764198493445e-07,
      "learning_rate": 3.826444397849628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7320104,
      "step": 12880
    },
    {
      "epoch": 226.05309734513276,
      "grad_norm": 3.618867367549683e-06,
      "learning_rate": 3.825612130634439e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7322248,
      "step": 12885
    },
    {
      "epoch": 226.141592920354,
      "grad_norm": 1.1901518064405536e-06,
      "learning_rate": 3.824779658993202e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7325208,
      "step": 12890
    },
    {
      "epoch": 226.23008849557522,
      "grad_norm": 1.6478421684951172e-06,
      "learning_rate": 3.823946983054292e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7327992,
      "step": 12895
    },
    {
      "epoch": 226.31858407079645,
      "grad_norm": 1.42299359140452e-06,
      "learning_rate": 3.82311410294612e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7330936,
      "step": 12900
    },
    {
      "epoch": 226.4070796460177,
      "grad_norm": 1.6453803937110933e-06,
      "learning_rate": 3.822281018797127e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7333400,
      "step": 12905
    },
    {
      "epoch": 226.49557522123894,
      "grad_norm": 1.6350649048035848e-06,
      "learning_rate": 3.821447730735783e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7336056,
      "step": 12910
    },
    {
      "epoch": 226.58407079646017,
      "grad_norm": 4.3206992472732964e-07,
      "learning_rate": 3.820614238890592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7339064,
      "step": 12915
    },
    {
      "epoch": 226.67256637168143,
      "grad_norm": 1.6514557046320988e-06,
      "learning_rate": 3.819780543390091e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7342264,
      "step": 12920
    },
    {
      "epoch": 226.76106194690266,
      "grad_norm": 1.6390118844356039e-06,
      "learning_rate": 3.818946644362844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7344872,
      "step": 12925
    },
    {
      "epoch": 226.8495575221239,
      "grad_norm": 1.349856120214099e-06,
      "learning_rate": 3.81811254193745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7347752,
      "step": 12930
    },
    {
      "epoch": 226.93805309734512,
      "grad_norm": 8.673656566315913e-07,
      "learning_rate": 3.8172782362425366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7350632,
      "step": 12935
    },
    {
      "epoch": 227.01769911504425,
      "grad_norm": 9.85411702458805e-07,
      "learning_rate": 3.816443727406765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7353528,
      "step": 12940
    },
    {
      "epoch": 227.10619469026548,
      "grad_norm": 2.2378812900569756e-06,
      "learning_rate": 3.815609015558829e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7356520,
      "step": 12945
    },
    {
      "epoch": 227.1946902654867,
      "grad_norm": 9.184637406178808e-07,
      "learning_rate": 3.814774100827448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7359640,
      "step": 12950
    },
    {
      "epoch": 227.28318584070797,
      "grad_norm": 2.105520252371207e-06,
      "learning_rate": 3.813938983341379e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7362504,
      "step": 12955
    },
    {
      "epoch": 227.3716814159292,
      "grad_norm": 1.2004961718048435e-06,
      "learning_rate": 3.813103663229407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7365048,
      "step": 12960
    },
    {
      "epoch": 227.46017699115043,
      "grad_norm": 4.4627293505072885e-07,
      "learning_rate": 3.812268140620349e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7367976,
      "step": 12965
    },
    {
      "epoch": 227.5486725663717,
      "grad_norm": 1.688885845396726e-06,
      "learning_rate": 3.811432415643051e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7370664,
      "step": 12970
    },
    {
      "epoch": 227.63716814159292,
      "grad_norm": 1.4694172705276287e-06,
      "learning_rate": 3.8105964884263954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7373832,
      "step": 12975
    },
    {
      "epoch": 227.72566371681415,
      "grad_norm": 8.371094963877113e-07,
      "learning_rate": 3.809760359099291e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7376888,
      "step": 12980
    },
    {
      "epoch": 227.81415929203538,
      "grad_norm": 3.949799520341912e-06,
      "learning_rate": 3.8089240277906804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7379784,
      "step": 12985
    },
    {
      "epoch": 227.90265486725664,
      "grad_norm": 1.5554012406937545e-06,
      "learning_rate": 3.808087494629535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7382792,
      "step": 12990
    },
    {
      "epoch": 227.99115044247787,
      "grad_norm": 1.9280162177892635e-06,
      "learning_rate": 3.8072507597448595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7385400,
      "step": 12995
    },
    {
      "epoch": 228.07079646017698,
      "grad_norm": 1.2517446066340199e-06,
      "learning_rate": 3.806413823265689e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7387544,
      "step": 13000
    },
    {
      "epoch": 228.07079646017698,
      "eval_loss": 0.4593886733055115,
      "eval_runtime": 1.0616,
      "eval_samples_per_second": 23.549,
      "eval_steps_per_second": 12.246,
      "num_input_tokens_seen": 7387544,
      "step": 13000
    },
    {
      "epoch": 228.15929203539824,
      "grad_norm": 2.0330062397988513e-06,
      "learning_rate": 3.805576685321089e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7390040,
      "step": 13005
    },
    {
      "epoch": 228.24778761061947,
      "grad_norm": 9.486487329013471e-07,
      "learning_rate": 3.804739346040158e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7392728,
      "step": 13010
    },
    {
      "epoch": 228.3362831858407,
      "grad_norm": 1.6403772633566405e-06,
      "learning_rate": 3.8039018055520234e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7395432,
      "step": 13015
    },
    {
      "epoch": 228.42477876106196,
      "grad_norm": 6.240757102204952e-07,
      "learning_rate": 3.803064063985844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7398024,
      "step": 13020
    },
    {
      "epoch": 228.5132743362832,
      "grad_norm": 1.9470153347356245e-06,
      "learning_rate": 3.802226121470811e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7401032,
      "step": 13025
    },
    {
      "epoch": 228.60176991150442,
      "grad_norm": 1.3290062952364678e-06,
      "learning_rate": 3.801387978136145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7403880,
      "step": 13030
    },
    {
      "epoch": 228.69026548672565,
      "grad_norm": 3.2959926556941355e-06,
      "learning_rate": 3.800549634111099e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7407064,
      "step": 13035
    },
    {
      "epoch": 228.7787610619469,
      "grad_norm": 7.483793069695821e-07,
      "learning_rate": 3.799711089524955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7409832,
      "step": 13040
    },
    {
      "epoch": 228.86725663716814,
      "grad_norm": 1.1142831226607086e-06,
      "learning_rate": 3.7988723445070285e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7412776,
      "step": 13045
    },
    {
      "epoch": 228.95575221238937,
      "grad_norm": 8.729688829589577e-07,
      "learning_rate": 3.798033399186663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7415768,
      "step": 13050
    },
    {
      "epoch": 229.0353982300885,
      "grad_norm": 8.712609655958659e-07,
      "learning_rate": 3.797194253693237e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7418296,
      "step": 13055
    },
    {
      "epoch": 229.12389380530973,
      "grad_norm": 9.598800261301221e-07,
      "learning_rate": 3.796354908156153e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7420952,
      "step": 13060
    },
    {
      "epoch": 229.21238938053096,
      "grad_norm": 5.95958738358604e-07,
      "learning_rate": 3.795515362704853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7423912,
      "step": 13065
    },
    {
      "epoch": 229.30088495575222,
      "grad_norm": 1.2550272003863938e-06,
      "learning_rate": 3.794675617468803e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7426264,
      "step": 13070
    },
    {
      "epoch": 229.38938053097345,
      "grad_norm": 1.2438703151929076e-06,
      "learning_rate": 3.793835672577503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7429368,
      "step": 13075
    },
    {
      "epoch": 229.47787610619469,
      "grad_norm": 3.2212381029239623e-06,
      "learning_rate": 3.7929955281604826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7432120,
      "step": 13080
    },
    {
      "epoch": 229.56637168141592,
      "grad_norm": 9.337744018012017e-07,
      "learning_rate": 3.7921551843473036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7435144,
      "step": 13085
    },
    {
      "epoch": 229.65486725663717,
      "grad_norm": 1.7881644680528552e-06,
      "learning_rate": 3.791314641267557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7438040,
      "step": 13090
    },
    {
      "epoch": 229.7433628318584,
      "grad_norm": 7.279609235411044e-07,
      "learning_rate": 3.790473899050864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7441160,
      "step": 13095
    },
    {
      "epoch": 229.83185840707964,
      "grad_norm": 8.58565954331425e-07,
      "learning_rate": 3.7896329578268794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7444408,
      "step": 13100
    },
    {
      "epoch": 229.9203539823009,
      "grad_norm": 2.2501646981254453e-06,
      "learning_rate": 3.7887918177252855e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7447208,
      "step": 13105
    },
    {
      "epoch": 230.0,
      "grad_norm": 4.553703547571786e-06,
      "learning_rate": 3.787950478875798e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7449656,
      "step": 13110
    },
    {
      "epoch": 230.08849557522123,
      "grad_norm": 5.349847924662754e-07,
      "learning_rate": 3.787108941408162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7452312,
      "step": 13115
    },
    {
      "epoch": 230.1769911504425,
      "grad_norm": 1.5176960914686788e-06,
      "learning_rate": 3.786267205452151e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7454904,
      "step": 13120
    },
    {
      "epoch": 230.26548672566372,
      "grad_norm": 6.416432825062657e-07,
      "learning_rate": 3.785425271137573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7457400,
      "step": 13125
    },
    {
      "epoch": 230.35398230088495,
      "grad_norm": 1.8708827838054276e-06,
      "learning_rate": 3.7845831385942655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7460424,
      "step": 13130
    },
    {
      "epoch": 230.44247787610618,
      "grad_norm": 1.149542185885366e-06,
      "learning_rate": 3.7837408079520944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7463832,
      "step": 13135
    },
    {
      "epoch": 230.53097345132744,
      "grad_norm": 1.5197947504930198e-06,
      "learning_rate": 3.782898279340957e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7466520,
      "step": 13140
    },
    {
      "epoch": 230.61946902654867,
      "grad_norm": 2.6903576326731127e-06,
      "learning_rate": 3.782055552890784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7469576,
      "step": 13145
    },
    {
      "epoch": 230.7079646017699,
      "grad_norm": 9.448323226024513e-07,
      "learning_rate": 3.781212628731534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7472120,
      "step": 13150
    },
    {
      "epoch": 230.79646017699116,
      "grad_norm": 8.864692517818185e-07,
      "learning_rate": 3.7803695069931946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7475128,
      "step": 13155
    },
    {
      "epoch": 230.8849557522124,
      "grad_norm": 4.4142026922600053e-07,
      "learning_rate": 3.779526187805789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7478616,
      "step": 13160
    },
    {
      "epoch": 230.97345132743362,
      "grad_norm": 6.368143203872023e-07,
      "learning_rate": 3.778682671299364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7481336,
      "step": 13165
    },
    {
      "epoch": 231.05309734513276,
      "grad_norm": 1.7829045191319892e-06,
      "learning_rate": 3.777838957604003e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7483656,
      "step": 13170
    },
    {
      "epoch": 231.141592920354,
      "grad_norm": 6.451006129282177e-07,
      "learning_rate": 3.776995046849816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7486552,
      "step": 13175
    },
    {
      "epoch": 231.23008849557522,
      "grad_norm": 5.609809363704699e-07,
      "learning_rate": 3.776150939166945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7489160,
      "step": 13180
    },
    {
      "epoch": 231.31858407079645,
      "grad_norm": 7.999739182196208e-07,
      "learning_rate": 3.775306634685562e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7492088,
      "step": 13185
    },
    {
      "epoch": 231.4070796460177,
      "grad_norm": 1.252972538168251e-06,
      "learning_rate": 3.7744621335358696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7494856,
      "step": 13190
    },
    {
      "epoch": 231.49557522123894,
      "grad_norm": 2.059890903183259e-06,
      "learning_rate": 3.7736174358481e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7497224,
      "step": 13195
    },
    {
      "epoch": 231.58407079646017,
      "grad_norm": 6.952368494239636e-07,
      "learning_rate": 3.7727725417525175e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7500200,
      "step": 13200
    },
    {
      "epoch": 231.58407079646017,
      "eval_loss": 0.46606552600860596,
      "eval_runtime": 1.0644,
      "eval_samples_per_second": 23.488,
      "eval_steps_per_second": 12.214,
      "num_input_tokens_seen": 7500200,
      "step": 13200
    },
    {
      "epoch": 231.67256637168143,
      "grad_norm": 1.1395616184017854e-06,
      "learning_rate": 3.771927451379414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7503064,
      "step": 13205
    },
    {
      "epoch": 231.76106194690266,
      "grad_norm": 2.8364304398564855e-06,
      "learning_rate": 3.7710821648591135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7506248,
      "step": 13210
    },
    {
      "epoch": 231.8495575221239,
      "grad_norm": 1.7702395780361257e-06,
      "learning_rate": 3.7702366823219694e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7509816,
      "step": 13215
    },
    {
      "epoch": 231.93805309734512,
      "grad_norm": 1.0474377631908283e-06,
      "learning_rate": 3.769391003898366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7512536,
      "step": 13220
    },
    {
      "epoch": 232.01769911504425,
      "grad_norm": 1.5490763871639501e-06,
      "learning_rate": 3.768545129718718e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7514808,
      "step": 13225
    },
    {
      "epoch": 232.10619469026548,
      "grad_norm": 6.559849339282664e-07,
      "learning_rate": 3.7676990599134686e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7517448,
      "step": 13230
    },
    {
      "epoch": 232.1946902654867,
      "grad_norm": 1.5119252338990918e-06,
      "learning_rate": 3.766852794613095e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7520392,
      "step": 13235
    },
    {
      "epoch": 232.28318584070797,
      "grad_norm": 2.3873317331890576e-06,
      "learning_rate": 3.766006333948099e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7523544,
      "step": 13240
    },
    {
      "epoch": 232.3716814159292,
      "grad_norm": 8.506002586727845e-07,
      "learning_rate": 3.765159678049017e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7526312,
      "step": 13245
    },
    {
      "epoch": 232.46017699115043,
      "grad_norm": 8.159379376593279e-07,
      "learning_rate": 3.7643128270464134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7529512,
      "step": 13250
    },
    {
      "epoch": 232.5486725663717,
      "grad_norm": 1.7127646287917742e-06,
      "learning_rate": 3.763465781070884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7532680,
      "step": 13255
    },
    {
      "epoch": 232.63716814159292,
      "grad_norm": 1.7919651327247266e-06,
      "learning_rate": 3.762618540253052e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7535384,
      "step": 13260
    },
    {
      "epoch": 232.72566371681415,
      "grad_norm": 2.1106623080413556e-06,
      "learning_rate": 3.761771104723576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7538376,
      "step": 13265
    },
    {
      "epoch": 232.81415929203538,
      "grad_norm": 1.3715816749026999e-06,
      "learning_rate": 3.7609234746131386e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7540728,
      "step": 13270
    },
    {
      "epoch": 232.90265486725664,
      "grad_norm": 5.768422965957143e-07,
      "learning_rate": 3.7600756500524556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7543848,
      "step": 13275
    },
    {
      "epoch": 232.99115044247787,
      "grad_norm": 2.5621950499044033e-06,
      "learning_rate": 3.759227631172271e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7546680,
      "step": 13280
    },
    {
      "epoch": 233.07079646017698,
      "grad_norm": 5.32456681412441e-07,
      "learning_rate": 3.758379418103363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7548792,
      "step": 13285
    },
    {
      "epoch": 233.15929203539824,
      "grad_norm": 1.2628332797248731e-06,
      "learning_rate": 3.757531010976534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7552184,
      "step": 13290
    },
    {
      "epoch": 233.24778761061947,
      "grad_norm": 1.366167339256208e-06,
      "learning_rate": 3.75668240992262e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7555480,
      "step": 13295
    },
    {
      "epoch": 233.3362831858407,
      "grad_norm": 2.5514993922115536e-06,
      "learning_rate": 3.7558336150724865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7558584,
      "step": 13300
    },
    {
      "epoch": 233.42477876106196,
      "grad_norm": 1.81331881776714e-06,
      "learning_rate": 3.754984626557028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7561160,
      "step": 13305
    },
    {
      "epoch": 233.5132743362832,
      "grad_norm": 8.063354357545904e-07,
      "learning_rate": 3.754135444507168e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7563464,
      "step": 13310
    },
    {
      "epoch": 233.60176991150442,
      "grad_norm": 5.4141492000781e-07,
      "learning_rate": 3.753286069053863e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7566280,
      "step": 13315
    },
    {
      "epoch": 233.69026548672565,
      "grad_norm": 1.5299043525374145e-06,
      "learning_rate": 3.7524365003280945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7569448,
      "step": 13320
    },
    {
      "epoch": 233.7787610619469,
      "grad_norm": 1.069998461389332e-06,
      "learning_rate": 3.75158673846088e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7572344,
      "step": 13325
    },
    {
      "epoch": 233.86725663716814,
      "grad_norm": 1.5625962532794802e-06,
      "learning_rate": 3.750736783583262e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7575016,
      "step": 13330
    },
    {
      "epoch": 233.95575221238937,
      "grad_norm": 1.7432522554372554e-06,
      "learning_rate": 3.7498866358263144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7577656,
      "step": 13335
    },
    {
      "epoch": 234.0353982300885,
      "grad_norm": 7.397883337034727e-07,
      "learning_rate": 3.74903629532114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7580016,
      "step": 13340
    },
    {
      "epoch": 234.12389380530973,
      "grad_norm": 5.88808404700103e-07,
      "learning_rate": 3.748185762198873e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7583104,
      "step": 13345
    },
    {
      "epoch": 234.21238938053096,
      "grad_norm": 2.079188561765477e-06,
      "learning_rate": 3.747335036590676e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7585952,
      "step": 13350
    },
    {
      "epoch": 234.30088495575222,
      "grad_norm": 8.279375265374256e-07,
      "learning_rate": 3.7464841186277405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7588480,
      "step": 13355
    },
    {
      "epoch": 234.38938053097345,
      "grad_norm": 1.0424375886941561e-06,
      "learning_rate": 3.7456330084412896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7591248,
      "step": 13360
    },
    {
      "epoch": 234.47787610619469,
      "grad_norm": 3.999244881924824e-07,
      "learning_rate": 3.744781706162576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7594544,
      "step": 13365
    },
    {
      "epoch": 234.56637168141592,
      "grad_norm": 4.603709840012016e-07,
      "learning_rate": 3.743930211922879e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7597600,
      "step": 13370
    },
    {
      "epoch": 234.65486725663717,
      "grad_norm": 8.864686833476298e-07,
      "learning_rate": 3.743078525853513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7600336,
      "step": 13375
    },
    {
      "epoch": 234.7433628318584,
      "grad_norm": 3.1023826068121707e-06,
      "learning_rate": 3.7422266480858154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7603536,
      "step": 13380
    },
    {
      "epoch": 234.83185840707964,
      "grad_norm": 6.540820322697982e-06,
      "learning_rate": 3.741374578751158e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7606048,
      "step": 13385
    },
    {
      "epoch": 234.9203539823009,
      "grad_norm": 1.9841777429974172e-06,
      "learning_rate": 3.740522317980941e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7609008,
      "step": 13390
    },
    {
      "epoch": 235.0,
      "grad_norm": 9.829376722336747e-06,
      "learning_rate": 3.739669865906593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7611336,
      "step": 13395
    },
    {
      "epoch": 235.08849557522123,
      "grad_norm": 1.7049720781869837e-06,
      "learning_rate": 3.738817222659573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7614696,
      "step": 13400
    },
    {
      "epoch": 235.08849557522123,
      "eval_loss": 0.4695309102535248,
      "eval_runtime": 1.0724,
      "eval_samples_per_second": 23.312,
      "eval_steps_per_second": 12.122,
      "num_input_tokens_seen": 7614696,
      "step": 13400
    },
    {
      "epoch": 235.1769911504425,
      "grad_norm": 1.4250153981265612e-06,
      "learning_rate": 3.73796438837137e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7617704,
      "step": 13405
    },
    {
      "epoch": 235.26548672566372,
      "grad_norm": 1.208253934237291e-06,
      "learning_rate": 3.7371113631735e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7620856,
      "step": 13410
    },
    {
      "epoch": 235.35398230088495,
      "grad_norm": 3.974824096530938e-07,
      "learning_rate": 3.736258147197512e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7623880,
      "step": 13415
    },
    {
      "epoch": 235.44247787610618,
      "grad_norm": 7.27597239347233e-07,
      "learning_rate": 3.735404740574981e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7626696,
      "step": 13420
    },
    {
      "epoch": 235.53097345132744,
      "grad_norm": 5.276070282889123e-07,
      "learning_rate": 3.7345511434375145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7629208,
      "step": 13425
    },
    {
      "epoch": 235.61946902654867,
      "grad_norm": 1.9327080735820346e-06,
      "learning_rate": 3.733697355916748e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7631656,
      "step": 13430
    },
    {
      "epoch": 235.7079646017699,
      "grad_norm": 1.7100727518482017e-06,
      "learning_rate": 3.732843378144345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7634376,
      "step": 13435
    },
    {
      "epoch": 235.79646017699116,
      "grad_norm": 1.009639277071983e-06,
      "learning_rate": 3.7319892102519995e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7637336,
      "step": 13440
    },
    {
      "epoch": 235.8849557522124,
      "grad_norm": 6.85647080445051e-07,
      "learning_rate": 3.731134852371436e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7640328,
      "step": 13445
    },
    {
      "epoch": 235.97345132743362,
      "grad_norm": 9.770242286322173e-07,
      "learning_rate": 3.730280304634408e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7643384,
      "step": 13450
    },
    {
      "epoch": 236.05309734513276,
      "grad_norm": 9.418139370609424e-07,
      "learning_rate": 3.729425567172696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7645784,
      "step": 13455
    },
    {
      "epoch": 236.141592920354,
      "grad_norm": 4.942175564792706e-06,
      "learning_rate": 3.728570640118111e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7649064,
      "step": 13460
    },
    {
      "epoch": 236.23008849557522,
      "grad_norm": 1.2525943020591512e-06,
      "learning_rate": 3.727715523602494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7651896,
      "step": 13465
    },
    {
      "epoch": 236.31858407079645,
      "grad_norm": 7.843142384444945e-07,
      "learning_rate": 3.726860217757715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7655032,
      "step": 13470
    },
    {
      "epoch": 236.4070796460177,
      "grad_norm": 1.3217191963121877e-06,
      "learning_rate": 3.726004722715673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7657464,
      "step": 13475
    },
    {
      "epoch": 236.49557522123894,
      "grad_norm": 1.1980602039329824e-06,
      "learning_rate": 3.725149038608296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7660840,
      "step": 13480
    },
    {
      "epoch": 236.58407079646017,
      "grad_norm": 1.2687361277130549e-06,
      "learning_rate": 3.7242931655675404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7663400,
      "step": 13485
    },
    {
      "epoch": 236.67256637168143,
      "grad_norm": 4.2959609913850727e-07,
      "learning_rate": 3.7234371037253937e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7666120,
      "step": 13490
    },
    {
      "epoch": 236.76106194690266,
      "grad_norm": 7.72847101870866e-07,
      "learning_rate": 3.7225808532138705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7668920,
      "step": 13495
    },
    {
      "epoch": 236.8495575221239,
      "grad_norm": 1.5720609098934801e-06,
      "learning_rate": 3.721724414165016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7671784,
      "step": 13500
    },
    {
      "epoch": 236.93805309734512,
      "grad_norm": 9.035372272592213e-07,
      "learning_rate": 3.720867786710904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7674408,
      "step": 13505
    },
    {
      "epoch": 237.01769911504425,
      "grad_norm": 8.153015187417623e-07,
      "learning_rate": 3.7200109709836366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7676832,
      "step": 13510
    },
    {
      "epoch": 237.10619469026548,
      "grad_norm": 1.4160083310343907e-06,
      "learning_rate": 3.7191539671153465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7679792,
      "step": 13515
    },
    {
      "epoch": 237.1946902654867,
      "grad_norm": 7.76658964696253e-07,
      "learning_rate": 3.718296775238193e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7682752,
      "step": 13520
    },
    {
      "epoch": 237.28318584070797,
      "grad_norm": 8.443180377071258e-07,
      "learning_rate": 3.7174393954843675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7685760,
      "step": 13525
    },
    {
      "epoch": 237.3716814159292,
      "grad_norm": 8.516114462508995e-07,
      "learning_rate": 3.716581827986087e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7689168,
      "step": 13530
    },
    {
      "epoch": 237.46017699115043,
      "grad_norm": 1.2517556342572789e-06,
      "learning_rate": 3.7157240728756004e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7691856,
      "step": 13535
    },
    {
      "epoch": 237.5486725663717,
      "grad_norm": 9.093058679354726e-07,
      "learning_rate": 3.714866130285184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7694384,
      "step": 13540
    },
    {
      "epoch": 237.63716814159292,
      "grad_norm": 1.9189035356248496e-06,
      "learning_rate": 3.714008000347143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7697440,
      "step": 13545
    },
    {
      "epoch": 237.72566371681415,
      "grad_norm": 1.055081156664528e-06,
      "learning_rate": 3.7131496831938126e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7700336,
      "step": 13550
    },
    {
      "epoch": 237.81415929203538,
      "grad_norm": 1.7655921737969038e-06,
      "learning_rate": 3.7122911789575565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7703616,
      "step": 13555
    },
    {
      "epoch": 237.90265486725664,
      "grad_norm": 2.162460077670403e-06,
      "learning_rate": 3.711432487770765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7706144,
      "step": 13560
    },
    {
      "epoch": 237.99115044247787,
      "grad_norm": 1.8784912754199468e-06,
      "learning_rate": 3.710573609765861e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7708704,
      "step": 13565
    },
    {
      "epoch": 238.07079646017698,
      "grad_norm": 1.0358459121562191e-06,
      "learning_rate": 3.709714545075292e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7710984,
      "step": 13570
    },
    {
      "epoch": 238.15929203539824,
      "grad_norm": 8.94805737061688e-07,
      "learning_rate": 3.708855293831538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7713608,
      "step": 13575
    },
    {
      "epoch": 238.24778761061947,
      "grad_norm": 2.01993452719762e-06,
      "learning_rate": 3.707995856167107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7716488,
      "step": 13580
    },
    {
      "epoch": 238.3362831858407,
      "grad_norm": 5.744242912442132e-07,
      "learning_rate": 3.707136232214534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7719640,
      "step": 13585
    },
    {
      "epoch": 238.42477876106196,
      "grad_norm": 7.58659950861329e-07,
      "learning_rate": 3.7062764221063844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7722200,
      "step": 13590
    },
    {
      "epoch": 238.5132743362832,
      "grad_norm": 5.216949148234562e-07,
      "learning_rate": 3.705416425975252e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7724840,
      "step": 13595
    },
    {
      "epoch": 238.60176991150442,
      "grad_norm": 1.484738845647371e-06,
      "learning_rate": 3.704556243953758e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7727608,
      "step": 13600
    },
    {
      "epoch": 238.60176991150442,
      "eval_loss": 0.4754555821418762,
      "eval_runtime": 1.0688,
      "eval_samples_per_second": 23.391,
      "eval_steps_per_second": 12.163,
      "num_input_tokens_seen": 7727608,
      "step": 13600
    },
    {
      "epoch": 238.69026548672565,
      "grad_norm": 2.0921884242852684e-06,
      "learning_rate": 3.7036958761745535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7730376,
      "step": 13605
    },
    {
      "epoch": 238.7787610619469,
      "grad_norm": 9.036306778398284e-07,
      "learning_rate": 3.702835322770318e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7733128,
      "step": 13610
    },
    {
      "epoch": 238.86725663716814,
      "grad_norm": 1.790496412468201e-06,
      "learning_rate": 3.701974583873761e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7736056,
      "step": 13615
    },
    {
      "epoch": 238.95575221238937,
      "grad_norm": 3.3417614986319677e-07,
      "learning_rate": 3.701113659617618e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7739048,
      "step": 13620
    },
    {
      "epoch": 239.0353982300885,
      "grad_norm": 1.7870515875983983e-06,
      "learning_rate": 3.7002525501346535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7741696,
      "step": 13625
    },
    {
      "epoch": 239.12389380530973,
      "grad_norm": 9.627401595935225e-07,
      "learning_rate": 3.699391255557664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7744384,
      "step": 13630
    },
    {
      "epoch": 239.21238938053096,
      "grad_norm": 1.4462339095189236e-06,
      "learning_rate": 3.69852977601947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7747776,
      "step": 13635
    },
    {
      "epoch": 239.30088495575222,
      "grad_norm": 1.5020747241578647e-06,
      "learning_rate": 3.697668111652922e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7750400,
      "step": 13640
    },
    {
      "epoch": 239.38938053097345,
      "grad_norm": 2.0450963802431943e-06,
      "learning_rate": 3.6968062625909005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7753376,
      "step": 13645
    },
    {
      "epoch": 239.47787610619469,
      "grad_norm": 4.251244831721124e-07,
      "learning_rate": 3.6959442289663135e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7756048,
      "step": 13650
    },
    {
      "epoch": 239.56637168141592,
      "grad_norm": 6.239287131393212e-07,
      "learning_rate": 3.695082010912098e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7758720,
      "step": 13655
    },
    {
      "epoch": 239.65486725663717,
      "grad_norm": 3.5315829904902785e-07,
      "learning_rate": 3.694219608561217e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7762096,
      "step": 13660
    },
    {
      "epoch": 239.7433628318584,
      "grad_norm": 3.052698048122693e-06,
      "learning_rate": 3.693357022046665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7764944,
      "step": 13665
    },
    {
      "epoch": 239.83185840707964,
      "grad_norm": 2.000187123485375e-06,
      "learning_rate": 3.6924942515014644e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7767696,
      "step": 13670
    },
    {
      "epoch": 239.9203539823009,
      "grad_norm": 7.314051231332996e-07,
      "learning_rate": 3.691631297058664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7770096,
      "step": 13675
    },
    {
      "epoch": 240.0,
      "grad_norm": 3.7262518048919446e-07,
      "learning_rate": 3.6907681588513424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7772536,
      "step": 13680
    },
    {
      "epoch": 240.08849557522123,
      "grad_norm": 2.231989356005215e-06,
      "learning_rate": 3.689904837012606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7775784,
      "step": 13685
    },
    {
      "epoch": 240.1769911504425,
      "grad_norm": 7.773019206069876e-07,
      "learning_rate": 3.689041331675591e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7778312,
      "step": 13690
    },
    {
      "epoch": 240.26548672566372,
      "grad_norm": 7.45084548725572e-07,
      "learning_rate": 3.688177642973461e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7781112,
      "step": 13695
    },
    {
      "epoch": 240.35398230088495,
      "grad_norm": 2.243640437882277e-06,
      "learning_rate": 3.687313771039406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7783944,
      "step": 13700
    },
    {
      "epoch": 240.44247787610618,
      "grad_norm": 1.165514390777389e-06,
      "learning_rate": 3.686449716006647e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7787192,
      "step": 13705
    },
    {
      "epoch": 240.53097345132744,
      "grad_norm": 2.5552262741257437e-06,
      "learning_rate": 3.685585478008432e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7790280,
      "step": 13710
    },
    {
      "epoch": 240.61946902654867,
      "grad_norm": 5.351683398657769e-07,
      "learning_rate": 3.6847210571780364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7793176,
      "step": 13715
    },
    {
      "epoch": 240.7079646017699,
      "grad_norm": 9.337817346022348e-07,
      "learning_rate": 3.683856453648767e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7796072,
      "step": 13720
    },
    {
      "epoch": 240.79646017699116,
      "grad_norm": 6.346003829094116e-07,
      "learning_rate": 3.682991667553954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7798968,
      "step": 13725
    },
    {
      "epoch": 240.8849557522124,
      "grad_norm": 6.370779601638787e-07,
      "learning_rate": 3.6821266990269606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7801640,
      "step": 13730
    },
    {
      "epoch": 240.97345132743362,
      "grad_norm": 1.4618060504290042e-06,
      "learning_rate": 3.681261548201174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7804472,
      "step": 13735
    },
    {
      "epoch": 241.05309734513276,
      "grad_norm": 2.7856378892465727e-06,
      "learning_rate": 3.6803962152100125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7806696,
      "step": 13740
    },
    {
      "epoch": 241.141592920354,
      "grad_norm": 1.264136471945676e-06,
      "learning_rate": 3.67953070018692e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7809304,
      "step": 13745
    },
    {
      "epoch": 241.23008849557522,
      "grad_norm": 1.097200765798334e-06,
      "learning_rate": 3.678665003265371e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7811912,
      "step": 13750
    },
    {
      "epoch": 241.31858407079645,
      "grad_norm": 1.1287295365036698e-06,
      "learning_rate": 3.677799124578867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7815256,
      "step": 13755
    },
    {
      "epoch": 241.4070796460177,
      "grad_norm": 5.343149496184196e-07,
      "learning_rate": 3.676933064260937e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7818152,
      "step": 13760
    },
    {
      "epoch": 241.49557522123894,
      "grad_norm": 1.17719946501893e-06,
      "learning_rate": 3.6760668224451365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7821752,
      "step": 13765
    },
    {
      "epoch": 241.58407079646017,
      "grad_norm": 1.146703993981646e-06,
      "learning_rate": 3.675200399265054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7824344,
      "step": 13770
    },
    {
      "epoch": 241.67256637168143,
      "grad_norm": 1.4527934126817854e-06,
      "learning_rate": 3.6743337948543014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7827464,
      "step": 13775
    },
    {
      "epoch": 241.76106194690266,
      "grad_norm": 9.922717936206027e-07,
      "learning_rate": 3.6734670093465204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7830376,
      "step": 13780
    },
    {
      "epoch": 241.8495575221239,
      "grad_norm": 8.761279559621471e-07,
      "learning_rate": 3.672600042875379e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7832856,
      "step": 13785
    },
    {
      "epoch": 241.93805309734512,
      "grad_norm": 8.520059964212123e-07,
      "learning_rate": 3.671732895574575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7835800,
      "step": 13790
    },
    {
      "epoch": 242.01769911504425,
      "grad_norm": 1.7427056491214898e-06,
      "learning_rate": 3.670865567577834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7838072,
      "step": 13795
    },
    {
      "epoch": 242.10619469026548,
      "grad_norm": 3.8630065546385595e-07,
      "learning_rate": 3.669998059018909e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7840696,
      "step": 13800
    },
    {
      "epoch": 242.10619469026548,
      "eval_loss": 0.48372510075569153,
      "eval_runtime": 1.0563,
      "eval_samples_per_second": 23.668,
      "eval_steps_per_second": 12.308,
      "num_input_tokens_seen": 7840696,
      "step": 13800
    },
    {
      "epoch": 242.1946902654867,
      "grad_norm": 8.218935931836313e-07,
      "learning_rate": 3.6691303700315796e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7843656,
      "step": 13805
    },
    {
      "epoch": 242.28318584070797,
      "grad_norm": 9.043690170074115e-07,
      "learning_rate": 3.668262500749655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7846568,
      "step": 13810
    },
    {
      "epoch": 242.3716814159292,
      "grad_norm": 8.284820296466933e-07,
      "learning_rate": 3.667394451306971e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7849208,
      "step": 13815
    },
    {
      "epoch": 242.46017699115043,
      "grad_norm": 1.0312234053344582e-06,
      "learning_rate": 3.666526221837393e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7851832,
      "step": 13820
    },
    {
      "epoch": 242.5486725663717,
      "grad_norm": 1.5500837662330014e-06,
      "learning_rate": 3.665657812474812e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7854456,
      "step": 13825
    },
    {
      "epoch": 242.63716814159292,
      "grad_norm": 6.761245572306507e-07,
      "learning_rate": 3.664789223353147e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7857464,
      "step": 13830
    },
    {
      "epoch": 242.72566371681415,
      "grad_norm": 2.186584652008605e-06,
      "learning_rate": 3.663920454606347e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7860760,
      "step": 13835
    },
    {
      "epoch": 242.81415929203538,
      "grad_norm": 8.08399590823683e-07,
      "learning_rate": 3.6630515063683856e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7863448,
      "step": 13840
    },
    {
      "epoch": 242.90265486725664,
      "grad_norm": 2.128546839230694e-06,
      "learning_rate": 3.662182378773267e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7866360,
      "step": 13845
    },
    {
      "epoch": 242.99115044247787,
      "grad_norm": 1.2812089380531688e-06,
      "learning_rate": 3.66131307195502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7870088,
      "step": 13850
    },
    {
      "epoch": 243.07079646017698,
      "grad_norm": 1.511199911874428e-06,
      "learning_rate": 3.6604435860477034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7872152,
      "step": 13855
    },
    {
      "epoch": 243.15929203539824,
      "grad_norm": 4.865718210567138e-07,
      "learning_rate": 3.6595739211854025e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7874824,
      "step": 13860
    },
    {
      "epoch": 243.24778761061947,
      "grad_norm": 1.5103654504855513e-06,
      "learning_rate": 3.658704077502231e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7877576,
      "step": 13865
    },
    {
      "epoch": 243.3362831858407,
      "grad_norm": 7.806640383023478e-07,
      "learning_rate": 3.65783405513233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7880216,
      "step": 13870
    },
    {
      "epoch": 243.42477876106196,
      "grad_norm": 1.4441063740378013e-06,
      "learning_rate": 3.656963854209867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7883096,
      "step": 13875
    },
    {
      "epoch": 243.5132743362832,
      "grad_norm": 1.0309302069799742e-06,
      "learning_rate": 3.656093474869038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7886120,
      "step": 13880
    },
    {
      "epoch": 243.60176991150442,
      "grad_norm": 1.0106649597219075e-06,
      "learning_rate": 3.655222917244068e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7889000,
      "step": 13885
    },
    {
      "epoch": 243.69026548672565,
      "grad_norm": 6.942470918147592e-07,
      "learning_rate": 3.6543521814692054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7892168,
      "step": 13890
    },
    {
      "epoch": 243.7787610619469,
      "grad_norm": 8.736521976970835e-07,
      "learning_rate": 3.653481267678731e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7895192,
      "step": 13895
    },
    {
      "epoch": 243.86725663716814,
      "grad_norm": 1.3661533557751682e-06,
      "learning_rate": 3.652610176006949e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7898216,
      "step": 13900
    },
    {
      "epoch": 243.95575221238937,
      "grad_norm": 1.4953864138078643e-06,
      "learning_rate": 3.6517389065881925e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7901288,
      "step": 13905
    },
    {
      "epoch": 244.0353982300885,
      "grad_norm": 2.2591432298213476e-06,
      "learning_rate": 3.650867459556824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7903720,
      "step": 13910
    },
    {
      "epoch": 244.12389380530973,
      "grad_norm": 7.89429861924873e-07,
      "learning_rate": 3.64999583504723e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7906184,
      "step": 13915
    },
    {
      "epoch": 244.21238938053096,
      "grad_norm": 4.4204679738868435e-07,
      "learning_rate": 3.649124033193827e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7908888,
      "step": 13920
    },
    {
      "epoch": 244.30088495575222,
      "grad_norm": 1.998054585783393e-06,
      "learning_rate": 3.648252054131057e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7911736,
      "step": 13925
    },
    {
      "epoch": 244.38938053097345,
      "grad_norm": 8.709384928806685e-07,
      "learning_rate": 3.647379897993391e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7914840,
      "step": 13930
    },
    {
      "epoch": 244.47787610619469,
      "grad_norm": 1.2018890629406087e-06,
      "learning_rate": 3.646507564915325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7917976,
      "step": 13935
    },
    {
      "epoch": 244.56637168141592,
      "grad_norm": 1.974688302652794e-06,
      "learning_rate": 3.645635055031385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7920792,
      "step": 13940
    },
    {
      "epoch": 244.65486725663717,
      "grad_norm": 1.1920797078346368e-06,
      "learning_rate": 3.6447623684761224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7923960,
      "step": 13945
    },
    {
      "epoch": 244.7433628318584,
      "grad_norm": 8.968785891738662e-07,
      "learning_rate": 3.643889505384117e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7926728,
      "step": 13950
    },
    {
      "epoch": 244.83185840707964,
      "grad_norm": 1.3345052138902247e-06,
      "learning_rate": 3.6430164658899744e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7929432,
      "step": 13955
    },
    {
      "epoch": 244.9203539823009,
      "grad_norm": 4.160708158451598e-06,
      "learning_rate": 3.642143250128329e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7932200,
      "step": 13960
    },
    {
      "epoch": 245.0,
      "grad_norm": 1.4449606169364415e-06,
      "learning_rate": 3.641269858233841e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7934424,
      "step": 13965
    },
    {
      "epoch": 245.08849557522123,
      "grad_norm": 6.896372042319854e-07,
      "learning_rate": 3.640396290341199e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7937624,
      "step": 13970
    },
    {
      "epoch": 245.1769911504425,
      "grad_norm": 8.166101679307758e-07,
      "learning_rate": 3.639522546585118e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7940504,
      "step": 13975
    },
    {
      "epoch": 245.26548672566372,
      "grad_norm": 2.248408009108971e-06,
      "learning_rate": 3.6386486271003404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7943528,
      "step": 13980
    },
    {
      "epoch": 245.35398230088495,
      "grad_norm": 1.2323736200414714e-06,
      "learning_rate": 3.6377745320216346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7946728,
      "step": 13985
    },
    {
      "epoch": 245.44247787610618,
      "grad_norm": 6.479905891865201e-07,
      "learning_rate": 3.636900261483798e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7949480,
      "step": 13990
    },
    {
      "epoch": 245.53097345132744,
      "grad_norm": 9.177971378449001e-07,
      "learning_rate": 3.636025815621654e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7952056,
      "step": 13995
    },
    {
      "epoch": 245.61946902654867,
      "grad_norm": 2.7243784188613063e-06,
      "learning_rate": 3.635151194570054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7954632,
      "step": 14000
    },
    {
      "epoch": 245.61946902654867,
      "eval_loss": 0.4702353775501251,
      "eval_runtime": 1.0608,
      "eval_samples_per_second": 23.566,
      "eval_steps_per_second": 12.255,
      "num_input_tokens_seen": 7954632,
      "step": 14000
    },
    {
      "epoch": 245.7079646017699,
      "grad_norm": 9.929890438797884e-07,
      "learning_rate": 3.634276398463873e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7957656,
      "step": 14005
    },
    {
      "epoch": 245.79646017699116,
      "grad_norm": 1.1783135960286018e-06,
      "learning_rate": 3.633401427438018e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7960280,
      "step": 14010
    },
    {
      "epoch": 245.8849557522124,
      "grad_norm": 2.073913492495194e-06,
      "learning_rate": 3.63252628162742e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7962648,
      "step": 14015
    },
    {
      "epoch": 245.97345132743362,
      "grad_norm": 8.969759619503748e-07,
      "learning_rate": 3.6316509611670364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7965832,
      "step": 14020
    },
    {
      "epoch": 246.05309734513276,
      "grad_norm": 6.094156219660363e-07,
      "learning_rate": 3.630775466191854e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7968384,
      "step": 14025
    },
    {
      "epoch": 246.141592920354,
      "grad_norm": 5.703475380869349e-07,
      "learning_rate": 3.629899796836884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7971264,
      "step": 14030
    },
    {
      "epoch": 246.23008849557522,
      "grad_norm": 8.189538220904069e-07,
      "learning_rate": 3.6290239532371666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7974224,
      "step": 14035
    },
    {
      "epoch": 246.31858407079645,
      "grad_norm": 9.654517043600208e-07,
      "learning_rate": 3.628147935527767e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7977392,
      "step": 14040
    },
    {
      "epoch": 246.4070796460177,
      "grad_norm": 1.2764477332893875e-06,
      "learning_rate": 3.627271743843779e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7980240,
      "step": 14045
    },
    {
      "epoch": 246.49557522123894,
      "grad_norm": 1.0348113619329524e-06,
      "learning_rate": 3.626395378320321e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7982688,
      "step": 14050
    },
    {
      "epoch": 246.58407079646017,
      "grad_norm": 1.9860826796502806e-06,
      "learning_rate": 3.625518839092541e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7985552,
      "step": 14055
    },
    {
      "epoch": 246.67256637168143,
      "grad_norm": 6.954404057069041e-07,
      "learning_rate": 3.624642126295612e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7988720,
      "step": 14060
    },
    {
      "epoch": 246.76106194690266,
      "grad_norm": 5.007534582546214e-07,
      "learning_rate": 3.6237652400647345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7991920,
      "step": 14065
    },
    {
      "epoch": 246.8495575221239,
      "grad_norm": 6.215942107701267e-07,
      "learning_rate": 3.622888180535134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7994512,
      "step": 14070
    },
    {
      "epoch": 246.93805309734512,
      "grad_norm": 2.3752693323331187e-06,
      "learning_rate": 3.6220109478420655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7997504,
      "step": 14075
    },
    {
      "epoch": 247.01769911504425,
      "grad_norm": 1.7092156667786185e-06,
      "learning_rate": 3.6211335421208084e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 7999840,
      "step": 14080
    },
    {
      "epoch": 247.10619469026548,
      "grad_norm": 6.269997925301141e-07,
      "learning_rate": 3.62025596350667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8002880,
      "step": 14085
    },
    {
      "epoch": 247.1946902654867,
      "grad_norm": 6.494014428426453e-07,
      "learning_rate": 3.619378212134984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8005616,
      "step": 14090
    },
    {
      "epoch": 247.28318584070797,
      "grad_norm": 1.370974359815591e-06,
      "learning_rate": 3.618500288141111e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8008224,
      "step": 14095
    },
    {
      "epoch": 247.3716814159292,
      "grad_norm": 4.461067874217406e-06,
      "learning_rate": 3.617622191660438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8010832,
      "step": 14100
    },
    {
      "epoch": 247.46017699115043,
      "grad_norm": 6.07671950092481e-07,
      "learning_rate": 3.616743922828377e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8013952,
      "step": 14105
    },
    {
      "epoch": 247.5486725663717,
      "grad_norm": 3.126677029285929e-06,
      "learning_rate": 3.615865481780371e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8016352,
      "step": 14110
    },
    {
      "epoch": 247.63716814159292,
      "grad_norm": 1.2296134173084283e-06,
      "learning_rate": 3.614986868651883e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8019968,
      "step": 14115
    },
    {
      "epoch": 247.72566371681415,
      "grad_norm": 8.57390659803059e-07,
      "learning_rate": 3.614108083578409e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8022880,
      "step": 14120
    },
    {
      "epoch": 247.81415929203538,
      "grad_norm": 1.400309201926575e-06,
      "learning_rate": 3.613229126695467e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8025632,
      "step": 14125
    },
    {
      "epoch": 247.90265486725664,
      "grad_norm": 2.327981519556488e-06,
      "learning_rate": 3.612349998138605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8028320,
      "step": 14130
    },
    {
      "epoch": 247.99115044247787,
      "grad_norm": 1.452418246117304e-06,
      "learning_rate": 3.6114706980433946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8031440,
      "step": 14135
    },
    {
      "epoch": 248.07079646017698,
      "grad_norm": 5.149865955900168e-07,
      "learning_rate": 3.610591226545435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8034088,
      "step": 14140
    },
    {
      "epoch": 248.15929203539824,
      "grad_norm": 1.182127789434162e-06,
      "learning_rate": 3.6097115837803505e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8036616,
      "step": 14145
    },
    {
      "epoch": 248.24778761061947,
      "grad_norm": 1.931734686877462e-06,
      "learning_rate": 3.608831769883795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8040040,
      "step": 14150
    },
    {
      "epoch": 248.3362831858407,
      "grad_norm": 2.498170715625747e-06,
      "learning_rate": 3.607951784991446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8042968,
      "step": 14155
    },
    {
      "epoch": 248.42477876106196,
      "grad_norm": 1.5428807955686352e-06,
      "learning_rate": 3.6070716292390085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8045512,
      "step": 14160
    },
    {
      "epoch": 248.5132743362832,
      "grad_norm": 1.2768938404406072e-06,
      "learning_rate": 3.606191302762213e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8048696,
      "step": 14165
    },
    {
      "epoch": 248.60176991150442,
      "grad_norm": 5.73025261019211e-07,
      "learning_rate": 3.605310805696818e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8051192,
      "step": 14170
    },
    {
      "epoch": 248.69026548672565,
      "grad_norm": 1.4016783325132565e-06,
      "learning_rate": 3.6044301381786067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8054232,
      "step": 14175
    },
    {
      "epoch": 248.7787610619469,
      "grad_norm": 5.811000391986454e-07,
      "learning_rate": 3.6035493003433883e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8056824,
      "step": 14180
    },
    {
      "epoch": 248.86725663716814,
      "grad_norm": 2.396063564447104e-06,
      "learning_rate": 3.6026682923269994e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8059928,
      "step": 14185
    },
    {
      "epoch": 248.95575221238937,
      "grad_norm": 1.5667994830437237e-06,
      "learning_rate": 3.6017871142653034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8062632,
      "step": 14190
    },
    {
      "epoch": 249.0353982300885,
      "grad_norm": 9.000637533063127e-07,
      "learning_rate": 3.600905766294189e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8065512,
      "step": 14195
    },
    {
      "epoch": 249.12389380530973,
      "grad_norm": 1.8884835526478128e-06,
      "learning_rate": 3.60002424854957e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8068648,
      "step": 14200
    },
    {
      "epoch": 249.12389380530973,
      "eval_loss": 0.49092504382133484,
      "eval_runtime": 1.0607,
      "eval_samples_per_second": 23.57,
      "eval_steps_per_second": 12.256,
      "num_input_tokens_seen": 8068648,
      "step": 14200
    },
    {
      "epoch": 249.21238938053096,
      "grad_norm": 6.98466578796797e-07,
      "learning_rate": 3.5991425611673876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8071144,
      "step": 14205
    },
    {
      "epoch": 249.30088495575222,
      "grad_norm": 1.6701726508472348e-06,
      "learning_rate": 3.5982607042836105e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8073992,
      "step": 14210
    },
    {
      "epoch": 249.38938053097345,
      "grad_norm": 7.15975488674303e-07,
      "learning_rate": 3.597378678034231e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8076584,
      "step": 14215
    },
    {
      "epoch": 249.47787610619469,
      "grad_norm": 3.4066820262523834e-06,
      "learning_rate": 3.596496482555269e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8079144,
      "step": 14220
    },
    {
      "epoch": 249.56637168141592,
      "grad_norm": 2.577981149443076e-06,
      "learning_rate": 3.595614117982769e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8082184,
      "step": 14225
    },
    {
      "epoch": 249.65486725663717,
      "grad_norm": 1.0573278359515825e-06,
      "learning_rate": 3.594731584452805e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8085208,
      "step": 14230
    },
    {
      "epoch": 249.7433628318584,
      "grad_norm": 8.011470526980702e-07,
      "learning_rate": 3.593848882101472e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8088440,
      "step": 14235
    },
    {
      "epoch": 249.83185840707964,
      "grad_norm": 1.3170886177249486e-06,
      "learning_rate": 3.592966011064896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8090952,
      "step": 14240
    },
    {
      "epoch": 249.9203539823009,
      "grad_norm": 2.764266810117988e-06,
      "learning_rate": 3.592082971479226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8093816,
      "step": 14245
    },
    {
      "epoch": 250.0,
      "grad_norm": 9.86954091786174e-07,
      "learning_rate": 3.5911997634806385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8096192,
      "step": 14250
    },
    {
      "epoch": 250.08849557522123,
      "grad_norm": 1.2436502174750785e-06,
      "learning_rate": 3.5903163872053336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8098672,
      "step": 14255
    },
    {
      "epoch": 250.1769911504425,
      "grad_norm": 2.198921492890804e-06,
      "learning_rate": 3.58943284278954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8101504,
      "step": 14260
    },
    {
      "epoch": 250.26548672566372,
      "grad_norm": 1.302450414186751e-06,
      "learning_rate": 3.588549130369512e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8104944,
      "step": 14265
    },
    {
      "epoch": 250.35398230088495,
      "grad_norm": 6.106538421590813e-07,
      "learning_rate": 3.5876652500815274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8107904,
      "step": 14270
    },
    {
      "epoch": 250.44247787610618,
      "grad_norm": 8.623568419352523e-07,
      "learning_rate": 3.586781202061894e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8110656,
      "step": 14275
    },
    {
      "epoch": 250.53097345132744,
      "grad_norm": 1.1733783367162687e-06,
      "learning_rate": 3.585896986446942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8113280,
      "step": 14280
    },
    {
      "epoch": 250.61946902654867,
      "grad_norm": 3.6656579140981194e-06,
      "learning_rate": 3.585012603373028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8115984,
      "step": 14285
    },
    {
      "epoch": 250.7079646017699,
      "grad_norm": 3.723320673998387e-07,
      "learning_rate": 3.584128052976535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8119008,
      "step": 14290
    },
    {
      "epoch": 250.79646017699116,
      "grad_norm": 2.246128360638977e-06,
      "learning_rate": 3.5832433353938724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8121792,
      "step": 14295
    },
    {
      "epoch": 250.8849557522124,
      "grad_norm": 5.243124974185775e-07,
      "learning_rate": 3.5823584507614746e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8124832,
      "step": 14300
    },
    {
      "epoch": 250.97345132743362,
      "grad_norm": 8.050123483371863e-07,
      "learning_rate": 3.581473399215802e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8127536,
      "step": 14305
    },
    {
      "epoch": 251.05309734513276,
      "grad_norm": 3.7256788232298277e-07,
      "learning_rate": 3.580588180893341e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8129992,
      "step": 14310
    },
    {
      "epoch": 251.141592920354,
      "grad_norm": 1.2387779406708432e-06,
      "learning_rate": 3.579702795930602e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8133000,
      "step": 14315
    },
    {
      "epoch": 251.23008849557522,
      "grad_norm": 1.1343959158693906e-06,
      "learning_rate": 3.578817244464125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8135608,
      "step": 14320
    },
    {
      "epoch": 251.31858407079645,
      "grad_norm": 2.6019422421086347e-06,
      "learning_rate": 3.577931526630471e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8138792,
      "step": 14325
    },
    {
      "epoch": 251.4070796460177,
      "grad_norm": 1.1689319308061386e-06,
      "learning_rate": 3.577045642566229e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8141400,
      "step": 14330
    },
    {
      "epoch": 251.49557522123894,
      "grad_norm": 1.0622857189446222e-06,
      "learning_rate": 3.576159592408014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8144280,
      "step": 14335
    },
    {
      "epoch": 251.58407079646017,
      "grad_norm": 9.667742233432364e-07,
      "learning_rate": 3.575273376292466e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8147272,
      "step": 14340
    },
    {
      "epoch": 251.67256637168143,
      "grad_norm": 1.3918644299337757e-06,
      "learning_rate": 3.574386994356251e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8150232,
      "step": 14345
    },
    {
      "epoch": 251.76106194690266,
      "grad_norm": 8.215726552407432e-07,
      "learning_rate": 3.573500446736059e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8153000,
      "step": 14350
    },
    {
      "epoch": 251.8495575221239,
      "grad_norm": 5.683331778527645e-07,
      "learning_rate": 3.5726137335686094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8155768,
      "step": 14355
    },
    {
      "epoch": 251.93805309734512,
      "grad_norm": 1.3333841479834518e-06,
      "learning_rate": 3.571726854990642e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8158712,
      "step": 14360
    },
    {
      "epoch": 252.01769911504425,
      "grad_norm": 1.175010538645438e-06,
      "learning_rate": 3.570839811138925e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8161616,
      "step": 14365
    },
    {
      "epoch": 252.10619469026548,
      "grad_norm": 6.684330742245947e-07,
      "learning_rate": 3.569952602150252e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8164576,
      "step": 14370
    },
    {
      "epoch": 252.1946902654867,
      "grad_norm": 2.452637090755161e-06,
      "learning_rate": 3.569065228161442e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8167376,
      "step": 14375
    },
    {
      "epoch": 252.28318584070797,
      "grad_norm": 5.259312274574768e-07,
      "learning_rate": 3.5681776893093395e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8170064,
      "step": 14380
    },
    {
      "epoch": 252.3716814159292,
      "grad_norm": 1.5585585515509592e-06,
      "learning_rate": 3.5672899857308134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8172304,
      "step": 14385
    },
    {
      "epoch": 252.46017699115043,
      "grad_norm": 2.718475343499449e-06,
      "learning_rate": 3.566402117562759e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8175664,
      "step": 14390
    },
    {
      "epoch": 252.5486725663717,
      "grad_norm": 1.8720680827755132e-06,
      "learning_rate": 3.565514084942097e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8178800,
      "step": 14395
    },
    {
      "epoch": 252.63716814159292,
      "grad_norm": 6.952332114451565e-07,
      "learning_rate": 3.564625888005773e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8181840,
      "step": 14400
    },
    {
      "epoch": 252.63716814159292,
      "eval_loss": 0.48222678899765015,
      "eval_runtime": 1.0586,
      "eval_samples_per_second": 23.617,
      "eval_steps_per_second": 12.281,
      "num_input_tokens_seen": 8181840,
      "step": 14400
    },
    {
      "epoch": 252.72566371681415,
      "grad_norm": 1.5255036487360485e-06,
      "learning_rate": 3.563737526890759e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8185568,
      "step": 14405
    },
    {
      "epoch": 252.81415929203538,
      "grad_norm": 5.211735469856649e-07,
      "learning_rate": 3.562849001734049e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8188272,
      "step": 14410
    },
    {
      "epoch": 252.90265486725664,
      "grad_norm": 1.4385176427822444e-06,
      "learning_rate": 3.561960312672667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8190960,
      "step": 14415
    },
    {
      "epoch": 252.99115044247787,
      "grad_norm": 1.8412550844004727e-06,
      "learning_rate": 3.5610714598436596e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8193312,
      "step": 14420
    },
    {
      "epoch": 253.07079646017698,
      "grad_norm": 1.7384606962878024e-06,
      "learning_rate": 3.5601824433840986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8195488,
      "step": 14425
    },
    {
      "epoch": 253.15929203539824,
      "grad_norm": 5.394878712650097e-07,
      "learning_rate": 3.559293263431082e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8198384,
      "step": 14430
    },
    {
      "epoch": 253.24778761061947,
      "grad_norm": 5.836478180754057e-07,
      "learning_rate": 3.558403920121732e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8200880,
      "step": 14435
    },
    {
      "epoch": 253.3362831858407,
      "grad_norm": 4.6905583417355956e-07,
      "learning_rate": 3.557514413593197e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8203680,
      "step": 14440
    },
    {
      "epoch": 253.42477876106196,
      "grad_norm": 9.348192975267011e-07,
      "learning_rate": 3.55662474398265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8206480,
      "step": 14445
    },
    {
      "epoch": 253.5132743362832,
      "grad_norm": 1.984589971470996e-06,
      "learning_rate": 3.555734911427288e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8209584,
      "step": 14450
    },
    {
      "epoch": 253.60176991150442,
      "grad_norm": 3.477433381249284e-07,
      "learning_rate": 3.5548449160643363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8212720,
      "step": 14455
    },
    {
      "epoch": 253.69026548672565,
      "grad_norm": 1.2757685681208386e-06,
      "learning_rate": 3.553954758031043e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8215296,
      "step": 14460
    },
    {
      "epoch": 253.7787610619469,
      "grad_norm": 1.203301167151949e-06,
      "learning_rate": 3.5530644374646815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8218864,
      "step": 14465
    },
    {
      "epoch": 253.86725663716814,
      "grad_norm": 1.3746000604442088e-06,
      "learning_rate": 3.552173954502549e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8221824,
      "step": 14470
    },
    {
      "epoch": 253.95575221238937,
      "grad_norm": 9.324435268354136e-07,
      "learning_rate": 3.55128330928197e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8224288,
      "step": 14475
    },
    {
      "epoch": 254.0353982300885,
      "grad_norm": 1.684396920609288e-06,
      "learning_rate": 3.550392501940294e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8226832,
      "step": 14480
    },
    {
      "epoch": 254.12389380530973,
      "grad_norm": 1.433025886399264e-06,
      "learning_rate": 3.5495015326148945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8229712,
      "step": 14485
    },
    {
      "epoch": 254.21238938053096,
      "grad_norm": 4.816583896172233e-07,
      "learning_rate": 3.548610401443169e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8232016,
      "step": 14490
    },
    {
      "epoch": 254.30088495575222,
      "grad_norm": 2.165244723073556e-06,
      "learning_rate": 3.547719108562543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8235040,
      "step": 14495
    },
    {
      "epoch": 254.38938053097345,
      "grad_norm": 3.827458385785576e-06,
      "learning_rate": 3.546827654110464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8238784,
      "step": 14500
    },
    {
      "epoch": 254.47787610619469,
      "grad_norm": 2.246403255412588e-06,
      "learning_rate": 3.545936038224405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8241408,
      "step": 14505
    },
    {
      "epoch": 254.56637168141592,
      "grad_norm": 6.262943088586326e-07,
      "learning_rate": 3.545044261041864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8244320,
      "step": 14510
    },
    {
      "epoch": 254.65486725663717,
      "grad_norm": 1.8211693486591685e-06,
      "learning_rate": 3.5441523227003657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8246896,
      "step": 14515
    },
    {
      "epoch": 254.7433628318584,
      "grad_norm": 2.9559721497207647e-06,
      "learning_rate": 3.543260223337459e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8249840,
      "step": 14520
    },
    {
      "epoch": 254.83185840707964,
      "grad_norm": 4.6969034883659333e-07,
      "learning_rate": 3.542367963090714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8252672,
      "step": 14525
    },
    {
      "epoch": 254.9203539823009,
      "grad_norm": 2.6917384730040794e-06,
      "learning_rate": 3.5414755420977295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8255904,
      "step": 14530
    },
    {
      "epoch": 255.0,
      "grad_norm": 1.6744473896324052e-06,
      "learning_rate": 3.54058296049613e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8258352,
      "step": 14535
    },
    {
      "epoch": 255.08849557522123,
      "grad_norm": 1.374637577100657e-06,
      "learning_rate": 3.53969021842356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8261040,
      "step": 14540
    },
    {
      "epoch": 255.1769911504425,
      "grad_norm": 8.028592333175766e-07,
      "learning_rate": 3.5387973160176926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8263712,
      "step": 14545
    },
    {
      "epoch": 255.26548672566372,
      "grad_norm": 1.6707207350918907e-06,
      "learning_rate": 3.537904253416224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8266016,
      "step": 14550
    },
    {
      "epoch": 255.35398230088495,
      "grad_norm": 1.6460634242321248e-06,
      "learning_rate": 3.537011030756878e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8268928,
      "step": 14555
    },
    {
      "epoch": 255.44247787610618,
      "grad_norm": 2.587397148090531e-06,
      "learning_rate": 3.536117648177399e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8272432,
      "step": 14560
    },
    {
      "epoch": 255.53097345132744,
      "grad_norm": 1.1895590432686731e-06,
      "learning_rate": 3.535224105815558e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8274912,
      "step": 14565
    },
    {
      "epoch": 255.61946902654867,
      "grad_norm": 4.6311021151268505e-07,
      "learning_rate": 3.5343304038091494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8278032,
      "step": 14570
    },
    {
      "epoch": 255.7079646017699,
      "grad_norm": 2.8466138246585615e-07,
      "learning_rate": 3.5334365422959955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8280896,
      "step": 14575
    },
    {
      "epoch": 255.79646017699116,
      "grad_norm": 8.347872721969907e-07,
      "learning_rate": 3.5325425214139396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8283616,
      "step": 14580
    },
    {
      "epoch": 255.8849557522124,
      "grad_norm": 6.731918915647839e-07,
      "learning_rate": 3.531648341300851e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8286352,
      "step": 14585
    },
    {
      "epoch": 255.97345132743362,
      "grad_norm": 6.225282049854286e-07,
      "learning_rate": 3.530754002094623e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8289728,
      "step": 14590
    },
    {
      "epoch": 256.05309734513276,
      "grad_norm": 5.313100928105996e-07,
      "learning_rate": 3.529859503933175e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8291776,
      "step": 14595
    },
    {
      "epoch": 256.14159292035396,
      "grad_norm": 8.133189908221539e-07,
      "learning_rate": 3.52896484695445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8294896,
      "step": 14600
    },
    {
      "epoch": 256.14159292035396,
      "eval_loss": 0.4791489839553833,
      "eval_runtime": 1.0569,
      "eval_samples_per_second": 23.654,
      "eval_steps_per_second": 12.3,
      "num_input_tokens_seen": 8294896,
      "step": 14600
    },
    {
      "epoch": 256.2300884955752,
      "grad_norm": 2.602055701572681e-06,
      "learning_rate": 3.528070031296414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8297696,
      "step": 14605
    },
    {
      "epoch": 256.3185840707965,
      "grad_norm": 1.08228994122328e-06,
      "learning_rate": 3.5271750570970605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8300496,
      "step": 14610
    },
    {
      "epoch": 256.4070796460177,
      "grad_norm": 4.161539663982694e-07,
      "learning_rate": 3.526279924494405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8303488,
      "step": 14615
    },
    {
      "epoch": 256.49557522123894,
      "grad_norm": 2.3231484647112666e-06,
      "learning_rate": 3.5253846336264874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8306624,
      "step": 14620
    },
    {
      "epoch": 256.5840707964602,
      "grad_norm": 4.0792971844894055e-07,
      "learning_rate": 3.5244891846313736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8309264,
      "step": 14625
    },
    {
      "epoch": 256.6725663716814,
      "grad_norm": 1.0403635997136007e-06,
      "learning_rate": 3.5235935776471527e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8312208,
      "step": 14630
    },
    {
      "epoch": 256.76106194690266,
      "grad_norm": 5.330630870048481e-07,
      "learning_rate": 3.522697812811939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8314656,
      "step": 14635
    },
    {
      "epoch": 256.8495575221239,
      "grad_norm": 8.795990993348823e-07,
      "learning_rate": 3.521801890263871e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8317728,
      "step": 14640
    },
    {
      "epoch": 256.9380530973451,
      "grad_norm": 6.694728540423966e-07,
      "learning_rate": 3.5209058101411114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8321104,
      "step": 14645
    },
    {
      "epoch": 257.01769911504425,
      "grad_norm": 2.3820514343242394e-06,
      "learning_rate": 3.520009572581845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8323296,
      "step": 14650
    },
    {
      "epoch": 257.1061946902655,
      "grad_norm": 1.568815946484392e-06,
      "learning_rate": 3.519113177724285e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8325968,
      "step": 14655
    },
    {
      "epoch": 257.1946902654867,
      "grad_norm": 4.48227183369454e-07,
      "learning_rate": 3.5182166257066656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8328688,
      "step": 14660
    },
    {
      "epoch": 257.283185840708,
      "grad_norm": 7.404619850603922e-07,
      "learning_rate": 3.517319916667247e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8331392,
      "step": 14665
    },
    {
      "epoch": 257.37168141592923,
      "grad_norm": 2.2128392629383598e-06,
      "learning_rate": 3.516423050744313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8334592,
      "step": 14670
    },
    {
      "epoch": 257.46017699115043,
      "grad_norm": 1.8167374946642667e-06,
      "learning_rate": 3.5155260280761704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8337504,
      "step": 14675
    },
    {
      "epoch": 257.5486725663717,
      "grad_norm": 1.3133751508576097e-06,
      "learning_rate": 3.514628848801154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8340528,
      "step": 14680
    },
    {
      "epoch": 257.6371681415929,
      "grad_norm": 7.271790991580929e-07,
      "learning_rate": 3.5137315130576174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8343168,
      "step": 14685
    },
    {
      "epoch": 257.72566371681415,
      "grad_norm": 2.5028509753610706e-06,
      "learning_rate": 3.512834020983942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8346304,
      "step": 14690
    },
    {
      "epoch": 257.8141592920354,
      "grad_norm": 1.455070332667674e-06,
      "learning_rate": 3.5119363727185334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8349152,
      "step": 14695
    },
    {
      "epoch": 257.9026548672566,
      "grad_norm": 6.929033133928897e-07,
      "learning_rate": 3.511038568399819e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8351952,
      "step": 14700
    },
    {
      "epoch": 257.9911504424779,
      "grad_norm": 1.2542064951048815e-06,
      "learning_rate": 3.510140608166251e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8355408,
      "step": 14705
    },
    {
      "epoch": 258.070796460177,
      "grad_norm": 2.087663233396597e-06,
      "learning_rate": 3.509242492156308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8357384,
      "step": 14710
    },
    {
      "epoch": 258.1592920353982,
      "grad_norm": 1.329169208474923e-06,
      "learning_rate": 3.5083442205084896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8360360,
      "step": 14715
    },
    {
      "epoch": 258.24778761061947,
      "grad_norm": 5.441485200208263e-07,
      "learning_rate": 3.507445793361321e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8363064,
      "step": 14720
    },
    {
      "epoch": 258.3362831858407,
      "grad_norm": 1.4323593404697021e-06,
      "learning_rate": 3.5065472108533505e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8366088,
      "step": 14725
    },
    {
      "epoch": 258.42477876106193,
      "grad_norm": 1.2842502883358975e-06,
      "learning_rate": 3.5056484731231504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8368568,
      "step": 14730
    },
    {
      "epoch": 258.5132743362832,
      "grad_norm": 1.2898842669528676e-06,
      "learning_rate": 3.504749580309319e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8372072,
      "step": 14735
    },
    {
      "epoch": 258.60176991150445,
      "grad_norm": 1.0181779543927405e-06,
      "learning_rate": 3.5038505325504753e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8375512,
      "step": 14740
    },
    {
      "epoch": 258.69026548672565,
      "grad_norm": 4.297435225453228e-07,
      "learning_rate": 3.502951329985264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8378376,
      "step": 14745
    },
    {
      "epoch": 258.7787610619469,
      "grad_norm": 4.175441858933482e-07,
      "learning_rate": 3.502051972752354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8380824,
      "step": 14750
    },
    {
      "epoch": 258.86725663716817,
      "grad_norm": 7.868840725677728e-07,
      "learning_rate": 3.5011524609904374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8383688,
      "step": 14755
    },
    {
      "epoch": 258.95575221238937,
      "grad_norm": 8.320688493768102e-07,
      "learning_rate": 3.50025279483823e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8386616,
      "step": 14760
    },
    {
      "epoch": 259.0353982300885,
      "grad_norm": 1.2337935686446144e-06,
      "learning_rate": 3.499352974434472e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8389072,
      "step": 14765
    },
    {
      "epoch": 259.12389380530976,
      "grad_norm": 9.666533742347383e-07,
      "learning_rate": 3.498452999917926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8392064,
      "step": 14770
    },
    {
      "epoch": 259.21238938053096,
      "grad_norm": 3.011133003383293e-06,
      "learning_rate": 3.4975528714273795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8394864,
      "step": 14775
    },
    {
      "epoch": 259.3008849557522,
      "grad_norm": 1.9461290321487468e-06,
      "learning_rate": 3.4966525891016454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8397840,
      "step": 14780
    },
    {
      "epoch": 259.3893805309734,
      "grad_norm": 5.232922717368638e-07,
      "learning_rate": 3.495752153079557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8400288,
      "step": 14785
    },
    {
      "epoch": 259.4778761061947,
      "grad_norm": 1.6585144066993962e-06,
      "learning_rate": 3.494851563499974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8402816,
      "step": 14790
    },
    {
      "epoch": 259.56637168141594,
      "grad_norm": 5.823263791171485e-07,
      "learning_rate": 3.493950820501777e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8405616,
      "step": 14795
    },
    {
      "epoch": 259.65486725663715,
      "grad_norm": 8.602528396295384e-07,
      "learning_rate": 3.493049924223872e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8408512,
      "step": 14800
    },
    {
      "epoch": 259.65486725663715,
      "eval_loss": 0.4914831817150116,
      "eval_runtime": 1.0637,
      "eval_samples_per_second": 23.503,
      "eval_steps_per_second": 12.221,
      "num_input_tokens_seen": 8408512,
      "step": 14800
    },
    {
      "epoch": 259.7433628318584,
      "grad_norm": 1.6144919072758057e-06,
      "learning_rate": 3.49214887480519e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8411120,
      "step": 14805
    },
    {
      "epoch": 259.83185840707966,
      "grad_norm": 3.7479262005035707e-07,
      "learning_rate": 3.4912476723846834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8414224,
      "step": 14810
    },
    {
      "epoch": 259.92035398230087,
      "grad_norm": 1.1142145694975625e-06,
      "learning_rate": 3.490346317101328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8417392,
      "step": 14815
    },
    {
      "epoch": 260.0,
      "grad_norm": 3.1243462217389606e-06,
      "learning_rate": 3.4894448090941266e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8420256,
      "step": 14820
    },
    {
      "epoch": 260.08849557522126,
      "grad_norm": 1.0424693073218805e-06,
      "learning_rate": 3.488543148502101e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8423168,
      "step": 14825
    },
    {
      "epoch": 260.17699115044246,
      "grad_norm": 1.2681972521022544e-06,
      "learning_rate": 3.487641335464299e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8425760,
      "step": 14830
    },
    {
      "epoch": 260.2654867256637,
      "grad_norm": 1.9051331037189811e-06,
      "learning_rate": 3.4867393701197914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8428720,
      "step": 14835
    },
    {
      "epoch": 260.353982300885,
      "grad_norm": 1.691775537437934e-06,
      "learning_rate": 3.485837252607673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8431392,
      "step": 14840
    },
    {
      "epoch": 260.4424778761062,
      "grad_norm": 3.5796772408502875e-06,
      "learning_rate": 3.4849349830670615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8434208,
      "step": 14845
    },
    {
      "epoch": 260.53097345132744,
      "grad_norm": 5.93705294704705e-07,
      "learning_rate": 3.4840325616370976e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8436864,
      "step": 14850
    },
    {
      "epoch": 260.6194690265487,
      "grad_norm": 7.043175287435588e-07,
      "learning_rate": 3.483129988456947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8440288,
      "step": 14855
    },
    {
      "epoch": 260.7079646017699,
      "grad_norm": 6.574559847649653e-07,
      "learning_rate": 3.482227263665797e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8442832,
      "step": 14860
    },
    {
      "epoch": 260.79646017699116,
      "grad_norm": 1.5357049960584845e-06,
      "learning_rate": 3.48132438740286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8446208,
      "step": 14865
    },
    {
      "epoch": 260.88495575221236,
      "grad_norm": 2.681051398667478e-07,
      "learning_rate": 3.48042135980737e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8448752,
      "step": 14870
    },
    {
      "epoch": 260.9734513274336,
      "grad_norm": 1.0401020063000033e-06,
      "learning_rate": 3.479518181018586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8451504,
      "step": 14875
    },
    {
      "epoch": 261.05309734513276,
      "grad_norm": 4.937883772981877e-07,
      "learning_rate": 3.4786148511757886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8454168,
      "step": 14880
    },
    {
      "epoch": 261.14159292035396,
      "grad_norm": 2.0234047042322345e-06,
      "learning_rate": 3.477711370418284e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8457336,
      "step": 14885
    },
    {
      "epoch": 261.2300884955752,
      "grad_norm": 9.177801985060796e-07,
      "learning_rate": 3.476807738885399e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8460856,
      "step": 14890
    },
    {
      "epoch": 261.3185840707965,
      "grad_norm": 1.2371817774692317e-06,
      "learning_rate": 3.475903956716485e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8464136,
      "step": 14895
    },
    {
      "epoch": 261.4070796460177,
      "grad_norm": 2.8097217636968708e-06,
      "learning_rate": 3.475000024050917e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8466488,
      "step": 14900
    },
    {
      "epoch": 261.49557522123894,
      "grad_norm": 3.449051746429177e-06,
      "learning_rate": 3.4740959410280926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8469592,
      "step": 14905
    },
    {
      "epoch": 261.5840707964602,
      "grad_norm": 5.460715328808874e-06,
      "learning_rate": 3.4731917077874324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8472168,
      "step": 14910
    },
    {
      "epoch": 261.6725663716814,
      "grad_norm": 6.441593995987205e-07,
      "learning_rate": 3.4722873244683816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8474760,
      "step": 14915
    },
    {
      "epoch": 261.76106194690266,
      "grad_norm": 1.542122504361032e-06,
      "learning_rate": 3.4713827912104065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8477768,
      "step": 14920
    },
    {
      "epoch": 261.8495575221239,
      "grad_norm": 7.7458884106818e-07,
      "learning_rate": 3.470478108152998e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8480360,
      "step": 14925
    },
    {
      "epoch": 261.9380530973451,
      "grad_norm": 1.2497988564064144e-06,
      "learning_rate": 3.4695732754356695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8483336,
      "step": 14930
    },
    {
      "epoch": 262.01769911504425,
      "grad_norm": 1.107884031625872e-06,
      "learning_rate": 3.4686682931979576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8485600,
      "step": 14935
    },
    {
      "epoch": 262.1061946902655,
      "grad_norm": 1.1375398116797442e-06,
      "learning_rate": 3.467763161579422e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8488384,
      "step": 14940
    },
    {
      "epoch": 262.1946902654867,
      "grad_norm": 1.3165021073291427e-06,
      "learning_rate": 3.466857880719645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8491600,
      "step": 14945
    },
    {
      "epoch": 262.283185840708,
      "grad_norm": 1.2815564787160838e-06,
      "learning_rate": 3.465952450758233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8494528,
      "step": 14950
    },
    {
      "epoch": 262.37168141592923,
      "grad_norm": 1.1369298817953677e-06,
      "learning_rate": 3.4650468718348126e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8497312,
      "step": 14955
    },
    {
      "epoch": 262.46017699115043,
      "grad_norm": 7.962932500049646e-07,
      "learning_rate": 3.464141144089038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8500192,
      "step": 14960
    },
    {
      "epoch": 262.5486725663717,
      "grad_norm": 3.0405644793063402e-06,
      "learning_rate": 3.463235267660583e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8503280,
      "step": 14965
    },
    {
      "epoch": 262.6371681415929,
      "grad_norm": 9.543251735522063e-07,
      "learning_rate": 3.462329242689145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8506448,
      "step": 14970
    },
    {
      "epoch": 262.72566371681415,
      "grad_norm": 2.2175909180077724e-06,
      "learning_rate": 3.461423069314444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8508960,
      "step": 14975
    },
    {
      "epoch": 262.8141592920354,
      "grad_norm": 3.8094043475211947e-07,
      "learning_rate": 3.460516747676224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8511840,
      "step": 14980
    },
    {
      "epoch": 262.9026548672566,
      "grad_norm": 9.673341310190153e-07,
      "learning_rate": 3.459610277914251e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8514992,
      "step": 14985
    },
    {
      "epoch": 262.9911504424779,
      "grad_norm": 1.5998923572624335e-06,
      "learning_rate": 3.458703660168314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8517376,
      "step": 14990
    },
    {
      "epoch": 263.070796460177,
      "grad_norm": 8.393830057684681e-07,
      "learning_rate": 3.457796894578224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8520168,
      "step": 14995
    },
    {
      "epoch": 263.1592920353982,
      "grad_norm": 1.338665242656134e-06,
      "learning_rate": 3.456889981283817e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8522664,
      "step": 15000
    },
    {
      "epoch": 263.1592920353982,
      "eval_loss": 0.4853701889514923,
      "eval_runtime": 1.0635,
      "eval_samples_per_second": 23.507,
      "eval_steps_per_second": 12.224,
      "num_input_tokens_seen": 8522664,
      "step": 15000
    },
    {
      "epoch": 263.24778761061947,
      "grad_norm": 7.675539563933853e-07,
      "learning_rate": 3.45598292042495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8525480,
      "step": 15005
    },
    {
      "epoch": 263.3362831858407,
      "grad_norm": 1.4099059626460075e-06,
      "learning_rate": 3.4550757121415035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8528440,
      "step": 15010
    },
    {
      "epoch": 263.42477876106193,
      "grad_norm": 1.9466376670607133e-06,
      "learning_rate": 3.454168356573378e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8531224,
      "step": 15015
    },
    {
      "epoch": 263.5132743362832,
      "grad_norm": 1.0763701538962778e-06,
      "learning_rate": 3.453260853860503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8534152,
      "step": 15020
    },
    {
      "epoch": 263.60176991150445,
      "grad_norm": 9.397218150297704e-07,
      "learning_rate": 3.452353204142824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8537240,
      "step": 15025
    },
    {
      "epoch": 263.69026548672565,
      "grad_norm": 2.942961373264552e-06,
      "learning_rate": 3.4514454075603136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8539960,
      "step": 15030
    },
    {
      "epoch": 263.7787610619469,
      "grad_norm": 5.308393156155944e-07,
      "learning_rate": 3.450537464252964e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8543400,
      "step": 15035
    },
    {
      "epoch": 263.86725663716817,
      "grad_norm": 1.484595031797653e-06,
      "learning_rate": 3.4496293743607925e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8545784,
      "step": 15040
    },
    {
      "epoch": 263.95575221238937,
      "grad_norm": 1.3838721315551084e-06,
      "learning_rate": 3.448721138023838e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8548536,
      "step": 15045
    },
    {
      "epoch": 264.0353982300885,
      "grad_norm": 1.4170252597978106e-06,
      "learning_rate": 3.447812755382162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8550744,
      "step": 15050
    },
    {
      "epoch": 264.12389380530976,
      "grad_norm": 2.0554389266180806e-06,
      "learning_rate": 3.446904226575847e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8553656,
      "step": 15055
    },
    {
      "epoch": 264.21238938053096,
      "grad_norm": 9.458401564188534e-07,
      "learning_rate": 3.445995551745002e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8556616,
      "step": 15060
    },
    {
      "epoch": 264.3008849557522,
      "grad_norm": 1.7509718190922285e-06,
      "learning_rate": 3.445086731029753e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8559176,
      "step": 15065
    },
    {
      "epoch": 264.3893805309734,
      "grad_norm": 5.002589773539512e-07,
      "learning_rate": 3.444177764570255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8562088,
      "step": 15070
    },
    {
      "epoch": 264.4778761061947,
      "grad_norm": 1.7799068245949456e-06,
      "learning_rate": 3.44326865250668e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8564936,
      "step": 15075
    },
    {
      "epoch": 264.56637168141594,
      "grad_norm": 5.381925802794285e-07,
      "learning_rate": 3.442359394979225e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8567544,
      "step": 15080
    },
    {
      "epoch": 264.65486725663715,
      "grad_norm": 7.112867024261504e-07,
      "learning_rate": 3.441449992128108e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8570008,
      "step": 15085
    },
    {
      "epoch": 264.7433628318584,
      "grad_norm": 4.221373330892675e-07,
      "learning_rate": 3.440540444093573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8572808,
      "step": 15090
    },
    {
      "epoch": 264.83185840707966,
      "grad_norm": 7.103952270881564e-07,
      "learning_rate": 3.43963075101588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8575864,
      "step": 15095
    },
    {
      "epoch": 264.92035398230087,
      "grad_norm": 9.542341103951912e-07,
      "learning_rate": 3.438720913035318e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8578712,
      "step": 15100
    },
    {
      "epoch": 265.0,
      "grad_norm": 1.3121024267093162e-06,
      "learning_rate": 3.437810930292195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8581640,
      "step": 15105
    },
    {
      "epoch": 265.08849557522126,
      "grad_norm": 4.4089344441999856e-07,
      "learning_rate": 3.43690080292684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8584232,
      "step": 15110
    },
    {
      "epoch": 265.17699115044246,
      "grad_norm": 3.3191989246006415e-07,
      "learning_rate": 3.435990531079608e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8586904,
      "step": 15115
    },
    {
      "epoch": 265.2654867256637,
      "grad_norm": 1.5036617924124585e-06,
      "learning_rate": 3.435080114890874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8589480,
      "step": 15120
    },
    {
      "epoch": 265.353982300885,
      "grad_norm": 8.959405022324063e-07,
      "learning_rate": 3.434169554501035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8592392,
      "step": 15125
    },
    {
      "epoch": 265.4424778761062,
      "grad_norm": 6.865363957331283e-07,
      "learning_rate": 3.433258850050511e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8595688,
      "step": 15130
    },
    {
      "epoch": 265.53097345132744,
      "grad_norm": 5.737164201491396e-07,
      "learning_rate": 3.4323480016797446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8598392,
      "step": 15135
    },
    {
      "epoch": 265.6194690265487,
      "grad_norm": 2.260910605400568e-06,
      "learning_rate": 3.4314370095291995e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8601064,
      "step": 15140
    },
    {
      "epoch": 265.7079646017699,
      "grad_norm": 9.63571437750943e-07,
      "learning_rate": 3.430525873739363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8603752,
      "step": 15145
    },
    {
      "epoch": 265.79646017699116,
      "grad_norm": 5.697297069673368e-07,
      "learning_rate": 3.429614594450743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8606504,
      "step": 15150
    },
    {
      "epoch": 265.88495575221236,
      "grad_norm": 1.6471558410557918e-06,
      "learning_rate": 3.428703171803869e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8609864,
      "step": 15155
    },
    {
      "epoch": 265.9734513274336,
      "grad_norm": 1.50068728999031e-06,
      "learning_rate": 3.4277916059392964e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8613224,
      "step": 15160
    },
    {
      "epoch": 266.05309734513276,
      "grad_norm": 1.4102569139140542e-06,
      "learning_rate": 3.426879896997598e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8615648,
      "step": 15165
    },
    {
      "epoch": 266.14159292035396,
      "grad_norm": 2.381353624514304e-06,
      "learning_rate": 3.425968045119372e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8618576,
      "step": 15170
    },
    {
      "epoch": 266.2300884955752,
      "grad_norm": 1.0292740171280457e-06,
      "learning_rate": 3.425056050445237e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8621280,
      "step": 15175
    },
    {
      "epoch": 266.3185840707965,
      "grad_norm": 4.5609741050611774e-07,
      "learning_rate": 3.4241439131158336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8624064,
      "step": 15180
    },
    {
      "epoch": 266.4070796460177,
      "grad_norm": 1.5530437167399214e-06,
      "learning_rate": 3.423231633271825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8626768,
      "step": 15185
    },
    {
      "epoch": 266.49557522123894,
      "grad_norm": 1.2744596915581496e-06,
      "learning_rate": 3.4223192110538985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8630224,
      "step": 15190
    },
    {
      "epoch": 266.5840707964602,
      "grad_norm": 1.0613531458147918e-06,
      "learning_rate": 3.4214066466027575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8633296,
      "step": 15195
    },
    {
      "epoch": 266.6725663716814,
      "grad_norm": 4.975373144588957e-07,
      "learning_rate": 3.4204939400591325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8636032,
      "step": 15200
    },
    {
      "epoch": 266.6725663716814,
      "eval_loss": 0.501243531703949,
      "eval_runtime": 1.063,
      "eval_samples_per_second": 23.518,
      "eval_steps_per_second": 12.23,
      "num_input_tokens_seen": 8636032,
      "step": 15200
    },
    {
      "epoch": 266.76106194690266,
      "grad_norm": 1.0023895811173134e-06,
      "learning_rate": 3.419581091563775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8638768,
      "step": 15205
    },
    {
      "epoch": 266.8495575221239,
      "grad_norm": 9.16552437502105e-07,
      "learning_rate": 3.418668101257456e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8641520,
      "step": 15210
    },
    {
      "epoch": 266.9380530973451,
      "grad_norm": 8.743572834646329e-07,
      "learning_rate": 3.417754969280971e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8644592,
      "step": 15215
    },
    {
      "epoch": 267.01769911504425,
      "grad_norm": 6.147636781861365e-07,
      "learning_rate": 3.416841695775137e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8647352,
      "step": 15220
    },
    {
      "epoch": 267.1061946902655,
      "grad_norm": 1.8611965515447082e-06,
      "learning_rate": 3.415928280880792e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8650632,
      "step": 15225
    },
    {
      "epoch": 267.1946902654867,
      "grad_norm": 7.769605190333095e-07,
      "learning_rate": 3.4150147247387965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8653288,
      "step": 15230
    },
    {
      "epoch": 267.283185840708,
      "grad_norm": 1.4341896985570202e-06,
      "learning_rate": 3.4141010274900306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8656376,
      "step": 15235
    },
    {
      "epoch": 267.37168141592923,
      "grad_norm": 8.122426606860245e-07,
      "learning_rate": 3.413187189275399e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8659080,
      "step": 15240
    },
    {
      "epoch": 267.46017699115043,
      "grad_norm": 1.9687772692122962e-06,
      "learning_rate": 3.4122732102358265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8662184,
      "step": 15245
    },
    {
      "epoch": 267.5486725663717,
      "grad_norm": 6.960215159779182e-07,
      "learning_rate": 3.411359090512261e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8664328,
      "step": 15250
    },
    {
      "epoch": 267.6371681415929,
      "grad_norm": 5.843225153512321e-06,
      "learning_rate": 3.410444830245672e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8667192,
      "step": 15255
    },
    {
      "epoch": 267.72566371681415,
      "grad_norm": 8.683188070790493e-07,
      "learning_rate": 3.409530429577048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8669912,
      "step": 15260
    },
    {
      "epoch": 267.8141592920354,
      "grad_norm": 3.9637293411942665e-06,
      "learning_rate": 3.408615888647402e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8672904,
      "step": 15265
    },
    {
      "epoch": 267.9026548672566,
      "grad_norm": 7.368043384303746e-07,
      "learning_rate": 3.4077012075977675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8675608,
      "step": 15270
    },
    {
      "epoch": 267.9911504424779,
      "grad_norm": 2.4011296773096547e-06,
      "learning_rate": 3.4067863865692e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8678808,
      "step": 15275
    },
    {
      "epoch": 268.070796460177,
      "grad_norm": 6.186106702443794e-07,
      "learning_rate": 3.4058714257027755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8681520,
      "step": 15280
    },
    {
      "epoch": 268.1592920353982,
      "grad_norm": 1.7642863667788333e-06,
      "learning_rate": 3.404956325139594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8684240,
      "step": 15285
    },
    {
      "epoch": 268.24778761061947,
      "grad_norm": 5.661693194269901e-07,
      "learning_rate": 3.404041085020775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8687232,
      "step": 15290
    },
    {
      "epoch": 268.3362831858407,
      "grad_norm": 1.0711438562793774e-06,
      "learning_rate": 3.403125705487459e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8690064,
      "step": 15295
    },
    {
      "epoch": 268.42477876106193,
      "grad_norm": 2.2653252074178454e-07,
      "learning_rate": 3.402210186680811e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8692912,
      "step": 15300
    },
    {
      "epoch": 268.5132743362832,
      "grad_norm": 6.40461280454474e-07,
      "learning_rate": 3.4012945287420137e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8695648,
      "step": 15305
    },
    {
      "epoch": 268.60176991150445,
      "grad_norm": 6.79984850648907e-07,
      "learning_rate": 3.400378731812274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8698512,
      "step": 15310
    },
    {
      "epoch": 268.69026548672565,
      "grad_norm": 9.924475534717203e-07,
      "learning_rate": 3.399462796032817e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8701232,
      "step": 15315
    },
    {
      "epoch": 268.7787610619469,
      "grad_norm": 3.893907773999672e-07,
      "learning_rate": 3.3985467215448954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8704144,
      "step": 15320
    },
    {
      "epoch": 268.86725663716817,
      "grad_norm": 1.6327379626090988e-06,
      "learning_rate": 3.3976305084897776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8707104,
      "step": 15325
    },
    {
      "epoch": 268.95575221238937,
      "grad_norm": 1.6303425809383043e-06,
      "learning_rate": 3.3967141570087544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8709664,
      "step": 15330
    },
    {
      "epoch": 269.0353982300885,
      "grad_norm": 1.7092507960114744e-06,
      "learning_rate": 3.39579766724314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8711928,
      "step": 15335
    },
    {
      "epoch": 269.12389380530976,
      "grad_norm": 1.4318029570858926e-06,
      "learning_rate": 3.3948810393342677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8714520,
      "step": 15340
    },
    {
      "epoch": 269.21238938053096,
      "grad_norm": 1.0631599707267014e-06,
      "learning_rate": 3.3939642734234936e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8717688,
      "step": 15345
    },
    {
      "epoch": 269.3008849557522,
      "grad_norm": 7.27307281067624e-07,
      "learning_rate": 3.393047369652194e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8720408,
      "step": 15350
    },
    {
      "epoch": 269.3893805309734,
      "grad_norm": 4.7551696980008273e-07,
      "learning_rate": 3.3921303281617664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8723752,
      "step": 15355
    },
    {
      "epoch": 269.4778761061947,
      "grad_norm": 5.302986778588092e-07,
      "learning_rate": 3.391213149093632e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8726568,
      "step": 15360
    },
    {
      "epoch": 269.56637168141594,
      "grad_norm": 1.5207909882519743e-06,
      "learning_rate": 3.3902958325892303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8729064,
      "step": 15365
    },
    {
      "epoch": 269.65486725663715,
      "grad_norm": 2.8786798793589696e-06,
      "learning_rate": 3.389378378790023e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8731960,
      "step": 15370
    },
    {
      "epoch": 269.7433628318584,
      "grad_norm": 1.1388520988475648e-06,
      "learning_rate": 3.388460787837493e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8734936,
      "step": 15375
    },
    {
      "epoch": 269.83185840707966,
      "grad_norm": 7.786312039570475e-07,
      "learning_rate": 3.387543059873145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8738168,
      "step": 15380
    },
    {
      "epoch": 269.92035398230087,
      "grad_norm": 5.513125529432727e-07,
      "learning_rate": 3.386625195038503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8740680,
      "step": 15385
    },
    {
      "epoch": 270.0,
      "grad_norm": 1.384447273267142e-06,
      "learning_rate": 3.3857071934751136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8743392,
      "step": 15390
    },
    {
      "epoch": 270.08849557522126,
      "grad_norm": 1.4997228845459176e-06,
      "learning_rate": 3.384789055324544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8746000,
      "step": 15395
    },
    {
      "epoch": 270.17699115044246,
      "grad_norm": 3.6848379636467143e-07,
      "learning_rate": 3.3838707807283843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8748624,
      "step": 15400
    },
    {
      "epoch": 270.17699115044246,
      "eval_loss": 0.5021628737449646,
      "eval_runtime": 1.0702,
      "eval_samples_per_second": 23.359,
      "eval_steps_per_second": 12.147,
      "num_input_tokens_seen": 8748624,
      "step": 15400
    },
    {
      "epoch": 270.2654867256637,
      "grad_norm": 1.2962563005203265e-06,
      "learning_rate": 3.382952369828243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8751984,
      "step": 15405
    },
    {
      "epoch": 270.353982300885,
      "grad_norm": 6.411088406821364e-07,
      "learning_rate": 3.38203382276575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8754560,
      "step": 15410
    },
    {
      "epoch": 270.4424778761062,
      "grad_norm": 1.4747315617569257e-06,
      "learning_rate": 3.381115139682557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8757600,
      "step": 15415
    },
    {
      "epoch": 270.53097345132744,
      "grad_norm": 6.171783866193437e-07,
      "learning_rate": 3.3801963207203366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8760640,
      "step": 15420
    },
    {
      "epoch": 270.6194690265487,
      "grad_norm": 8.57032091516885e-07,
      "learning_rate": 3.379277366020782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8763744,
      "step": 15425
    },
    {
      "epoch": 270.7079646017699,
      "grad_norm": 1.8929370071418816e-06,
      "learning_rate": 3.3783582757256085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8766160,
      "step": 15430
    },
    {
      "epoch": 270.79646017699116,
      "grad_norm": 1.3408608765530516e-06,
      "learning_rate": 3.3774390499765504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8768832,
      "step": 15435
    },
    {
      "epoch": 270.88495575221236,
      "grad_norm": 1.5817959138075821e-06,
      "learning_rate": 3.376519688915364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8771968,
      "step": 15440
    },
    {
      "epoch": 270.9734513274336,
      "grad_norm": 5.397769768933358e-07,
      "learning_rate": 3.3756001926838273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8775184,
      "step": 15445
    },
    {
      "epoch": 271.05309734513276,
      "grad_norm": 4.906142407890002e-07,
      "learning_rate": 3.374680561423737e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8777984,
      "step": 15450
    },
    {
      "epoch": 271.14159292035396,
      "grad_norm": 1.1294732757960446e-06,
      "learning_rate": 3.373760795276912e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8780960,
      "step": 15455
    },
    {
      "epoch": 271.2300884955752,
      "grad_norm": 4.994552682546782e-07,
      "learning_rate": 3.372840894385192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8784320,
      "step": 15460
    },
    {
      "epoch": 271.3185840707965,
      "grad_norm": 9.530830880066787e-07,
      "learning_rate": 3.3719208588904375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8787312,
      "step": 15465
    },
    {
      "epoch": 271.4070796460177,
      "grad_norm": 6.596332013941719e-07,
      "learning_rate": 3.371000688934529e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8790288,
      "step": 15470
    },
    {
      "epoch": 271.49557522123894,
      "grad_norm": 1.5237113757393672e-06,
      "learning_rate": 3.370080384659369e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8792736,
      "step": 15475
    },
    {
      "epoch": 271.5840707964602,
      "grad_norm": 6.999346737757151e-07,
      "learning_rate": 3.36915994620688e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8795216,
      "step": 15480
    },
    {
      "epoch": 271.6725663716814,
      "grad_norm": 1.4608107221647515e-06,
      "learning_rate": 3.3682393737190035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8798064,
      "step": 15485
    },
    {
      "epoch": 271.76106194690266,
      "grad_norm": 7.1567251325177494e-06,
      "learning_rate": 3.3673186673377054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8800928,
      "step": 15490
    },
    {
      "epoch": 271.8495575221239,
      "grad_norm": 6.672054837508767e-07,
      "learning_rate": 3.366397827204969e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8803616,
      "step": 15495
    },
    {
      "epoch": 271.9380530973451,
      "grad_norm": 7.553189220743661e-07,
      "learning_rate": 3.3654768534628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8806624,
      "step": 15500
    },
    {
      "epoch": 272.01769911504425,
      "grad_norm": 3.684883040477871e-06,
      "learning_rate": 3.3645557462532245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8808880,
      "step": 15505
    },
    {
      "epoch": 272.1061946902655,
      "grad_norm": 5.658937993757718e-07,
      "learning_rate": 3.363634505718288e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8812192,
      "step": 15510
    },
    {
      "epoch": 272.1946902654867,
      "grad_norm": 6.873604547763534e-07,
      "learning_rate": 3.362713132000057e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8814656,
      "step": 15515
    },
    {
      "epoch": 272.283185840708,
      "grad_norm": 5.231508453107381e-07,
      "learning_rate": 3.36179162524062e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8817760,
      "step": 15520
    },
    {
      "epoch": 272.37168141592923,
      "grad_norm": 1.1599772733461577e-06,
      "learning_rate": 3.3608699855820846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8820352,
      "step": 15525
    },
    {
      "epoch": 272.46017699115043,
      "grad_norm": 7.108569661795627e-07,
      "learning_rate": 3.359948213166578e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8823968,
      "step": 15530
    },
    {
      "epoch": 272.5486725663717,
      "grad_norm": 4.893118443760613e-07,
      "learning_rate": 3.359026308136252e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8826720,
      "step": 15535
    },
    {
      "epoch": 272.6371681415929,
      "grad_norm": 9.879646540866815e-07,
      "learning_rate": 3.358104270633272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8829792,
      "step": 15540
    },
    {
      "epoch": 272.72566371681415,
      "grad_norm": 8.290688242595934e-07,
      "learning_rate": 3.357182100799831e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8832256,
      "step": 15545
    },
    {
      "epoch": 272.8141592920354,
      "grad_norm": 1.0971989468089305e-06,
      "learning_rate": 3.3562597987781384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8835088,
      "step": 15550
    },
    {
      "epoch": 272.9026548672566,
      "grad_norm": 1.498668325439212e-06,
      "learning_rate": 3.355337364710424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8837888,
      "step": 15555
    },
    {
      "epoch": 272.9911504424779,
      "grad_norm": 1.0825960998772644e-06,
      "learning_rate": 3.354414798738939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8840656,
      "step": 15560
    },
    {
      "epoch": 273.070796460177,
      "grad_norm": 5.833647946928977e-07,
      "learning_rate": 3.353492101005955e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8842656,
      "step": 15565
    },
    {
      "epoch": 273.1592920353982,
      "grad_norm": 8.880616064743663e-07,
      "learning_rate": 3.352569271653763e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8845168,
      "step": 15570
    },
    {
      "epoch": 273.24778761061947,
      "grad_norm": 1.192117224491085e-06,
      "learning_rate": 3.351646310824675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8848032,
      "step": 15575
    },
    {
      "epoch": 273.3362831858407,
      "grad_norm": 1.862271119534853e-06,
      "learning_rate": 3.350723218661023e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8851328,
      "step": 15580
    },
    {
      "epoch": 273.42477876106193,
      "grad_norm": 4.4218421635378036e-07,
      "learning_rate": 3.349799995305162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8854064,
      "step": 15585
    },
    {
      "epoch": 273.5132743362832,
      "grad_norm": 1.4347044725582236e-06,
      "learning_rate": 3.348876640899461e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8856848,
      "step": 15590
    },
    {
      "epoch": 273.60176991150445,
      "grad_norm": 4.69350311504968e-07,
      "learning_rate": 3.3479531555863144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8860368,
      "step": 15595
    },
    {
      "epoch": 273.69026548672565,
      "grad_norm": 1.3705152923648711e-06,
      "learning_rate": 3.3470295395081344e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8863248,
      "step": 15600
    },
    {
      "epoch": 273.69026548672565,
      "eval_loss": 0.5094857811927795,
      "eval_runtime": 1.0601,
      "eval_samples_per_second": 23.583,
      "eval_steps_per_second": 12.263,
      "num_input_tokens_seen": 8863248,
      "step": 15600
    },
    {
      "epoch": 273.7787610619469,
      "grad_norm": 9.82200162980007e-07,
      "learning_rate": 3.3461057928073556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8865936,
      "step": 15605
    },
    {
      "epoch": 273.86725663716817,
      "grad_norm": 2.915595814556582e-06,
      "learning_rate": 3.345181915626431e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8868976,
      "step": 15610
    },
    {
      "epoch": 273.95575221238937,
      "grad_norm": 2.38595043811074e-06,
      "learning_rate": 3.344257908107834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8871712,
      "step": 15615
    },
    {
      "epoch": 274.0353982300885,
      "grad_norm": 7.141329660953488e-07,
      "learning_rate": 3.343333770394058e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8874184,
      "step": 15620
    },
    {
      "epoch": 274.12389380530976,
      "grad_norm": 1.393739580635156e-06,
      "learning_rate": 3.342409502627616e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8877032,
      "step": 15625
    },
    {
      "epoch": 274.21238938053096,
      "grad_norm": 1.1468459888419602e-06,
      "learning_rate": 3.341485104951043e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8879656,
      "step": 15630
    },
    {
      "epoch": 274.3008849557522,
      "grad_norm": 1.6063620478234952e-06,
      "learning_rate": 3.340560577506892e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8882664,
      "step": 15635
    },
    {
      "epoch": 274.3893805309734,
      "grad_norm": 4.2058439930769964e-07,
      "learning_rate": 3.339635920437735e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8885672,
      "step": 15640
    },
    {
      "epoch": 274.4778761061947,
      "grad_norm": 7.600471576552081e-07,
      "learning_rate": 3.338711133886169e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8888840,
      "step": 15645
    },
    {
      "epoch": 274.56637168141594,
      "grad_norm": 4.4720351866089914e-07,
      "learning_rate": 3.3377862179948064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8891448,
      "step": 15650
    },
    {
      "epoch": 274.65486725663715,
      "grad_norm": 3.132570043362648e-07,
      "learning_rate": 3.336861172906281e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8894760,
      "step": 15655
    },
    {
      "epoch": 274.7433628318584,
      "grad_norm": 6.517016686302668e-07,
      "learning_rate": 3.335935998763245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8897688,
      "step": 15660
    },
    {
      "epoch": 274.83185840707966,
      "grad_norm": 1.4741120821781806e-06,
      "learning_rate": 3.3350106957083744e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8900424,
      "step": 15665
    },
    {
      "epoch": 274.92035398230087,
      "grad_norm": 1.8263013998875977e-06,
      "learning_rate": 3.33408526388436e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8903096,
      "step": 15670
    },
    {
      "epoch": 275.0,
      "grad_norm": 9.390805075781827e-07,
      "learning_rate": 3.3331597034339166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8905392,
      "step": 15675
    },
    {
      "epoch": 275.08849557522126,
      "grad_norm": 4.294447819574998e-07,
      "learning_rate": 3.3322340144997764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8907856,
      "step": 15680
    },
    {
      "epoch": 275.17699115044246,
      "grad_norm": 1.1475516430436983e-06,
      "learning_rate": 3.331308197224693e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8910768,
      "step": 15685
    },
    {
      "epoch": 275.2654867256637,
      "grad_norm": 4.4681820554615115e-07,
      "learning_rate": 3.330382251751438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8913536,
      "step": 15690
    },
    {
      "epoch": 275.353982300885,
      "grad_norm": 4.2662847477004107e-07,
      "learning_rate": 3.3294561782228054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8916816,
      "step": 15695
    },
    {
      "epoch": 275.4424778761062,
      "grad_norm": 2.8529791507025948e-06,
      "learning_rate": 3.328529976781607e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8919168,
      "step": 15700
    },
    {
      "epoch": 275.53097345132744,
      "grad_norm": 1.1408376394683728e-06,
      "learning_rate": 3.327603647570673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8922000,
      "step": 15705
    },
    {
      "epoch": 275.6194690265487,
      "grad_norm": 1.2516484275693074e-06,
      "learning_rate": 3.326677190732857e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8924960,
      "step": 15710
    },
    {
      "epoch": 275.7079646017699,
      "grad_norm": 7.087045105436118e-06,
      "learning_rate": 3.325750606411029e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8928016,
      "step": 15715
    },
    {
      "epoch": 275.79646017699116,
      "grad_norm": 6.439099138333404e-07,
      "learning_rate": 3.3248238947480804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8931008,
      "step": 15720
    },
    {
      "epoch": 275.88495575221236,
      "grad_norm": 7.661120093871432e-07,
      "learning_rate": 3.323897055886922e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8933776,
      "step": 15725
    },
    {
      "epoch": 275.9734513274336,
      "grad_norm": 8.387464731640648e-07,
      "learning_rate": 3.322970089970484e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8936688,
      "step": 15730
    },
    {
      "epoch": 276.05309734513276,
      "grad_norm": 2.5153556180157466e-06,
      "learning_rate": 3.3220429971417165e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8938744,
      "step": 15735
    },
    {
      "epoch": 276.14159292035396,
      "grad_norm": 5.852097046954441e-07,
      "learning_rate": 3.321115777543588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8941256,
      "step": 15740
    },
    {
      "epoch": 276.2300884955752,
      "grad_norm": 9.979319202102488e-07,
      "learning_rate": 3.320188431319088e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8943992,
      "step": 15745
    },
    {
      "epoch": 276.3185840707965,
      "grad_norm": 1.3208023119659629e-06,
      "learning_rate": 3.319260958611224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8946424,
      "step": 15750
    },
    {
      "epoch": 276.4070796460177,
      "grad_norm": 1.1044562597817276e-06,
      "learning_rate": 3.3183333595630256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8949752,
      "step": 15755
    },
    {
      "epoch": 276.49557522123894,
      "grad_norm": 2.936084229077096e-07,
      "learning_rate": 3.317405634317538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8953256,
      "step": 15760
    },
    {
      "epoch": 276.5840707964602,
      "grad_norm": 5.597229346676613e-07,
      "learning_rate": 3.3164777830178315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8956360,
      "step": 15765
    },
    {
      "epoch": 276.6725663716814,
      "grad_norm": 7.268486115208361e-07,
      "learning_rate": 3.315549805806989e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8959352,
      "step": 15770
    },
    {
      "epoch": 276.76106194690266,
      "grad_norm": 1.688387214926479e-06,
      "learning_rate": 3.314621702828118e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8962104,
      "step": 15775
    },
    {
      "epoch": 276.8495575221239,
      "grad_norm": 1.3892813512939028e-06,
      "learning_rate": 3.313693474224342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8964984,
      "step": 15780
    },
    {
      "epoch": 276.9380530973451,
      "grad_norm": 2.71526187134441e-06,
      "learning_rate": 3.312765120138809e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8968008,
      "step": 15785
    },
    {
      "epoch": 277.01769911504425,
      "grad_norm": 1.586946041243209e-06,
      "learning_rate": 3.311836640714679e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8970792,
      "step": 15790
    },
    {
      "epoch": 277.1061946902655,
      "grad_norm": 5.979243837828108e-07,
      "learning_rate": 3.310908036095137e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8973832,
      "step": 15795
    },
    {
      "epoch": 277.1946902654867,
      "grad_norm": 1.327471977674577e-06,
      "learning_rate": 3.309979306423386e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8976424,
      "step": 15800
    },
    {
      "epoch": 277.1946902654867,
      "eval_loss": 0.5140652060508728,
      "eval_runtime": 1.0594,
      "eval_samples_per_second": 23.599,
      "eval_steps_per_second": 12.272,
      "num_input_tokens_seen": 8976424,
      "step": 15800
    },
    {
      "epoch": 277.283185840708,
      "grad_norm": 2.4096329980238806e-06,
      "learning_rate": 3.309050451842647e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8978760,
      "step": 15805
    },
    {
      "epoch": 277.37168141592923,
      "grad_norm": 4.4429219769881456e-07,
      "learning_rate": 3.3081214724961604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8981528,
      "step": 15810
    },
    {
      "epoch": 277.46017699115043,
      "grad_norm": 1.915629809445818e-06,
      "learning_rate": 3.307192368527188e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8984568,
      "step": 15815
    },
    {
      "epoch": 277.5486725663717,
      "grad_norm": 2.701150606299052e-06,
      "learning_rate": 3.306263140079008e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8987656,
      "step": 15820
    },
    {
      "epoch": 277.6371681415929,
      "grad_norm": 8.715499575373542e-07,
      "learning_rate": 3.30533378729492e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8990296,
      "step": 15825
    },
    {
      "epoch": 277.72566371681415,
      "grad_norm": 8.635685730951081e-07,
      "learning_rate": 3.304404310318242e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8993656,
      "step": 15830
    },
    {
      "epoch": 277.8141592920354,
      "grad_norm": 1.2038567547278944e-06,
      "learning_rate": 3.3034747092923105e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8996840,
      "step": 15835
    },
    {
      "epoch": 277.9026548672566,
      "grad_norm": 8.482833777634369e-07,
      "learning_rate": 3.3025449843604806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 8999512,
      "step": 15840
    },
    {
      "epoch": 277.9911504424779,
      "grad_norm": 9.658384669819498e-07,
      "learning_rate": 3.30161513566613e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9001832,
      "step": 15845
    },
    {
      "epoch": 278.070796460177,
      "grad_norm": 5.325763936525618e-07,
      "learning_rate": 3.3006851633526506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9004264,
      "step": 15850
    },
    {
      "epoch": 278.1592920353982,
      "grad_norm": 1.0748542536020977e-06,
      "learning_rate": 3.2997550675634584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9006920,
      "step": 15855
    },
    {
      "epoch": 278.24778761061947,
      "grad_norm": 1.0047456271422561e-06,
      "learning_rate": 3.2988248484419825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9009544,
      "step": 15860
    },
    {
      "epoch": 278.3362831858407,
      "grad_norm": 2.0164952729828656e-06,
      "learning_rate": 3.2978945061316776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9012472,
      "step": 15865
    },
    {
      "epoch": 278.42477876106193,
      "grad_norm": 1.089029069589742e-06,
      "learning_rate": 3.296964040776013e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9015304,
      "step": 15870
    },
    {
      "epoch": 278.5132743362832,
      "grad_norm": 5.804213856208662e-07,
      "learning_rate": 3.296033452518478e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9017928,
      "step": 15875
    },
    {
      "epoch": 278.60176991150445,
      "grad_norm": 5.448178512779123e-07,
      "learning_rate": 3.2951027415025806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9021272,
      "step": 15880
    },
    {
      "epoch": 278.69026548672565,
      "grad_norm": 1.473043539590435e-06,
      "learning_rate": 3.294171907871849e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9024472,
      "step": 15885
    },
    {
      "epoch": 278.7787610619469,
      "grad_norm": 9.869465884548845e-07,
      "learning_rate": 3.293240951769828e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9027288,
      "step": 15890
    },
    {
      "epoch": 278.86725663716817,
      "grad_norm": 6.60270245589345e-07,
      "learning_rate": 3.2923098733400846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9030024,
      "step": 15895
    },
    {
      "epoch": 278.95575221238937,
      "grad_norm": 1.1849601833091583e-06,
      "learning_rate": 3.291378672726202e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9032984,
      "step": 15900
    },
    {
      "epoch": 279.0353982300885,
      "grad_norm": 1.1107351838290924e-06,
      "learning_rate": 3.2904473500717824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9035232,
      "step": 15905
    },
    {
      "epoch": 279.12389380530976,
      "grad_norm": 1.1681530622809078e-06,
      "learning_rate": 3.289515905520449e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9037920,
      "step": 15910
    },
    {
      "epoch": 279.21238938053096,
      "grad_norm": 2.2338783765007975e-06,
      "learning_rate": 3.288584339215841e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9040560,
      "step": 15915
    },
    {
      "epoch": 279.3008849557522,
      "grad_norm": 5.726114977733232e-07,
      "learning_rate": 3.287652651301617e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9043664,
      "step": 15920
    },
    {
      "epoch": 279.3893805309734,
      "grad_norm": 1.0143984354726854e-06,
      "learning_rate": 3.286720841921457e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9046544,
      "step": 15925
    },
    {
      "epoch": 279.4778761061947,
      "grad_norm": 1.9484793938318035e-06,
      "learning_rate": 3.285788911219056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9049456,
      "step": 15930
    },
    {
      "epoch": 279.56637168141594,
      "grad_norm": 1.0720398222474614e-06,
      "learning_rate": 3.284856859338131e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9052560,
      "step": 15935
    },
    {
      "epoch": 279.65486725663715,
      "grad_norm": 5.57223359010095e-07,
      "learning_rate": 3.283924686422414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9055312,
      "step": 15940
    },
    {
      "epoch": 279.7433628318584,
      "grad_norm": 9.212334362018737e-07,
      "learning_rate": 3.282992392615659e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9058080,
      "step": 15945
    },
    {
      "epoch": 279.83185840707966,
      "grad_norm": 1.0818927194122807e-06,
      "learning_rate": 3.282059978061638e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9061280,
      "step": 15950
    },
    {
      "epoch": 279.92035398230087,
      "grad_norm": 3.310052989036194e-06,
      "learning_rate": 3.28112744290414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9064128,
      "step": 15955
    },
    {
      "epoch": 280.0,
      "grad_norm": 4.206245193927316e-06,
      "learning_rate": 3.280194787286974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9066520,
      "step": 15960
    },
    {
      "epoch": 280.08849557522126,
      "grad_norm": 7.377299198196852e-07,
      "learning_rate": 3.2792620113539674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9069192,
      "step": 15965
    },
    {
      "epoch": 280.17699115044246,
      "grad_norm": 1.9646247437776765e-06,
      "learning_rate": 3.278329115248966e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9072264,
      "step": 15970
    },
    {
      "epoch": 280.2654867256637,
      "grad_norm": 1.7314649767286028e-06,
      "learning_rate": 3.277396099115834e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9074808,
      "step": 15975
    },
    {
      "epoch": 280.353982300885,
      "grad_norm": 2.378315230089356e-06,
      "learning_rate": 3.276462963098454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9077592,
      "step": 15980
    },
    {
      "epoch": 280.4424778761062,
      "grad_norm": 9.825591860135319e-07,
      "learning_rate": 3.275529707340728e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9080632,
      "step": 15985
    },
    {
      "epoch": 280.53097345132744,
      "grad_norm": 1.579379159011296e-06,
      "learning_rate": 3.274596331986574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9083816,
      "step": 15990
    },
    {
      "epoch": 280.6194690265487,
      "grad_norm": 1.9535707451723283e-06,
      "learning_rate": 3.273662837179932e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9086312,
      "step": 15995
    },
    {
      "epoch": 280.7079646017699,
      "grad_norm": 6.879994884911866e-07,
      "learning_rate": 3.272729223064758e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9088984,
      "step": 16000
    },
    {
      "epoch": 280.7079646017699,
      "eval_loss": 0.5121561288833618,
      "eval_runtime": 1.0581,
      "eval_samples_per_second": 23.628,
      "eval_steps_per_second": 12.287,
      "num_input_tokens_seen": 9088984,
      "step": 16000
    },
    {
      "epoch": 280.79646017699116,
      "grad_norm": 6.510948082905088e-07,
      "learning_rate": 3.2717954897850264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9091704,
      "step": 16005
    },
    {
      "epoch": 280.88495575221236,
      "grad_norm": 7.734411724413803e-07,
      "learning_rate": 3.270861637484733e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9095176,
      "step": 16010
    },
    {
      "epoch": 280.9734513274336,
      "grad_norm": 6.826583103247685e-07,
      "learning_rate": 3.2699276663078867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9098424,
      "step": 16015
    },
    {
      "epoch": 281.05309734513276,
      "grad_norm": 2.175582949348609e-06,
      "learning_rate": 3.268993576398519e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9100800,
      "step": 16020
    },
    {
      "epoch": 281.14159292035396,
      "grad_norm": 3.265126224505366e-06,
      "learning_rate": 3.268059367900678e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9103552,
      "step": 16025
    },
    {
      "epoch": 281.2300884955752,
      "grad_norm": 6.947736892470857e-07,
      "learning_rate": 3.26712504095843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9106368,
      "step": 16030
    },
    {
      "epoch": 281.3185840707965,
      "grad_norm": 8.712603403182584e-07,
      "learning_rate": 3.2661905957158615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9109008,
      "step": 16035
    },
    {
      "epoch": 281.4070796460177,
      "grad_norm": 5.732043177886226e-07,
      "learning_rate": 3.2652560323170734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9112544,
      "step": 16040
    },
    {
      "epoch": 281.49557522123894,
      "grad_norm": 1.016879878079635e-06,
      "learning_rate": 3.264321350906189e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9115280,
      "step": 16045
    },
    {
      "epoch": 281.5840707964602,
      "grad_norm": 1.1426333230701857e-06,
      "learning_rate": 3.263386551627346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9117760,
      "step": 16050
    },
    {
      "epoch": 281.6725663716814,
      "grad_norm": 1.1704610187734943e-06,
      "learning_rate": 3.2624516346247055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9120688,
      "step": 16055
    },
    {
      "epoch": 281.76106194690266,
      "grad_norm": 1.605028387530183e-06,
      "learning_rate": 3.2615166000424404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9123408,
      "step": 16060
    },
    {
      "epoch": 281.8495575221239,
      "grad_norm": 6.492506372524076e-07,
      "learning_rate": 3.260581448024745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9126368,
      "step": 16065
    },
    {
      "epoch": 281.9380530973451,
      "grad_norm": 5.966178378002951e-07,
      "learning_rate": 3.2596461787158335e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9129552,
      "step": 16070
    },
    {
      "epoch": 282.01769911504425,
      "grad_norm": 4.415790044731693e-06,
      "learning_rate": 3.258710792259934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9132240,
      "step": 16075
    },
    {
      "epoch": 282.1061946902655,
      "grad_norm": 1.7843932482719538e-06,
      "learning_rate": 3.257775288801296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9134768,
      "step": 16080
    },
    {
      "epoch": 282.1946902654867,
      "grad_norm": 7.211376100713096e-07,
      "learning_rate": 3.256839668484186e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9137856,
      "step": 16085
    },
    {
      "epoch": 282.283185840708,
      "grad_norm": 1.9664012143039145e-06,
      "learning_rate": 3.255903931452888e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9140448,
      "step": 16090
    },
    {
      "epoch": 282.37168141592923,
      "grad_norm": 1.3420547020359663e-06,
      "learning_rate": 3.2549680778517045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9143200,
      "step": 16095
    },
    {
      "epoch": 282.46017699115043,
      "grad_norm": 1.1700958566507325e-06,
      "learning_rate": 3.2540321078249556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9146016,
      "step": 16100
    },
    {
      "epoch": 282.5486725663717,
      "grad_norm": 8.937895472627133e-07,
      "learning_rate": 3.2530960215169795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9148576,
      "step": 16105
    },
    {
      "epoch": 282.6371681415929,
      "grad_norm": 2.4650180421303958e-06,
      "learning_rate": 3.2521598190721345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9151776,
      "step": 16110
    },
    {
      "epoch": 282.72566371681415,
      "grad_norm": 8.166788916241785e-07,
      "learning_rate": 3.251223500634792e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9154896,
      "step": 16115
    },
    {
      "epoch": 282.8141592920354,
      "grad_norm": 2.583158220659243e-06,
      "learning_rate": 3.2502870663493445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9157744,
      "step": 16120
    },
    {
      "epoch": 282.9026548672566,
      "grad_norm": 7.240138302222476e-07,
      "learning_rate": 3.249350516360203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9160704,
      "step": 16125
    },
    {
      "epoch": 282.9911504424779,
      "grad_norm": 3.151574901494314e-06,
      "learning_rate": 3.248413850811797e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9164096,
      "step": 16130
    },
    {
      "epoch": 283.070796460177,
      "grad_norm": 1.4464650348600117e-06,
      "learning_rate": 3.2474770698485677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9166512,
      "step": 16135
    },
    {
      "epoch": 283.1592920353982,
      "grad_norm": 2.5331978577014524e-06,
      "learning_rate": 3.246540173614983e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9169456,
      "step": 16140
    },
    {
      "epoch": 283.24778761061947,
      "grad_norm": 9.426923384125985e-07,
      "learning_rate": 3.2456031622555197e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9172336,
      "step": 16145
    },
    {
      "epoch": 283.3362831858407,
      "grad_norm": 6.078441856516292e-07,
      "learning_rate": 3.2446660359146794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9174784,
      "step": 16150
    },
    {
      "epoch": 283.42477876106193,
      "grad_norm": 1.038863501889864e-06,
      "learning_rate": 3.2437287947369786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9177456,
      "step": 16155
    },
    {
      "epoch": 283.5132743362832,
      "grad_norm": 9.619453749110107e-07,
      "learning_rate": 3.2427914388669525e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9179952,
      "step": 16160
    },
    {
      "epoch": 283.60176991150445,
      "grad_norm": 7.871205411902338e-07,
      "learning_rate": 3.241853968449151e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9182992,
      "step": 16165
    },
    {
      "epoch": 283.69026548672565,
      "grad_norm": 1.179625201075396e-06,
      "learning_rate": 3.240916383628144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9186336,
      "step": 16170
    },
    {
      "epoch": 283.7787610619469,
      "grad_norm": 1.4247759736463195e-06,
      "learning_rate": 3.239978684548521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9189264,
      "step": 16175
    },
    {
      "epoch": 283.86725663716817,
      "grad_norm": 2.8045587896485813e-06,
      "learning_rate": 3.239040871354885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9192480,
      "step": 16180
    },
    {
      "epoch": 283.95575221238937,
      "grad_norm": 1.093540049623698e-06,
      "learning_rate": 3.2381029441918596e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9195264,
      "step": 16185
    },
    {
      "epoch": 284.0353982300885,
      "grad_norm": 1.9199858343199594e-06,
      "learning_rate": 3.2371649032040845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9197744,
      "step": 16190
    },
    {
      "epoch": 284.12389380530976,
      "grad_norm": 1.1863971849379595e-06,
      "learning_rate": 3.2362267485362174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9200960,
      "step": 16195
    },
    {
      "epoch": 284.21238938053096,
      "grad_norm": 5.169611654309847e-07,
      "learning_rate": 3.235288480332934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9204128,
      "step": 16200
    },
    {
      "epoch": 284.21238938053096,
      "eval_loss": 0.5214510560035706,
      "eval_runtime": 1.0587,
      "eval_samples_per_second": 23.613,
      "eval_steps_per_second": 12.279,
      "num_input_tokens_seen": 9204128,
      "step": 16200
    },
    {
      "epoch": 284.3008849557522,
      "grad_norm": 1.1016830967491842e-06,
      "learning_rate": 3.234350098738927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9207184,
      "step": 16205
    },
    {
      "epoch": 284.3893805309734,
      "grad_norm": 2.0263303213141626e-06,
      "learning_rate": 3.233411603898906e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9210048,
      "step": 16210
    },
    {
      "epoch": 284.4778761061947,
      "grad_norm": 4.700532088008913e-07,
      "learning_rate": 3.232472995957599e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9212576,
      "step": 16215
    },
    {
      "epoch": 284.56637168141594,
      "grad_norm": 1.4840589983577956e-06,
      "learning_rate": 3.231534275059751e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9215312,
      "step": 16220
    },
    {
      "epoch": 284.65486725663715,
      "grad_norm": 3.080850092374021e-06,
      "learning_rate": 3.230595441350125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9217968,
      "step": 16225
    },
    {
      "epoch": 284.7433628318584,
      "grad_norm": 1.733026124384196e-06,
      "learning_rate": 3.2296564949735e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9220736,
      "step": 16230
    },
    {
      "epoch": 284.83185840707966,
      "grad_norm": 2.816967253238545e-06,
      "learning_rate": 3.228717436074675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9223760,
      "step": 16235
    },
    {
      "epoch": 284.92035398230087,
      "grad_norm": 2.1902692424191628e-06,
      "learning_rate": 3.227778264798463e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9226512,
      "step": 16240
    },
    {
      "epoch": 285.0,
      "grad_norm": 3.290863901384e-07,
      "learning_rate": 3.226838981289698e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9229264,
      "step": 16245
    },
    {
      "epoch": 285.08849557522126,
      "grad_norm": 9.287774673794047e-07,
      "learning_rate": 3.225899585693227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9232288,
      "step": 16250
    },
    {
      "epoch": 285.17699115044246,
      "grad_norm": 1.524810841146973e-06,
      "learning_rate": 3.224960078153918e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9235472,
      "step": 16255
    },
    {
      "epoch": 285.2654867256637,
      "grad_norm": 3.2717703106754925e-06,
      "learning_rate": 3.224020458816655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9238064,
      "step": 16260
    },
    {
      "epoch": 285.353982300885,
      "grad_norm": 2.171148707930115e-06,
      "learning_rate": 3.223080727826337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9241216,
      "step": 16265
    },
    {
      "epoch": 285.4424778761062,
      "grad_norm": 1.76386140537943e-06,
      "learning_rate": 3.222140885327885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9243888,
      "step": 16270
    },
    {
      "epoch": 285.53097345132744,
      "grad_norm": 2.7179279982192384e-07,
      "learning_rate": 3.221200931466234e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9247008,
      "step": 16275
    },
    {
      "epoch": 285.6194690265487,
      "grad_norm": 3.0025478281459073e-06,
      "learning_rate": 3.220260866386336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9250032,
      "step": 16280
    },
    {
      "epoch": 285.7079646017699,
      "grad_norm": 8.1753955782915e-07,
      "learning_rate": 3.21932069023316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9252544,
      "step": 16285
    },
    {
      "epoch": 285.79646017699116,
      "grad_norm": 1.0540803714320646e-06,
      "learning_rate": 3.218380403151695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9255056,
      "step": 16290
    },
    {
      "epoch": 285.88495575221236,
      "grad_norm": 4.060673859385133e-07,
      "learning_rate": 3.217440005286943e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9257888,
      "step": 16295
    },
    {
      "epoch": 285.9734513274336,
      "grad_norm": 1.0753738024504855e-06,
      "learning_rate": 3.216499496783928e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9260832,
      "step": 16300
    },
    {
      "epoch": 286.05309734513276,
      "grad_norm": 1.3168991017664666e-06,
      "learning_rate": 3.2155588777876856e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9263472,
      "step": 16305
    },
    {
      "epoch": 286.14159292035396,
      "grad_norm": 1.5268767583620502e-06,
      "learning_rate": 3.214618148443273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9266176,
      "step": 16310
    },
    {
      "epoch": 286.2300884955752,
      "grad_norm": 1.2986088222532999e-06,
      "learning_rate": 3.2136773088957595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9269056,
      "step": 16315
    },
    {
      "epoch": 286.3185840707965,
      "grad_norm": 4.938578399560356e-07,
      "learning_rate": 3.2127363592902374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9271840,
      "step": 16320
    },
    {
      "epoch": 286.4070796460177,
      "grad_norm": 1.778657406248385e-06,
      "learning_rate": 3.211795299771812e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9274848,
      "step": 16325
    },
    {
      "epoch": 286.49557522123894,
      "grad_norm": 4.7398333435921813e-07,
      "learning_rate": 3.210854130485605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9277568,
      "step": 16330
    },
    {
      "epoch": 286.5840707964602,
      "grad_norm": 1.973703319890774e-06,
      "learning_rate": 3.209912851576759e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9280544,
      "step": 16335
    },
    {
      "epoch": 286.6725663716814,
      "grad_norm": 1.1223507954127854e-06,
      "learning_rate": 3.208971463190431e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9283040,
      "step": 16340
    },
    {
      "epoch": 286.76106194690266,
      "grad_norm": 1.051148387887224e-06,
      "learning_rate": 3.208029965471793e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9286048,
      "step": 16345
    },
    {
      "epoch": 286.8495575221239,
      "grad_norm": 1.4751469734619604e-06,
      "learning_rate": 3.2070883585660364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9289392,
      "step": 16350
    },
    {
      "epoch": 286.9380530973451,
      "grad_norm": 2.8729359655699227e-06,
      "learning_rate": 3.20614664261837e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9291952,
      "step": 16355
    },
    {
      "epoch": 287.01769911504425,
      "grad_norm": 3.294242560514249e-06,
      "learning_rate": 3.205204817774016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9294184,
      "step": 16360
    },
    {
      "epoch": 287.1061946902655,
      "grad_norm": 3.5749285416386556e-06,
      "learning_rate": 3.204262884178218e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9297176,
      "step": 16365
    },
    {
      "epoch": 287.1946902654867,
      "grad_norm": 1.6593288592048339e-06,
      "learning_rate": 3.2033208419762314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9299880,
      "step": 16370
    },
    {
      "epoch": 287.283185840708,
      "grad_norm": 6.699500545437331e-07,
      "learning_rate": 3.2023786913133344e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9302536,
      "step": 16375
    },
    {
      "epoch": 287.37168141592923,
      "grad_norm": 6.646531574006076e-07,
      "learning_rate": 3.201436432334816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9305528,
      "step": 16380
    },
    {
      "epoch": 287.46017699115043,
      "grad_norm": 7.596395903419761e-07,
      "learning_rate": 3.2004940651859844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9308488,
      "step": 16385
    },
    {
      "epoch": 287.5486725663717,
      "grad_norm": 1.749114289850695e-06,
      "learning_rate": 3.1995515900121655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9311608,
      "step": 16390
    },
    {
      "epoch": 287.6371681415929,
      "grad_norm": 2.414506923287263e-07,
      "learning_rate": 3.1986090069587e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9314376,
      "step": 16395
    },
    {
      "epoch": 287.72566371681415,
      "grad_norm": 7.056906383695605e-07,
      "learning_rate": 3.1976663161709466e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9317208,
      "step": 16400
    },
    {
      "epoch": 287.72566371681415,
      "eval_loss": 0.5181902647018433,
      "eval_runtime": 1.0627,
      "eval_samples_per_second": 23.526,
      "eval_steps_per_second": 12.233,
      "num_input_tokens_seen": 9317208,
      "step": 16400
    },
    {
      "epoch": 287.8141592920354,
      "grad_norm": 1.9724432149814675e-06,
      "learning_rate": 3.196723517794279e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9320456,
      "step": 16405
    },
    {
      "epoch": 287.9026548672566,
      "grad_norm": 1.1142051334900316e-06,
      "learning_rate": 3.19578061197409e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9323400,
      "step": 16410
    },
    {
      "epoch": 287.9911504424779,
      "grad_norm": 1.4994557204772718e-06,
      "learning_rate": 3.194837598855787e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9325992,
      "step": 16415
    },
    {
      "epoch": 288.070796460177,
      "grad_norm": 1.6261154769381392e-06,
      "learning_rate": 3.193894478584794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9328424,
      "step": 16420
    },
    {
      "epoch": 288.1592920353982,
      "grad_norm": 1.2241856666150852e-06,
      "learning_rate": 3.192951251306553e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9331688,
      "step": 16425
    },
    {
      "epoch": 288.24778761061947,
      "grad_norm": 3.993187874584692e-06,
      "learning_rate": 3.192007917166521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9334360,
      "step": 16430
    },
    {
      "epoch": 288.3362831858407,
      "grad_norm": 1.8169067743656342e-06,
      "learning_rate": 3.191064476310171e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9337096,
      "step": 16435
    },
    {
      "epoch": 288.42477876106193,
      "grad_norm": 7.011648222032818e-07,
      "learning_rate": 3.1901209288829944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9340024,
      "step": 16440
    },
    {
      "epoch": 288.5132743362832,
      "grad_norm": 1.4988012253525085e-06,
      "learning_rate": 3.1891772750304985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9343224,
      "step": 16445
    },
    {
      "epoch": 288.60176991150445,
      "grad_norm": 1.8684812630453962e-06,
      "learning_rate": 3.188233514898206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9345896,
      "step": 16450
    },
    {
      "epoch": 288.69026548672565,
      "grad_norm": 7.103049597390054e-07,
      "learning_rate": 3.187289648631657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9349160,
      "step": 16455
    },
    {
      "epoch": 288.7787610619469,
      "grad_norm": 1.2215951983307605e-06,
      "learning_rate": 3.186345676376406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9351720,
      "step": 16460
    },
    {
      "epoch": 288.86725663716817,
      "grad_norm": 2.053832986348425e-06,
      "learning_rate": 3.1854015982780275e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9353976,
      "step": 16465
    },
    {
      "epoch": 288.95575221238937,
      "grad_norm": 1.409412448083458e-06,
      "learning_rate": 3.1844574144821084e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9356632,
      "step": 16470
    },
    {
      "epoch": 289.0353982300885,
      "grad_norm": 1.3445277318169246e-06,
      "learning_rate": 3.1835131251342554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9359328,
      "step": 16475
    },
    {
      "epoch": 289.12389380530976,
      "grad_norm": 2.883919307805627e-07,
      "learning_rate": 3.182568730380089e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9362272,
      "step": 16480
    },
    {
      "epoch": 289.21238938053096,
      "grad_norm": 9.071256954484852e-07,
      "learning_rate": 3.181624230365245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9364960,
      "step": 16485
    },
    {
      "epoch": 289.3008849557522,
      "grad_norm": 1.2151734836152173e-06,
      "learning_rate": 3.180679625235381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9368096,
      "step": 16490
    },
    {
      "epoch": 289.3893805309734,
      "grad_norm": 1.7367746067975531e-06,
      "learning_rate": 3.1797349151361646e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9371136,
      "step": 16495
    },
    {
      "epoch": 289.4778761061947,
      "grad_norm": 3.1797955557522073e-07,
      "learning_rate": 3.178790100213281e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9373904,
      "step": 16500
    },
    {
      "epoch": 289.56637168141594,
      "grad_norm": 4.639967983166571e-07,
      "learning_rate": 3.1778451806124346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9376608,
      "step": 16505
    },
    {
      "epoch": 289.65486725663715,
      "grad_norm": 2.229687424915028e-06,
      "learning_rate": 3.176900156479342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9379040,
      "step": 16510
    },
    {
      "epoch": 289.7433628318584,
      "grad_norm": 7.149408702389337e-06,
      "learning_rate": 3.17595502795974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9381904,
      "step": 16515
    },
    {
      "epoch": 289.83185840707966,
      "grad_norm": 9.987057865146198e-07,
      "learning_rate": 3.175009795199377e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9384944,
      "step": 16520
    },
    {
      "epoch": 289.92035398230087,
      "grad_norm": 7.731516120657034e-07,
      "learning_rate": 3.1740644583440224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9388400,
      "step": 16525
    },
    {
      "epoch": 290.0,
      "grad_norm": 5.602830128736969e-07,
      "learning_rate": 3.173119017539457e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9390648,
      "step": 16530
    },
    {
      "epoch": 290.08849557522126,
      "grad_norm": 1.3969643077871297e-06,
      "learning_rate": 3.172173472931479e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9393784,
      "step": 16535
    },
    {
      "epoch": 290.17699115044246,
      "grad_norm": 1.237016249433509e-06,
      "learning_rate": 3.1712278246659055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9396680,
      "step": 16540
    },
    {
      "epoch": 290.2654867256637,
      "grad_norm": 1.0807358421516255e-06,
      "learning_rate": 3.170282072888566e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9399784,
      "step": 16545
    },
    {
      "epoch": 290.353982300885,
      "grad_norm": 2.112258925990318e-06,
      "learning_rate": 3.169336217745307e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9402632,
      "step": 16550
    },
    {
      "epoch": 290.4424778761062,
      "grad_norm": 2.974258904941962e-06,
      "learning_rate": 3.1683902593819924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9405368,
      "step": 16555
    },
    {
      "epoch": 290.53097345132744,
      "grad_norm": 1.683799382590223e-06,
      "learning_rate": 3.1674441979445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9408456,
      "step": 16560
    },
    {
      "epoch": 290.6194690265487,
      "grad_norm": 6.875616236357018e-07,
      "learning_rate": 3.166498033578725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9410920,
      "step": 16565
    },
    {
      "epoch": 290.7079646017699,
      "grad_norm": 1.551527930132579e-06,
      "learning_rate": 3.165551766430578e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9414024,
      "step": 16570
    },
    {
      "epoch": 290.79646017699116,
      "grad_norm": 8.887794251677406e-07,
      "learning_rate": 3.164605396645984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9417144,
      "step": 16575
    },
    {
      "epoch": 290.88495575221236,
      "grad_norm": 1.4861410591038293e-06,
      "learning_rate": 3.163658924370886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9420024,
      "step": 16580
    },
    {
      "epoch": 290.9734513274336,
      "grad_norm": 8.37245579532464e-07,
      "learning_rate": 3.1627123497512415e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9422488,
      "step": 16585
    },
    {
      "epoch": 291.05309734513276,
      "grad_norm": 4.700165732174355e-07,
      "learning_rate": 3.1617656729330245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9424856,
      "step": 16590
    },
    {
      "epoch": 291.14159292035396,
      "grad_norm": 1.2727825833280804e-06,
      "learning_rate": 3.1608188940622255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9427880,
      "step": 16595
    },
    {
      "epoch": 291.2300884955752,
      "grad_norm": 2.6999975943908794e-06,
      "learning_rate": 3.159872013284847e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9431208,
      "step": 16600
    },
    {
      "epoch": 291.2300884955752,
      "eval_loss": 0.5424460172653198,
      "eval_runtime": 1.0625,
      "eval_samples_per_second": 23.53,
      "eval_steps_per_second": 12.236,
      "num_input_tokens_seen": 9431208,
      "step": 16600
    },
    {
      "epoch": 291.3185840707965,
      "grad_norm": 1.9147748844261514e-06,
      "learning_rate": 3.1589250307469134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9433976,
      "step": 16605
    },
    {
      "epoch": 291.4070796460177,
      "grad_norm": 2.2888766579853836e-06,
      "learning_rate": 3.1579779465944586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9436232,
      "step": 16610
    },
    {
      "epoch": 291.49557522123894,
      "grad_norm": 8.069893624451652e-07,
      "learning_rate": 3.1570307609735363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9439112,
      "step": 16615
    },
    {
      "epoch": 291.5840707964602,
      "grad_norm": 1.8778164303512312e-06,
      "learning_rate": 3.156083474030213e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9441736,
      "step": 16620
    },
    {
      "epoch": 291.6725663716814,
      "grad_norm": 2.785393917292822e-06,
      "learning_rate": 3.155136085910573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9444504,
      "step": 16625
    },
    {
      "epoch": 291.76106194690266,
      "grad_norm": 9.512132805866713e-07,
      "learning_rate": 3.154188596760717e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9447496,
      "step": 16630
    },
    {
      "epoch": 291.8495575221239,
      "grad_norm": 1.0330985560358386e-06,
      "learning_rate": 3.153241006726757e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9450648,
      "step": 16635
    },
    {
      "epoch": 291.9380530973451,
      "grad_norm": 3.1681886412116e-06,
      "learning_rate": 3.152293315954825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9453880,
      "step": 16640
    },
    {
      "epoch": 292.01769911504425,
      "grad_norm": 6.674888481938979e-07,
      "learning_rate": 3.1513455245910666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9455992,
      "step": 16645
    },
    {
      "epoch": 292.1061946902655,
      "grad_norm": 9.961535170077696e-07,
      "learning_rate": 3.150397632781643e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9458872,
      "step": 16650
    },
    {
      "epoch": 292.1946902654867,
      "grad_norm": 1.196546122628206e-06,
      "learning_rate": 3.149449640672731e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9461512,
      "step": 16655
    },
    {
      "epoch": 292.283185840708,
      "grad_norm": 1.9584231267799623e-06,
      "learning_rate": 3.148501548410523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9464216,
      "step": 16660
    },
    {
      "epoch": 292.37168141592923,
      "grad_norm": 7.180689749475277e-07,
      "learning_rate": 3.1475533561412256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9467160,
      "step": 16665
    },
    {
      "epoch": 292.46017699115043,
      "grad_norm": 2.4580242552474374e-06,
      "learning_rate": 3.146605064011065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9469880,
      "step": 16670
    },
    {
      "epoch": 292.5486725663717,
      "grad_norm": 6.889258656883612e-07,
      "learning_rate": 3.145656672166277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9472392,
      "step": 16675
    },
    {
      "epoch": 292.6371681415929,
      "grad_norm": 1.1569263733690605e-06,
      "learning_rate": 3.144708180753116e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9475096,
      "step": 16680
    },
    {
      "epoch": 292.72566371681415,
      "grad_norm": 1.1412012099754065e-06,
      "learning_rate": 3.143759589917851e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9478024,
      "step": 16685
    },
    {
      "epoch": 292.8141592920354,
      "grad_norm": 1.2834493645641487e-06,
      "learning_rate": 3.142810899806768e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9481416,
      "step": 16690
    },
    {
      "epoch": 292.9026548672566,
      "grad_norm": 7.805026598362019e-07,
      "learning_rate": 3.141862110566166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9484360,
      "step": 16695
    },
    {
      "epoch": 292.9911504424779,
      "grad_norm": 7.916277127151261e-07,
      "learning_rate": 3.1409132223423606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9487320,
      "step": 16700
    },
    {
      "epoch": 293.070796460177,
      "grad_norm": 1.4731695046066307e-06,
      "learning_rate": 3.139964235281682e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9489576,
      "step": 16705
    },
    {
      "epoch": 293.1592920353982,
      "grad_norm": 1.9928672827518312e-06,
      "learning_rate": 3.139015149530476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9492840,
      "step": 16710
    },
    {
      "epoch": 293.24778761061947,
      "grad_norm": 1.926946424646303e-06,
      "learning_rate": 3.1380659652351034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9495496,
      "step": 16715
    },
    {
      "epoch": 293.3362831858407,
      "grad_norm": 8.961255275607982e-07,
      "learning_rate": 3.137116682541941e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9498504,
      "step": 16720
    },
    {
      "epoch": 293.42477876106193,
      "grad_norm": 1.6759108802943956e-06,
      "learning_rate": 3.136167301597379e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9501544,
      "step": 16725
    },
    {
      "epoch": 293.5132743362832,
      "grad_norm": 1.2324686622378067e-06,
      "learning_rate": 3.1352178225478254e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9504392,
      "step": 16730
    },
    {
      "epoch": 293.60176991150445,
      "grad_norm": 1.2336273584878654e-06,
      "learning_rate": 3.1342682455396996e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9507256,
      "step": 16735
    },
    {
      "epoch": 293.69026548672565,
      "grad_norm": 4.0669905843060405e-07,
      "learning_rate": 3.133318570719441e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9510008,
      "step": 16740
    },
    {
      "epoch": 293.7787610619469,
      "grad_norm": 1.0269592394251958e-06,
      "learning_rate": 3.132368798233499e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9513064,
      "step": 16745
    },
    {
      "epoch": 293.86725663716817,
      "grad_norm": 2.331122686882736e-06,
      "learning_rate": 3.131418928228342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9516232,
      "step": 16750
    },
    {
      "epoch": 293.95575221238937,
      "grad_norm": 2.3443471945938654e-06,
      "learning_rate": 3.1304689608504514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9518824,
      "step": 16755
    },
    {
      "epoch": 294.0353982300885,
      "grad_norm": 6.51505956739129e-07,
      "learning_rate": 3.129518896246324e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9521096,
      "step": 16760
    },
    {
      "epoch": 294.12389380530976,
      "grad_norm": 6.606578608625568e-07,
      "learning_rate": 3.128568734562472e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9524568,
      "step": 16765
    },
    {
      "epoch": 294.21238938053096,
      "grad_norm": 1.6240217064478202e-06,
      "learning_rate": 3.127618475945421e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9527400,
      "step": 16770
    },
    {
      "epoch": 294.3008849557522,
      "grad_norm": 1.0096144933413598e-06,
      "learning_rate": 3.126668120541715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9530360,
      "step": 16775
    },
    {
      "epoch": 294.3893805309734,
      "grad_norm": 3.15551255880564e-06,
      "learning_rate": 3.1257176684979096e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9533032,
      "step": 16780
    },
    {
      "epoch": 294.4778761061947,
      "grad_norm": 1.942369863172644e-06,
      "learning_rate": 3.124767119960576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9535704,
      "step": 16785
    },
    {
      "epoch": 294.56637168141594,
      "grad_norm": 1.7425958276362508e-06,
      "learning_rate": 3.123816475076301e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9538488,
      "step": 16790
    },
    {
      "epoch": 294.65486725663715,
      "grad_norm": 7.977762379596243e-07,
      "learning_rate": 3.122865733991687e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9541448,
      "step": 16795
    },
    {
      "epoch": 294.7433628318584,
      "grad_norm": 7.011960860836552e-07,
      "learning_rate": 3.1219148968533486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9544328,
      "step": 16800
    },
    {
      "epoch": 294.7433628318584,
      "eval_loss": 0.5420147180557251,
      "eval_runtime": 1.0634,
      "eval_samples_per_second": 23.51,
      "eval_steps_per_second": 12.225,
      "num_input_tokens_seen": 9544328,
      "step": 16800
    },
    {
      "epoch": 294.83185840707966,
      "grad_norm": 9.625637176213786e-07,
      "learning_rate": 3.120963963807918e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9547144,
      "step": 16805
    },
    {
      "epoch": 294.92035398230087,
      "grad_norm": 8.763754522078671e-07,
      "learning_rate": 3.12001293500204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9549912,
      "step": 16810
    },
    {
      "epoch": 295.0,
      "grad_norm": 4.11900828112266e-07,
      "learning_rate": 3.1190618105823765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9552440,
      "step": 16815
    },
    {
      "epoch": 295.08849557522126,
      "grad_norm": 4.5300052420316206e-07,
      "learning_rate": 3.118110590695603e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9554984,
      "step": 16820
    },
    {
      "epoch": 295.17699115044246,
      "grad_norm": 1.5192769069471979e-06,
      "learning_rate": 3.117159275488407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9557992,
      "step": 16825
    },
    {
      "epoch": 295.2654867256637,
      "grad_norm": 2.3455584141629515e-06,
      "learning_rate": 3.1162078651074956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9560760,
      "step": 16830
    },
    {
      "epoch": 295.353982300885,
      "grad_norm": 8.894267580217274e-07,
      "learning_rate": 3.1152563596995885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9563624,
      "step": 16835
    },
    {
      "epoch": 295.4424778761062,
      "grad_norm": 2.0225104435667163e-06,
      "learning_rate": 3.1143047594114186e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9566248,
      "step": 16840
    },
    {
      "epoch": 295.53097345132744,
      "grad_norm": 1.5459144151463988e-06,
      "learning_rate": 3.113353064389734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9568904,
      "step": 16845
    },
    {
      "epoch": 295.6194690265487,
      "grad_norm": 7.020720431683003e-07,
      "learning_rate": 3.1124012747812993e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9571656,
      "step": 16850
    },
    {
      "epoch": 295.7079646017699,
      "grad_norm": 6.530772225232795e-07,
      "learning_rate": 3.1114493907328936e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9574584,
      "step": 16855
    },
    {
      "epoch": 295.79646017699116,
      "grad_norm": 1.4473524743152666e-06,
      "learning_rate": 3.110497412391306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9577592,
      "step": 16860
    },
    {
      "epoch": 295.88495575221236,
      "grad_norm": 4.132718061100604e-07,
      "learning_rate": 3.1095453399033466e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9580920,
      "step": 16865
    },
    {
      "epoch": 295.9734513274336,
      "grad_norm": 1.4663019101135433e-06,
      "learning_rate": 3.108593173415835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9583784,
      "step": 16870
    },
    {
      "epoch": 296.05309734513276,
      "grad_norm": 8.716729098523501e-07,
      "learning_rate": 3.107640913075609e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9586176,
      "step": 16875
    },
    {
      "epoch": 296.14159292035396,
      "grad_norm": 7.744275762888719e-07,
      "learning_rate": 3.106688559029517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9588896,
      "step": 16880
    },
    {
      "epoch": 296.2300884955752,
      "grad_norm": 8.62234060150513e-07,
      "learning_rate": 3.105736111424425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9591328,
      "step": 16885
    },
    {
      "epoch": 296.3185840707965,
      "grad_norm": 1.409135165886255e-06,
      "learning_rate": 3.1047835704072136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9593808,
      "step": 16890
    },
    {
      "epoch": 296.4070796460177,
      "grad_norm": 1.2875242418886046e-06,
      "learning_rate": 3.103830936124775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9596912,
      "step": 16895
    },
    {
      "epoch": 296.49557522123894,
      "grad_norm": 4.6794153263363114e-07,
      "learning_rate": 3.102878208724018e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9600096,
      "step": 16900
    },
    {
      "epoch": 296.5840707964602,
      "grad_norm": 9.683687949291198e-07,
      "learning_rate": 3.101925388351865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9602528,
      "step": 16905
    },
    {
      "epoch": 296.6725663716814,
      "grad_norm": 9.829885811996064e-07,
      "learning_rate": 3.1009724751552515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9605296,
      "step": 16910
    },
    {
      "epoch": 296.76106194690266,
      "grad_norm": 2.017098722717492e-06,
      "learning_rate": 3.100019469281131e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9608256,
      "step": 16915
    },
    {
      "epoch": 296.8495575221239,
      "grad_norm": 4.453855808606022e-07,
      "learning_rate": 3.0990663708764685e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9611536,
      "step": 16920
    },
    {
      "epoch": 296.9380530973451,
      "grad_norm": 1.299985456171271e-06,
      "learning_rate": 3.098113180088243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9614832,
      "step": 16925
    },
    {
      "epoch": 297.01769911504425,
      "grad_norm": 2.709610271267593e-06,
      "learning_rate": 3.097159897063448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9617344,
      "step": 16930
    },
    {
      "epoch": 297.1061946902655,
      "grad_norm": 2.05895526050881e-06,
      "learning_rate": 3.096206521949094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9620544,
      "step": 16935
    },
    {
      "epoch": 297.1946902654867,
      "grad_norm": 1.697064476502419e-06,
      "learning_rate": 3.0952530548922006e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9623280,
      "step": 16940
    },
    {
      "epoch": 297.283185840708,
      "grad_norm": 5.667947675647156e-07,
      "learning_rate": 3.0942994960398064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9626464,
      "step": 16945
    },
    {
      "epoch": 297.37168141592923,
      "grad_norm": 1.3473977560352068e-06,
      "learning_rate": 3.093345845538961e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9629072,
      "step": 16950
    },
    {
      "epoch": 297.46017699115043,
      "grad_norm": 6.550836815222283e-07,
      "learning_rate": 3.09239210353673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9631872,
      "step": 16955
    },
    {
      "epoch": 297.5486725663717,
      "grad_norm": 4.29609883667581e-07,
      "learning_rate": 3.0914382701801926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9634640,
      "step": 16960
    },
    {
      "epoch": 297.6371681415929,
      "grad_norm": 7.461546260856267e-07,
      "learning_rate": 3.090484345616441e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9637360,
      "step": 16965
    },
    {
      "epoch": 297.72566371681415,
      "grad_norm": 6.654567528130428e-07,
      "learning_rate": 3.0895303299925825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9640224,
      "step": 16970
    },
    {
      "epoch": 297.8141592920354,
      "grad_norm": 3.721112591392739e-07,
      "learning_rate": 3.0885762234557393e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9642800,
      "step": 16975
    },
    {
      "epoch": 297.9026548672566,
      "grad_norm": 1.0182983487538877e-06,
      "learning_rate": 3.087622026153045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9645600,
      "step": 16980
    },
    {
      "epoch": 297.9911504424779,
      "grad_norm": 6.469166464739828e-07,
      "learning_rate": 3.086667738231651e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9648912,
      "step": 16985
    },
    {
      "epoch": 298.070796460177,
      "grad_norm": 9.128315241468954e-07,
      "learning_rate": 3.085713359838718e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9651656,
      "step": 16990
    },
    {
      "epoch": 298.1592920353982,
      "grad_norm": 5.347613409867336e-07,
      "learning_rate": 3.084758891121425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9654552,
      "step": 16995
    },
    {
      "epoch": 298.24778761061947,
      "grad_norm": 1.8492423805582803e-06,
      "learning_rate": 3.083804332226963e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9657432,
      "step": 17000
    },
    {
      "epoch": 298.24778761061947,
      "eval_loss": 0.5455121397972107,
      "eval_runtime": 1.0609,
      "eval_samples_per_second": 23.566,
      "eval_steps_per_second": 12.254,
      "num_input_tokens_seen": 9657432,
      "step": 17000
    },
    {
      "epoch": 298.3362831858407,
      "grad_norm": 5.498915243151714e-07,
      "learning_rate": 3.082849683302536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9660344,
      "step": 17005
    },
    {
      "epoch": 298.42477876106193,
      "grad_norm": 9.933111186910537e-07,
      "learning_rate": 3.081894944495363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9662776,
      "step": 17010
    },
    {
      "epoch": 298.5132743362832,
      "grad_norm": 5.403323370956059e-07,
      "learning_rate": 3.080940115952677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9665656,
      "step": 17015
    },
    {
      "epoch": 298.60176991150445,
      "grad_norm": 8.967726898845285e-07,
      "learning_rate": 3.0799851978217245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9668296,
      "step": 17020
    },
    {
      "epoch": 298.69026548672565,
      "grad_norm": 9.69932784755656e-07,
      "learning_rate": 3.0790301902497666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9671640,
      "step": 17025
    },
    {
      "epoch": 298.7787610619469,
      "grad_norm": 4.204391643725103e-06,
      "learning_rate": 3.078075093384076e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9674872,
      "step": 17030
    },
    {
      "epoch": 298.86725663716817,
      "grad_norm": 1.2070097454852657e-06,
      "learning_rate": 3.077119907371942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9677672,
      "step": 17035
    },
    {
      "epoch": 298.95575221238937,
      "grad_norm": 1.5155591199800256e-06,
      "learning_rate": 3.076164632360666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9680264,
      "step": 17040
    },
    {
      "epoch": 299.0353982300885,
      "grad_norm": 9.63521983976534e-07,
      "learning_rate": 3.075209268497563e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9682552,
      "step": 17045
    },
    {
      "epoch": 299.12389380530976,
      "grad_norm": 2.9229561278043548e-06,
      "learning_rate": 3.074253815929961e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9685464,
      "step": 17050
    },
    {
      "epoch": 299.21238938053096,
      "grad_norm": 5.414025508798659e-06,
      "learning_rate": 3.0732982748052054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9688632,
      "step": 17055
    },
    {
      "epoch": 299.3008849557522,
      "grad_norm": 1.3223146879681735e-06,
      "learning_rate": 3.072342645270651e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9691288,
      "step": 17060
    },
    {
      "epoch": 299.3893805309734,
      "grad_norm": 4.389420666939259e-07,
      "learning_rate": 3.071386927473668e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9694168,
      "step": 17065
    },
    {
      "epoch": 299.4778761061947,
      "grad_norm": 4.4620682615459373e-07,
      "learning_rate": 3.0704311215616404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9697080,
      "step": 17070
    },
    {
      "epoch": 299.56637168141594,
      "grad_norm": 1.0112722748090164e-06,
      "learning_rate": 3.0694752276819656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9699832,
      "step": 17075
    },
    {
      "epoch": 299.65486725663715,
      "grad_norm": 1.0772970426842221e-06,
      "learning_rate": 3.068519245982054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9702712,
      "step": 17080
    },
    {
      "epoch": 299.7433628318584,
      "grad_norm": 1.6011496200007969e-06,
      "learning_rate": 3.0675631766093304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9705416,
      "step": 17085
    },
    {
      "epoch": 299.83185840707966,
      "grad_norm": 1.6218845075854915e-06,
      "learning_rate": 3.066607019711232e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9708824,
      "step": 17090
    },
    {
      "epoch": 299.92035398230087,
      "grad_norm": 6.068560765015718e-07,
      "learning_rate": 3.065650775435211e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9711688,
      "step": 17095
    },
    {
      "epoch": 300.0,
      "grad_norm": 5.762003638665192e-07,
      "learning_rate": 3.0646944439287326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9713848,
      "step": 17100
    },
    {
      "epoch": 300.08849557522126,
      "grad_norm": 7.975150992933777e-07,
      "learning_rate": 3.0637380253392736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9716568,
      "step": 17105
    },
    {
      "epoch": 300.17699115044246,
      "grad_norm": 7.0545734160987195e-06,
      "learning_rate": 3.062781519814327e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9719288,
      "step": 17110
    },
    {
      "epoch": 300.2654867256637,
      "grad_norm": 1.43282250064658e-06,
      "learning_rate": 3.0618249275013985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9722344,
      "step": 17115
    },
    {
      "epoch": 300.353982300885,
      "grad_norm": 8.77478498750861e-07,
      "learning_rate": 3.060868248548005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9725496,
      "step": 17120
    },
    {
      "epoch": 300.4424778761062,
      "grad_norm": 1.543543248772039e-06,
      "learning_rate": 3.0599114831016796e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9728456,
      "step": 17125
    },
    {
      "epoch": 300.53097345132744,
      "grad_norm": 6.093245588090213e-07,
      "learning_rate": 3.0589546313099666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9731016,
      "step": 17130
    },
    {
      "epoch": 300.6194690265487,
      "grad_norm": 1.656416202422406e-06,
      "learning_rate": 3.0579976933204255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9734408,
      "step": 17135
    },
    {
      "epoch": 300.7079646017699,
      "grad_norm": 6.623265562666347e-06,
      "learning_rate": 3.0570406692806284e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9737176,
      "step": 17140
    },
    {
      "epoch": 300.79646017699116,
      "grad_norm": 3.4044796848320402e-06,
      "learning_rate": 3.05608355933816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9739896,
      "step": 17145
    },
    {
      "epoch": 300.88495575221236,
      "grad_norm": 8.39759991322353e-07,
      "learning_rate": 3.055126363640618e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9742952,
      "step": 17150
    },
    {
      "epoch": 300.9734513274336,
      "grad_norm": 1.4454433312494075e-06,
      "learning_rate": 3.0541690823356146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9745896,
      "step": 17155
    },
    {
      "epoch": 301.05309734513276,
      "grad_norm": 5.751095955019991e-07,
      "learning_rate": 3.053211715570775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9748024,
      "step": 17160
    },
    {
      "epoch": 301.14159292035396,
      "grad_norm": 1.428059931640746e-06,
      "learning_rate": 3.052254263493736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9750664,
      "step": 17165
    },
    {
      "epoch": 301.2300884955752,
      "grad_norm": 8.896070085029351e-07,
      "learning_rate": 3.0512967262521498e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9753512,
      "step": 17170
    },
    {
      "epoch": 301.3185840707965,
      "grad_norm": 9.797839766179095e-07,
      "learning_rate": 3.0503391039936803e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9756664,
      "step": 17175
    },
    {
      "epoch": 301.4070796460177,
      "grad_norm": 2.993472890011617e-06,
      "learning_rate": 3.0493813968660056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9759736,
      "step": 17180
    },
    {
      "epoch": 301.49557522123894,
      "grad_norm": 2.7742366910388228e-06,
      "learning_rate": 3.0484236050168153e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9762728,
      "step": 17185
    },
    {
      "epoch": 301.5840707964602,
      "grad_norm": 6.324713922367664e-06,
      "learning_rate": 3.0474657285938123e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9765288,
      "step": 17190
    },
    {
      "epoch": 301.6725663716814,
      "grad_norm": 3.8881285036040936e-07,
      "learning_rate": 3.046507767744715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9768312,
      "step": 17195
    },
    {
      "epoch": 301.76106194690266,
      "grad_norm": 5.05826756125316e-06,
      "learning_rate": 3.045549722617252e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9770824,
      "step": 17200
    },
    {
      "epoch": 301.76106194690266,
      "eval_loss": 0.555571973323822,
      "eval_runtime": 1.0608,
      "eval_samples_per_second": 23.568,
      "eval_steps_per_second": 12.255,
      "num_input_tokens_seen": 9770824,
      "step": 17200
    },
    {
      "epoch": 301.8495575221239,
      "grad_norm": 2.541927187849069e-06,
      "learning_rate": 3.0445915933591658e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9773640,
      "step": 17205
    },
    {
      "epoch": 301.9380530973451,
      "grad_norm": 6.767741069779731e-07,
      "learning_rate": 3.0436333801182114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9777448,
      "step": 17210
    },
    {
      "epoch": 302.01769911504425,
      "grad_norm": 5.456876692733204e-07,
      "learning_rate": 3.0426750830421596e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9779976,
      "step": 17215
    },
    {
      "epoch": 302.1061946902655,
      "grad_norm": 7.923749762994703e-07,
      "learning_rate": 3.0417167022787897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9782440,
      "step": 17220
    },
    {
      "epoch": 302.1946902654867,
      "grad_norm": 7.331828442147525e-07,
      "learning_rate": 3.0407582379758966e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9785128,
      "step": 17225
    },
    {
      "epoch": 302.283185840708,
      "grad_norm": 1.4808546211497742e-06,
      "learning_rate": 3.039799690281287e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9788056,
      "step": 17230
    },
    {
      "epoch": 302.37168141592923,
      "grad_norm": 1.0272165127389599e-06,
      "learning_rate": 3.0388410593427823e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9791032,
      "step": 17235
    },
    {
      "epoch": 302.46017699115043,
      "grad_norm": 1.7385783621648443e-06,
      "learning_rate": 3.0378823453082146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9793784,
      "step": 17240
    },
    {
      "epoch": 302.5486725663717,
      "grad_norm": 1.5058875533213723e-06,
      "learning_rate": 3.03692354832543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9796808,
      "step": 17245
    },
    {
      "epoch": 302.6371681415929,
      "grad_norm": 5.92186950143514e-07,
      "learning_rate": 3.0359646685422865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9799528,
      "step": 17250
    },
    {
      "epoch": 302.72566371681415,
      "grad_norm": 1.5782187574586715e-06,
      "learning_rate": 3.035005706106656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9802040,
      "step": 17255
    },
    {
      "epoch": 302.8141592920354,
      "grad_norm": 2.7745597890316276e-06,
      "learning_rate": 3.034046661166422e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9804648,
      "step": 17260
    },
    {
      "epoch": 302.9026548672566,
      "grad_norm": 5.717251383430266e-07,
      "learning_rate": 3.033087533869482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9807784,
      "step": 17265
    },
    {
      "epoch": 302.9911504424779,
      "grad_norm": 4.961721629115345e-07,
      "learning_rate": 3.0321283243637444e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9811080,
      "step": 17270
    },
    {
      "epoch": 303.070796460177,
      "grad_norm": 1.1028567996618222e-06,
      "learning_rate": 3.0311690327971326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9813280,
      "step": 17275
    },
    {
      "epoch": 303.1592920353982,
      "grad_norm": 5.468065182867576e-07,
      "learning_rate": 3.030209659317581e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9816496,
      "step": 17280
    },
    {
      "epoch": 303.24778761061947,
      "grad_norm": 1.5382288438559044e-06,
      "learning_rate": 3.0292502040730362e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9819424,
      "step": 17285
    },
    {
      "epoch": 303.3362831858407,
      "grad_norm": 3.158313575113425e-06,
      "learning_rate": 3.0282906672114597e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9822464,
      "step": 17290
    },
    {
      "epoch": 303.42477876106193,
      "grad_norm": 8.272104992101958e-07,
      "learning_rate": 3.027331048880823e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9825504,
      "step": 17295
    },
    {
      "epoch": 303.5132743362832,
      "grad_norm": 2.1776156700070715e-06,
      "learning_rate": 3.0263713492291123e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9828240,
      "step": 17300
    },
    {
      "epoch": 303.60176991150445,
      "grad_norm": 1.3485008594216197e-06,
      "learning_rate": 3.0254115684043242e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9831120,
      "step": 17305
    },
    {
      "epoch": 303.69026548672565,
      "grad_norm": 3.0781643545196857e-07,
      "learning_rate": 3.024451706554469e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9834368,
      "step": 17310
    },
    {
      "epoch": 303.7787610619469,
      "grad_norm": 1.623189064048347e-06,
      "learning_rate": 3.0234917638275705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9836976,
      "step": 17315
    },
    {
      "epoch": 303.86725663716817,
      "grad_norm": 2.522536078686244e-06,
      "learning_rate": 3.0225317403716635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9839680,
      "step": 17320
    },
    {
      "epoch": 303.95575221238937,
      "grad_norm": 3.871222816087538e-06,
      "learning_rate": 3.0215716363347956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9842240,
      "step": 17325
    },
    {
      "epoch": 304.0353982300885,
      "grad_norm": 9.100845659304468e-07,
      "learning_rate": 3.0206114518650275e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9844824,
      "step": 17330
    },
    {
      "epoch": 304.12389380530976,
      "grad_norm": 6.344582743622595e-07,
      "learning_rate": 3.0196511871104304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9847880,
      "step": 17335
    },
    {
      "epoch": 304.21238938053096,
      "grad_norm": 1.8167991129303118e-06,
      "learning_rate": 3.01869084221909e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9850632,
      "step": 17340
    },
    {
      "epoch": 304.3008849557522,
      "grad_norm": 1.5708576484030345e-06,
      "learning_rate": 3.0177304173391037e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9853176,
      "step": 17345
    },
    {
      "epoch": 304.3893805309734,
      "grad_norm": 1.5493836826863117e-06,
      "learning_rate": 3.01676991261858e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9855736,
      "step": 17350
    },
    {
      "epoch": 304.4778761061947,
      "grad_norm": 3.2428863505629124e-06,
      "learning_rate": 3.015809328205642e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9858520,
      "step": 17355
    },
    {
      "epoch": 304.56637168141594,
      "grad_norm": 2.1159285097382963e-06,
      "learning_rate": 3.0148486642484248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9861288,
      "step": 17360
    },
    {
      "epoch": 304.65486725663715,
      "grad_norm": 1.4329273199109593e-06,
      "learning_rate": 3.0138879208950722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9863976,
      "step": 17365
    },
    {
      "epoch": 304.7433628318584,
      "grad_norm": 1.8912372752311057e-06,
      "learning_rate": 3.012927098293744e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9867192,
      "step": 17370
    },
    {
      "epoch": 304.83185840707966,
      "grad_norm": 4.2266211153219047e-07,
      "learning_rate": 3.0119661965926123e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9870552,
      "step": 17375
    },
    {
      "epoch": 304.92035398230087,
      "grad_norm": 3.6925203517057525e-07,
      "learning_rate": 3.0110052159398587e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9873416,
      "step": 17380
    },
    {
      "epoch": 305.0,
      "grad_norm": 1.8521863864862098e-07,
      "learning_rate": 3.0100441564836802e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9875816,
      "step": 17385
    },
    {
      "epoch": 305.08849557522126,
      "grad_norm": 7.85786255619314e-07,
      "learning_rate": 3.0090830183722817e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9879064,
      "step": 17390
    },
    {
      "epoch": 305.17699115044246,
      "grad_norm": 1.8090935327563784e-06,
      "learning_rate": 3.0081218017538852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9881928,
      "step": 17395
    },
    {
      "epoch": 305.2654867256637,
      "grad_norm": 6.819186069151328e-07,
      "learning_rate": 3.0071605067767212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9884648,
      "step": 17400
    },
    {
      "epoch": 305.2654867256637,
      "eval_loss": 0.5646206140518188,
      "eval_runtime": 1.0621,
      "eval_samples_per_second": 23.539,
      "eval_steps_per_second": 12.24,
      "num_input_tokens_seen": 9884648,
      "step": 17400
    },
    {
      "epoch": 305.353982300885,
      "grad_norm": 8.685897228133399e-07,
      "learning_rate": 3.006199133589034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9887384,
      "step": 17405
    },
    {
      "epoch": 305.4424778761062,
      "grad_norm": 4.190805157122668e-06,
      "learning_rate": 3.005237682339079e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9890344,
      "step": 17410
    },
    {
      "epoch": 305.53097345132744,
      "grad_norm": 2.406622343187337e-06,
      "learning_rate": 3.0042761531751228e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9893144,
      "step": 17415
    },
    {
      "epoch": 305.6194690265487,
      "grad_norm": 8.748916116019245e-07,
      "learning_rate": 3.0033145462454482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9895848,
      "step": 17420
    },
    {
      "epoch": 305.7079646017699,
      "grad_norm": 6.613349796680268e-06,
      "learning_rate": 3.002352861698345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9899160,
      "step": 17425
    },
    {
      "epoch": 305.79646017699116,
      "grad_norm": 8.664532629154564e-07,
      "learning_rate": 3.0013910996821178e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9902024,
      "step": 17430
    },
    {
      "epoch": 305.88495575221236,
      "grad_norm": 2.472962250976707e-06,
      "learning_rate": 3.0004292603450817e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9904728,
      "step": 17435
    },
    {
      "epoch": 305.9734513274336,
      "grad_norm": 1.8647297110874206e-06,
      "learning_rate": 2.9994673438355653e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9907576,
      "step": 17440
    },
    {
      "epoch": 306.05309734513276,
      "grad_norm": 9.598373935659765e-07,
      "learning_rate": 2.9985053503019078e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9909608,
      "step": 17445
    },
    {
      "epoch": 306.14159292035396,
      "grad_norm": 1.1827660273411311e-06,
      "learning_rate": 2.99754327989246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9912584,
      "step": 17450
    },
    {
      "epoch": 306.2300884955752,
      "grad_norm": 8.56582857977628e-07,
      "learning_rate": 2.9965811327555864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9915208,
      "step": 17455
    },
    {
      "epoch": 306.3185840707965,
      "grad_norm": 1.7923638324646163e-06,
      "learning_rate": 2.995618909039662e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9918184,
      "step": 17460
    },
    {
      "epoch": 306.4070796460177,
      "grad_norm": 5.924018182668078e-07,
      "learning_rate": 2.9946566088930727e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9920776,
      "step": 17465
    },
    {
      "epoch": 306.49557522123894,
      "grad_norm": 4.5858914745622315e-06,
      "learning_rate": 2.9936942324642192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9923400,
      "step": 17470
    },
    {
      "epoch": 306.5840707964602,
      "grad_norm": 1.0468916116224136e-06,
      "learning_rate": 2.9927317799015097e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9927256,
      "step": 17475
    },
    {
      "epoch": 306.6725663716814,
      "grad_norm": 7.454368073922524e-07,
      "learning_rate": 2.9917692513533685e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9929704,
      "step": 17480
    },
    {
      "epoch": 306.76106194690266,
      "grad_norm": 5.463374236569507e-06,
      "learning_rate": 2.990806646968229e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9932872,
      "step": 17485
    },
    {
      "epoch": 306.8495575221239,
      "grad_norm": 6.686409506073687e-07,
      "learning_rate": 2.989843966894536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9935608,
      "step": 17490
    },
    {
      "epoch": 306.9380530973451,
      "grad_norm": 9.86780946732324e-07,
      "learning_rate": 2.9888812112807472e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9938632,
      "step": 17495
    },
    {
      "epoch": 307.01769911504425,
      "grad_norm": 3.07316486214404e-06,
      "learning_rate": 2.987918380275333e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9940824,
      "step": 17500
    },
    {
      "epoch": 307.1061946902655,
      "grad_norm": 3.722282997387083e-07,
      "learning_rate": 2.9869554740267724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9943704,
      "step": 17505
    },
    {
      "epoch": 307.1946902654867,
      "grad_norm": 1.804698740670574e-06,
      "learning_rate": 2.9859924926835585e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9946184,
      "step": 17510
    },
    {
      "epoch": 307.283185840708,
      "grad_norm": 8.061292078309634e-07,
      "learning_rate": 2.9850294363941944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9949000,
      "step": 17515
    },
    {
      "epoch": 307.37168141592923,
      "grad_norm": 7.446913059538929e-07,
      "learning_rate": 2.9840663053071967e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9952072,
      "step": 17520
    },
    {
      "epoch": 307.46017699115043,
      "grad_norm": 6.710049092362169e-06,
      "learning_rate": 2.983103099571091e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9954440,
      "step": 17525
    },
    {
      "epoch": 307.5486725663717,
      "grad_norm": 5.3042089120936e-07,
      "learning_rate": 2.9821398193344164e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9957560,
      "step": 17530
    },
    {
      "epoch": 307.6371681415929,
      "grad_norm": 5.497168444890121e-07,
      "learning_rate": 2.9811764647457226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9960216,
      "step": 17535
    },
    {
      "epoch": 307.72566371681415,
      "grad_norm": 1.13327473627578e-06,
      "learning_rate": 2.9802130359535714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9963288,
      "step": 17540
    },
    {
      "epoch": 307.8141592920354,
      "grad_norm": 7.79229310410301e-07,
      "learning_rate": 2.979249533106535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9965960,
      "step": 17545
    },
    {
      "epoch": 307.9026548672566,
      "grad_norm": 1.5794079217812396e-06,
      "learning_rate": 2.9782859563531986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9969000,
      "step": 17550
    },
    {
      "epoch": 307.9911504424779,
      "grad_norm": 9.79000105871819e-07,
      "learning_rate": 2.977322305842156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9971928,
      "step": 17555
    },
    {
      "epoch": 308.070796460177,
      "grad_norm": 4.991553055333497e-07,
      "learning_rate": 2.9763585817220162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9974680,
      "step": 17560
    },
    {
      "epoch": 308.1592920353982,
      "grad_norm": 6.322704280137259e-07,
      "learning_rate": 2.975394784141397e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9977352,
      "step": 17565
    },
    {
      "epoch": 308.24778761061947,
      "grad_norm": 7.226452680697548e-07,
      "learning_rate": 2.974430913248928e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9980584,
      "step": 17570
    },
    {
      "epoch": 308.3362831858407,
      "grad_norm": 3.7983038509992184e-06,
      "learning_rate": 2.9734669691932497e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9983064,
      "step": 17575
    },
    {
      "epoch": 308.42477876106193,
      "grad_norm": 9.402024829796574e-07,
      "learning_rate": 2.9725029521230147e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9985768,
      "step": 17580
    },
    {
      "epoch": 308.5132743362832,
      "grad_norm": 7.585142611787887e-07,
      "learning_rate": 2.9715388621868873e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9988856,
      "step": 17585
    },
    {
      "epoch": 308.60176991150445,
      "grad_norm": 1.2466480256989598e-06,
      "learning_rate": 2.970574699533541e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9991352,
      "step": 17590
    },
    {
      "epoch": 308.69026548672565,
      "grad_norm": 7.61877174682013e-07,
      "learning_rate": 2.969610464311662e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9994584,
      "step": 17595
    },
    {
      "epoch": 308.7787610619469,
      "grad_norm": 6.600232040909759e-07,
      "learning_rate": 2.9686461566699487e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9997288,
      "step": 17600
    },
    {
      "epoch": 308.7787610619469,
      "eval_loss": 0.5576398372650146,
      "eval_runtime": 1.0646,
      "eval_samples_per_second": 23.484,
      "eval_steps_per_second": 12.212,
      "num_input_tokens_seen": 9997288,
      "step": 17600
    },
    {
      "epoch": 308.86725663716817,
      "grad_norm": 7.028733080005622e-07,
      "learning_rate": 2.9676817767571086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 9999976,
      "step": 17605
    },
    {
      "epoch": 308.95575221238937,
      "grad_norm": 9.754702432474005e-07,
      "learning_rate": 2.966717324721861e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10003320,
      "step": 17610
    },
    {
      "epoch": 309.0353982300885,
      "grad_norm": 2.0938423404004425e-06,
      "learning_rate": 2.9657528007129366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10005664,
      "step": 17615
    },
    {
      "epoch": 309.12389380530976,
      "grad_norm": 5.636761670757551e-07,
      "learning_rate": 2.9647882048790777e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10008288,
      "step": 17620
    },
    {
      "epoch": 309.21238938053096,
      "grad_norm": 4.753216217068257e-06,
      "learning_rate": 2.963823537369037e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10011440,
      "step": 17625
    },
    {
      "epoch": 309.3008849557522,
      "grad_norm": 4.7339790398837067e-07,
      "learning_rate": 2.9628587983315775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10014720,
      "step": 17630
    },
    {
      "epoch": 309.3893805309734,
      "grad_norm": 5.169814016880991e-07,
      "learning_rate": 2.9618939879154746e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10017408,
      "step": 17635
    },
    {
      "epoch": 309.4778761061947,
      "grad_norm": 1.1155262882311945e-06,
      "learning_rate": 2.9609291062695143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10020096,
      "step": 17640
    },
    {
      "epoch": 309.56637168141594,
      "grad_norm": 1.4191002719599055e-06,
      "learning_rate": 2.9599641535424938e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10022944,
      "step": 17645
    },
    {
      "epoch": 309.65486725663715,
      "grad_norm": 1.7338364841634757e-06,
      "learning_rate": 2.9589991298832202e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10025344,
      "step": 17650
    },
    {
      "epoch": 309.7433628318584,
      "grad_norm": 3.218521669623442e-06,
      "learning_rate": 2.958034035440513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10028656,
      "step": 17655
    },
    {
      "epoch": 309.83185840707966,
      "grad_norm": 6.855462970634107e-07,
      "learning_rate": 2.957068870363201e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10031616,
      "step": 17660
    },
    {
      "epoch": 309.92035398230087,
      "grad_norm": 4.0459732986164454e-07,
      "learning_rate": 2.956103634800126e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10034416,
      "step": 17665
    },
    {
      "epoch": 310.0,
      "grad_norm": 4.738609504784108e-07,
      "learning_rate": 2.9551383289001384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10036912,
      "step": 17670
    },
    {
      "epoch": 310.08849557522126,
      "grad_norm": 1.4482685628536274e-06,
      "learning_rate": 2.9541729528121005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10039408,
      "step": 17675
    },
    {
      "epoch": 310.17699115044246,
      "grad_norm": 6.404679879778996e-06,
      "learning_rate": 2.9532075066848856e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10042304,
      "step": 17680
    },
    {
      "epoch": 310.2654867256637,
      "grad_norm": 1.5648565749870613e-06,
      "learning_rate": 2.9522419906673786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10045232,
      "step": 17685
    },
    {
      "epoch": 310.353982300885,
      "grad_norm": 1.0869155175896594e-06,
      "learning_rate": 2.951276404908474e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10048608,
      "step": 17690
    },
    {
      "epoch": 310.4424778761062,
      "grad_norm": 9.206083291246614e-07,
      "learning_rate": 2.9503107495570752e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10051568,
      "step": 17695
    },
    {
      "epoch": 310.53097345132744,
      "grad_norm": 1.3158912679500645e-06,
      "learning_rate": 2.9493450247621003e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10054544,
      "step": 17700
    },
    {
      "epoch": 310.6194690265487,
      "grad_norm": 8.547874017494905e-07,
      "learning_rate": 2.948379230672476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10057088,
      "step": 17705
    },
    {
      "epoch": 310.7079646017699,
      "grad_norm": 3.56844657289912e-06,
      "learning_rate": 2.9474133674371396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10059776,
      "step": 17710
    },
    {
      "epoch": 310.79646017699116,
      "grad_norm": 2.091376245516585e-06,
      "learning_rate": 2.9464474352050387e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10062576,
      "step": 17715
    },
    {
      "epoch": 310.88495575221236,
      "grad_norm": 1.2913827731608762e-06,
      "learning_rate": 2.9454814341251336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10065632,
      "step": 17720
    },
    {
      "epoch": 310.9734513274336,
      "grad_norm": 2.4956973447842756e-06,
      "learning_rate": 2.9445153643463942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10068256,
      "step": 17725
    },
    {
      "epoch": 311.05309734513276,
      "grad_norm": 5.377239062909211e-07,
      "learning_rate": 2.943549226017798e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10070480,
      "step": 17730
    },
    {
      "epoch": 311.14159292035396,
      "grad_norm": 6.507028729174635e-07,
      "learning_rate": 2.942583019288337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10073136,
      "step": 17735
    },
    {
      "epoch": 311.2300884955752,
      "grad_norm": 9.172836143989116e-07,
      "learning_rate": 2.9416167443070132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10076160,
      "step": 17740
    },
    {
      "epoch": 311.3185840707965,
      "grad_norm": 7.732157882855972e-07,
      "learning_rate": 2.9406504012228375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10079312,
      "step": 17745
    },
    {
      "epoch": 311.4070796460177,
      "grad_norm": 2.9091031592543004e-06,
      "learning_rate": 2.939683990184832e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10082208,
      "step": 17750
    },
    {
      "epoch": 311.49557522123894,
      "grad_norm": 1.2488807215049746e-06,
      "learning_rate": 2.93871751134203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10085312,
      "step": 17755
    },
    {
      "epoch": 311.5840707964602,
      "grad_norm": 1.5784793276907294e-06,
      "learning_rate": 2.9377509648434752e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10088352,
      "step": 17760
    },
    {
      "epoch": 311.6725663716814,
      "grad_norm": 2.50619018515863e-06,
      "learning_rate": 2.9367843508382203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10091216,
      "step": 17765
    },
    {
      "epoch": 311.76106194690266,
      "grad_norm": 5.020452817916521e-07,
      "learning_rate": 2.9358176694753293e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10093760,
      "step": 17770
    },
    {
      "epoch": 311.8495575221239,
      "grad_norm": 7.080602131281921e-07,
      "learning_rate": 2.9348509209038766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10097136,
      "step": 17775
    },
    {
      "epoch": 311.9380530973451,
      "grad_norm": 8.435101335635409e-06,
      "learning_rate": 2.933884105272947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10099984,
      "step": 17780
    },
    {
      "epoch": 312.01769911504425,
      "grad_norm": 9.042262263392331e-07,
      "learning_rate": 2.9329172227316366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10102272,
      "step": 17785
    },
    {
      "epoch": 312.1061946902655,
      "grad_norm": 1.7718314211379038e-06,
      "learning_rate": 2.93195027342905e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10104768,
      "step": 17790
    },
    {
      "epoch": 312.1946902654867,
      "grad_norm": 4.270454212473851e-07,
      "learning_rate": 2.9309832575143024e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10108304,
      "step": 17795
    },
    {
      "epoch": 312.283185840708,
      "grad_norm": 8.860191087478597e-07,
      "learning_rate": 2.930016175136521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10111472,
      "step": 17800
    },
    {
      "epoch": 312.283185840708,
      "eval_loss": 0.5531618595123291,
      "eval_runtime": 1.0758,
      "eval_samples_per_second": 23.239,
      "eval_steps_per_second": 12.084,
      "num_input_tokens_seen": 10111472,
      "step": 17800
    },
    {
      "epoch": 312.37168141592923,
      "grad_norm": 3.177602593495976e-06,
      "learning_rate": 2.9290490264448412e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10113968,
      "step": 17805
    },
    {
      "epoch": 312.46017699115043,
      "grad_norm": 7.533496386713523e-07,
      "learning_rate": 2.9280818115884094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10116528,
      "step": 17810
    },
    {
      "epoch": 312.5486725663717,
      "grad_norm": 2.661473445186857e-06,
      "learning_rate": 2.9271145307163828e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10119520,
      "step": 17815
    },
    {
      "epoch": 312.6371681415929,
      "grad_norm": 1.3452151961246273e-06,
      "learning_rate": 2.9261471839779287e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10122592,
      "step": 17820
    },
    {
      "epoch": 312.72566371681415,
      "grad_norm": 4.619740820999141e-07,
      "learning_rate": 2.925179771522223e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10125376,
      "step": 17825
    },
    {
      "epoch": 312.8141592920354,
      "grad_norm": 6.737585067639884e-07,
      "learning_rate": 2.9242122934984535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10127920,
      "step": 17830
    },
    {
      "epoch": 312.9026548672566,
      "grad_norm": 1.249170509254327e-06,
      "learning_rate": 2.9232447500558176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10131136,
      "step": 17835
    },
    {
      "epoch": 312.9911504424779,
      "grad_norm": 6.940898060747713e-07,
      "learning_rate": 2.9222771413435225e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10133664,
      "step": 17840
    },
    {
      "epoch": 313.070796460177,
      "grad_norm": 2.2592307686863933e-06,
      "learning_rate": 2.9213094675107848e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10136024,
      "step": 17845
    },
    {
      "epoch": 313.1592920353982,
      "grad_norm": 6.629805398006283e-07,
      "learning_rate": 2.9203417287068335e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10139064,
      "step": 17850
    },
    {
      "epoch": 313.24778761061947,
      "grad_norm": 9.643739531384199e-07,
      "learning_rate": 2.9193739250809042e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10141944,
      "step": 17855
    },
    {
      "epoch": 313.3362831858407,
      "grad_norm": 1.2852412965003168e-06,
      "learning_rate": 2.9184060567822463e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10144392,
      "step": 17860
    },
    {
      "epoch": 313.42477876106193,
      "grad_norm": 2.0493368992902106e-06,
      "learning_rate": 2.9174381239601166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10146840,
      "step": 17865
    },
    {
      "epoch": 313.5132743362832,
      "grad_norm": 5.5501836868643295e-06,
      "learning_rate": 2.916470126763783e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10150104,
      "step": 17870
    },
    {
      "epoch": 313.60176991150445,
      "grad_norm": 1.2123887245252263e-06,
      "learning_rate": 2.9155020653425203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10152888,
      "step": 17875
    },
    {
      "epoch": 313.69026548672565,
      "grad_norm": 5.155259714229032e-06,
      "learning_rate": 2.9145339398456184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10155192,
      "step": 17880
    },
    {
      "epoch": 313.7787610619469,
      "grad_norm": 1.1533857104950584e-06,
      "learning_rate": 2.913565750422374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10157976,
      "step": 17885
    },
    {
      "epoch": 313.86725663716817,
      "grad_norm": 2.449812200211454e-06,
      "learning_rate": 2.9125974972220938e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10161144,
      "step": 17890
    },
    {
      "epoch": 313.95575221238937,
      "grad_norm": 4.829850013265968e-07,
      "learning_rate": 2.9116291803940932e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10164328,
      "step": 17895
    },
    {
      "epoch": 314.0353982300885,
      "grad_norm": 1.7505645928395097e-06,
      "learning_rate": 2.910660800087701e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10167128,
      "step": 17900
    },
    {
      "epoch": 314.12389380530976,
      "grad_norm": 2.602281483632396e-06,
      "learning_rate": 2.909692356452254e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10169848,
      "step": 17905
    },
    {
      "epoch": 314.21238938053096,
      "grad_norm": 7.613076036250277e-07,
      "learning_rate": 2.9087238496370962e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10172920,
      "step": 17910
    },
    {
      "epoch": 314.3008849557522,
      "grad_norm": 1.1976528639934259e-06,
      "learning_rate": 2.907755279791583e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10175576,
      "step": 17915
    },
    {
      "epoch": 314.3893805309734,
      "grad_norm": 1.2737613133140258e-06,
      "learning_rate": 2.906786647065083e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10178056,
      "step": 17920
    },
    {
      "epoch": 314.4778761061947,
      "grad_norm": 4.099559646419948e-06,
      "learning_rate": 2.9058179516069695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10181608,
      "step": 17925
    },
    {
      "epoch": 314.56637168141594,
      "grad_norm": 4.099232683074661e-06,
      "learning_rate": 2.9048491935666282e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10184792,
      "step": 17930
    },
    {
      "epoch": 314.65486725663715,
      "grad_norm": 5.755825895903399e-07,
      "learning_rate": 2.9038803730934534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10187576,
      "step": 17935
    },
    {
      "epoch": 314.7433628318584,
      "grad_norm": 1.1861843631777447e-06,
      "learning_rate": 2.9029114903368503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10189928,
      "step": 17940
    },
    {
      "epoch": 314.83185840707966,
      "grad_norm": 4.357857505965512e-07,
      "learning_rate": 2.9019425454462318e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10192616,
      "step": 17945
    },
    {
      "epoch": 314.92035398230087,
      "grad_norm": 5.306022785589448e-07,
      "learning_rate": 2.9009735385710212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10195192,
      "step": 17950
    },
    {
      "epoch": 315.0,
      "grad_norm": 9.639496965974104e-06,
      "learning_rate": 2.900004469860652e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10197744,
      "step": 17955
    },
    {
      "epoch": 315.08849557522126,
      "grad_norm": 4.879865400653216e-07,
      "learning_rate": 2.8990353394645668e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10200736,
      "step": 17960
    },
    {
      "epoch": 315.17699115044246,
      "grad_norm": 7.033920041976671e-07,
      "learning_rate": 2.8980661475322186e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10203952,
      "step": 17965
    },
    {
      "epoch": 315.2654867256637,
      "grad_norm": 2.621571411509649e-06,
      "learning_rate": 2.897096894213067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10206896,
      "step": 17970
    },
    {
      "epoch": 315.353982300885,
      "grad_norm": 3.89375418308191e-06,
      "learning_rate": 2.8961275796565845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10210000,
      "step": 17975
    },
    {
      "epoch": 315.4424778761062,
      "grad_norm": 1.2422812005752348e-06,
      "learning_rate": 2.8951582040122517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10212704,
      "step": 17980
    },
    {
      "epoch": 315.53097345132744,
      "grad_norm": 6.165014951875492e-07,
      "learning_rate": 2.894188767429557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10215136,
      "step": 17985
    },
    {
      "epoch": 315.6194690265487,
      "grad_norm": 1.5391234455819358e-06,
      "learning_rate": 2.8932192700580014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10218080,
      "step": 17990
    },
    {
      "epoch": 315.7079646017699,
      "grad_norm": 8.054987119976431e-06,
      "learning_rate": 2.8922497120470916e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10220752,
      "step": 17995
    },
    {
      "epoch": 315.79646017699116,
      "grad_norm": 3.7348193018260645e-06,
      "learning_rate": 2.891280093546348e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10223648,
      "step": 18000
    },
    {
      "epoch": 315.79646017699116,
      "eval_loss": 0.5568270087242126,
      "eval_runtime": 1.0711,
      "eval_samples_per_second": 23.34,
      "eval_steps_per_second": 12.137,
      "num_input_tokens_seen": 10223648,
      "step": 18000
    },
    {
      "epoch": 315.88495575221236,
      "grad_norm": 4.154700945946388e-06,
      "learning_rate": 2.890310414705297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10226400,
      "step": 18005
    },
    {
      "epoch": 315.9734513274336,
      "grad_norm": 4.478337984892278e-07,
      "learning_rate": 2.8893406756734742e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10229536,
      "step": 18010
    },
    {
      "epoch": 316.05309734513276,
      "grad_norm": 1.3149671076462255e-06,
      "learning_rate": 2.888370876600427e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10231656,
      "step": 18015
    },
    {
      "epoch": 316.14159292035396,
      "grad_norm": 6.766176738892682e-07,
      "learning_rate": 2.8874010176357104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10234248,
      "step": 18020
    },
    {
      "epoch": 316.2300884955752,
      "grad_norm": 1.386128587910207e-06,
      "learning_rate": 2.886431098928888e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10236808,
      "step": 18025
    },
    {
      "epoch": 316.3185840707965,
      "grad_norm": 8.695931228430709e-07,
      "learning_rate": 2.885461120629534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10239496,
      "step": 18030
    },
    {
      "epoch": 316.4070796460177,
      "grad_norm": 2.4240216589532793e-06,
      "learning_rate": 2.8844910828872317e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10242392,
      "step": 18035
    },
    {
      "epoch": 316.49557522123894,
      "grad_norm": 7.411364322251757e-07,
      "learning_rate": 2.8835209858515715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10245304,
      "step": 18040
    },
    {
      "epoch": 316.5840707964602,
      "grad_norm": 6.84788233229483e-07,
      "learning_rate": 2.8825508296721566e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10248088,
      "step": 18045
    },
    {
      "epoch": 316.6725663716814,
      "grad_norm": 2.9307796012290055e-06,
      "learning_rate": 2.881580614498596e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10251464,
      "step": 18050
    },
    {
      "epoch": 316.76106194690266,
      "grad_norm": 2.441283641019254e-06,
      "learning_rate": 2.8806103404805103e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10254424,
      "step": 18055
    },
    {
      "epoch": 316.8495575221239,
      "grad_norm": 8.102175002022705e-07,
      "learning_rate": 2.8796400077675257e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10257896,
      "step": 18060
    },
    {
      "epoch": 316.9380530973451,
      "grad_norm": 1.3682392818736844e-06,
      "learning_rate": 2.8786696165092812e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10260424,
      "step": 18065
    },
    {
      "epoch": 317.01769911504425,
      "grad_norm": 1.1790505141107133e-06,
      "learning_rate": 2.8776991668554236e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10263120,
      "step": 18070
    },
    {
      "epoch": 317.1061946902655,
      "grad_norm": 1.1321006923026289e-06,
      "learning_rate": 2.876728658955608e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10265568,
      "step": 18075
    },
    {
      "epoch": 317.1946902654867,
      "grad_norm": 1.3326971384231001e-06,
      "learning_rate": 2.8757580929594986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10268208,
      "step": 18080
    },
    {
      "epoch": 317.283185840708,
      "grad_norm": 2.1500229649973335e-06,
      "learning_rate": 2.87478746901677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10271376,
      "step": 18085
    },
    {
      "epoch": 317.37168141592923,
      "grad_norm": 2.5495228328509256e-06,
      "learning_rate": 2.873816787277103e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10274176,
      "step": 18090
    },
    {
      "epoch": 317.46017699115043,
      "grad_norm": 2.359766995141399e-06,
      "learning_rate": 2.8728460478901903e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10277264,
      "step": 18095
    },
    {
      "epoch": 317.5486725663717,
      "grad_norm": 1.6830838376336033e-06,
      "learning_rate": 2.8718752510057307e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10279744,
      "step": 18100
    },
    {
      "epoch": 317.6371681415929,
      "grad_norm": 9.57243969423871e-07,
      "learning_rate": 2.870904396773435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10283040,
      "step": 18105
    },
    {
      "epoch": 317.72566371681415,
      "grad_norm": 1.1569569551284076e-06,
      "learning_rate": 2.86993348534302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10285536,
      "step": 18110
    },
    {
      "epoch": 317.8141592920354,
      "grad_norm": 1.179107812276925e-06,
      "learning_rate": 2.868962516864212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10288672,
      "step": 18115
    },
    {
      "epoch": 317.9026548672566,
      "grad_norm": 1.3873768693883903e-06,
      "learning_rate": 2.8679914914867477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10291456,
      "step": 18120
    },
    {
      "epoch": 317.9911504424779,
      "grad_norm": 2.053332309515099e-06,
      "learning_rate": 2.8670204093603713e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10294480,
      "step": 18125
    },
    {
      "epoch": 318.070796460177,
      "grad_norm": 1.6165326996997464e-06,
      "learning_rate": 2.8660492706348357e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10296688,
      "step": 18130
    },
    {
      "epoch": 318.1592920353982,
      "grad_norm": 6.932355631761311e-07,
      "learning_rate": 2.8650780754599022e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10299552,
      "step": 18135
    },
    {
      "epoch": 318.24778761061947,
      "grad_norm": 1.74822935150587e-06,
      "learning_rate": 2.8641068239853407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10302240,
      "step": 18140
    },
    {
      "epoch": 318.3362831858407,
      "grad_norm": 7.77652132910589e-07,
      "learning_rate": 2.863135516360932e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10305200,
      "step": 18145
    },
    {
      "epoch": 318.42477876106193,
      "grad_norm": 7.518420375163259e-07,
      "learning_rate": 2.8621641527364633e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10308688,
      "step": 18150
    },
    {
      "epoch": 318.5132743362832,
      "grad_norm": 9.158231932815397e-07,
      "learning_rate": 2.8611927332617313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10311328,
      "step": 18155
    },
    {
      "epoch": 318.60176991150445,
      "grad_norm": 9.288137903240568e-07,
      "learning_rate": 2.8602212580865405e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10313952,
      "step": 18160
    },
    {
      "epoch": 318.69026548672565,
      "grad_norm": 2.0264340037101647e-06,
      "learning_rate": 2.859249727360705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10316656,
      "step": 18165
    },
    {
      "epoch": 318.7787610619469,
      "grad_norm": 1.9240410438214894e-06,
      "learning_rate": 2.8582781412340465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10319808,
      "step": 18170
    },
    {
      "epoch": 318.86725663716817,
      "grad_norm": 2.2954141059017275e-06,
      "learning_rate": 2.857306499856397e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10322336,
      "step": 18175
    },
    {
      "epoch": 318.95575221238937,
      "grad_norm": 1.6639240811855416e-06,
      "learning_rate": 2.856334803377594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10325248,
      "step": 18180
    },
    {
      "epoch": 319.0353982300885,
      "grad_norm": 8.787128535914235e-07,
      "learning_rate": 2.8553630519474867e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10327968,
      "step": 18185
    },
    {
      "epoch": 319.12389380530976,
      "grad_norm": 8.344580351149489e-07,
      "learning_rate": 2.8543912457159317e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10330800,
      "step": 18190
    },
    {
      "epoch": 319.21238938053096,
      "grad_norm": 1.5165948070716695e-06,
      "learning_rate": 2.853419384832792e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10333424,
      "step": 18195
    },
    {
      "epoch": 319.3008849557522,
      "grad_norm": 1.2671520153162419e-06,
      "learning_rate": 2.8524474694479423e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10336864,
      "step": 18200
    },
    {
      "epoch": 319.3008849557522,
      "eval_loss": 0.5882929563522339,
      "eval_runtime": 1.0624,
      "eval_samples_per_second": 23.532,
      "eval_steps_per_second": 12.237,
      "num_input_tokens_seen": 10336864,
      "step": 18200
    },
    {
      "epoch": 319.3893805309734,
      "grad_norm": 1.0696053323044907e-06,
      "learning_rate": 2.851475499711264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10339680,
      "step": 18205
    },
    {
      "epoch": 319.4778761061947,
      "grad_norm": 1.3497623285729787e-06,
      "learning_rate": 2.8505034757726468e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10342224,
      "step": 18210
    },
    {
      "epoch": 319.56637168141594,
      "grad_norm": 9.976616865969845e-07,
      "learning_rate": 2.8495313977819886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10344816,
      "step": 18215
    },
    {
      "epoch": 319.65486725663715,
      "grad_norm": 1.9696963136084378e-06,
      "learning_rate": 2.8485592658891956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10347456,
      "step": 18220
    },
    {
      "epoch": 319.7433628318584,
      "grad_norm": 2.5112738057941897e-06,
      "learning_rate": 2.8475870802441844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10350592,
      "step": 18225
    },
    {
      "epoch": 319.83185840707966,
      "grad_norm": 2.611019226606004e-06,
      "learning_rate": 2.8466148409968774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10353440,
      "step": 18230
    },
    {
      "epoch": 319.92035398230087,
      "grad_norm": 1.0201753184446716e-06,
      "learning_rate": 2.8456425482972067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10356976,
      "step": 18235
    },
    {
      "epoch": 320.0,
      "grad_norm": 6.646955625910778e-07,
      "learning_rate": 2.84467020229511e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10359368,
      "step": 18240
    },
    {
      "epoch": 320.08849557522126,
      "grad_norm": 6.781282991141779e-06,
      "learning_rate": 2.8436978031405375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10362504,
      "step": 18245
    },
    {
      "epoch": 320.17699115044246,
      "grad_norm": 8.502667014909093e-07,
      "learning_rate": 2.842725350983445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10364920,
      "step": 18250
    },
    {
      "epoch": 320.2654867256637,
      "grad_norm": 1.301095835515298e-06,
      "learning_rate": 2.8417528459737957e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10368104,
      "step": 18255
    },
    {
      "epoch": 320.353982300885,
      "grad_norm": 2.4505529836460482e-06,
      "learning_rate": 2.8407802882615624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10370952,
      "step": 18260
    },
    {
      "epoch": 320.4424778761062,
      "grad_norm": 1.255759343621321e-06,
      "learning_rate": 2.8398076779967277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10373432,
      "step": 18265
    },
    {
      "epoch": 320.53097345132744,
      "grad_norm": 9.126575264417625e-07,
      "learning_rate": 2.8388350153292774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10376280,
      "step": 18270
    },
    {
      "epoch": 320.6194690265487,
      "grad_norm": 1.6758589254095568e-06,
      "learning_rate": 2.8378623004092103e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10378824,
      "step": 18275
    },
    {
      "epoch": 320.7079646017699,
      "grad_norm": 8.328273679580889e-07,
      "learning_rate": 2.8368895333865302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10381816,
      "step": 18280
    },
    {
      "epoch": 320.79646017699116,
      "grad_norm": 1.3661614275406464e-06,
      "learning_rate": 2.835916714411251e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10384568,
      "step": 18285
    },
    {
      "epoch": 320.88495575221236,
      "grad_norm": 1.1215728363822564e-06,
      "learning_rate": 2.8349438436333926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10387528,
      "step": 18290
    },
    {
      "epoch": 320.9734513274336,
      "grad_norm": 1.5123996490729041e-06,
      "learning_rate": 2.833970921202984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10390376,
      "step": 18295
    },
    {
      "epoch": 321.05309734513276,
      "grad_norm": 1.585685367899714e-06,
      "learning_rate": 2.8329979472700628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10393016,
      "step": 18300
    },
    {
      "epoch": 321.14159292035396,
      "grad_norm": 8.91267120550765e-07,
      "learning_rate": 2.832024921984674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10396008,
      "step": 18305
    },
    {
      "epoch": 321.2300884955752,
      "grad_norm": 2.2632871150563005e-06,
      "learning_rate": 2.8310518454968693e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10399080,
      "step": 18310
    },
    {
      "epoch": 321.3185840707965,
      "grad_norm": 1.2368428770059836e-06,
      "learning_rate": 2.8300787179567095e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10402072,
      "step": 18315
    },
    {
      "epoch": 321.4070796460177,
      "grad_norm": 1.1955592071899446e-06,
      "learning_rate": 2.8291055395142636e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10405032,
      "step": 18320
    },
    {
      "epoch": 321.49557522123894,
      "grad_norm": 8.519264724782261e-07,
      "learning_rate": 2.8281323103196073e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10407528,
      "step": 18325
    },
    {
      "epoch": 321.5840707964602,
      "grad_norm": 8.566709652768623e-07,
      "learning_rate": 2.8271590305228256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10410184,
      "step": 18330
    },
    {
      "epoch": 321.6725663716814,
      "grad_norm": 1.3325699228516896e-06,
      "learning_rate": 2.82618570027401e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10413288,
      "step": 18335
    },
    {
      "epoch": 321.76106194690266,
      "grad_norm": 1.2315191497691558e-06,
      "learning_rate": 2.8252123197232604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10416376,
      "step": 18340
    },
    {
      "epoch": 321.8495575221239,
      "grad_norm": 1.1450886177044595e-06,
      "learning_rate": 2.8242388890206843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10419656,
      "step": 18345
    },
    {
      "epoch": 321.9380530973451,
      "grad_norm": 5.19954483024776e-06,
      "learning_rate": 2.8232654083163967e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10422376,
      "step": 18350
    },
    {
      "epoch": 322.01769911504425,
      "grad_norm": 2.1499818103620782e-06,
      "learning_rate": 2.822291877760521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10424320,
      "step": 18355
    },
    {
      "epoch": 322.1061946902655,
      "grad_norm": 9.519846457806125e-07,
      "learning_rate": 2.8213182975031864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10427376,
      "step": 18360
    },
    {
      "epoch": 322.1946902654867,
      "grad_norm": 6.563521424141072e-07,
      "learning_rate": 2.8203446676945337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10430240,
      "step": 18365
    },
    {
      "epoch": 322.283185840708,
      "grad_norm": 1.4628452618126175e-06,
      "learning_rate": 2.8193709884847075e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10433248,
      "step": 18370
    },
    {
      "epoch": 322.37168141592923,
      "grad_norm": 7.998144155862974e-07,
      "learning_rate": 2.8183972600238605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10436400,
      "step": 18375
    },
    {
      "epoch": 322.46017699115043,
      "grad_norm": 8.152090913426946e-07,
      "learning_rate": 2.817423482462156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10439600,
      "step": 18380
    },
    {
      "epoch": 322.5486725663717,
      "grad_norm": 1.6207687849600916e-06,
      "learning_rate": 2.8164496559497605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10442448,
      "step": 18385
    },
    {
      "epoch": 322.6371681415929,
      "grad_norm": 1.1398363994885585e-06,
      "learning_rate": 2.815475780636852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10444864,
      "step": 18390
    },
    {
      "epoch": 322.72566371681415,
      "grad_norm": 2.1260134417389054e-06,
      "learning_rate": 2.814501856673613e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10447424,
      "step": 18395
    },
    {
      "epoch": 322.8141592920354,
      "grad_norm": 5.257480097498046e-06,
      "learning_rate": 2.8135278842102353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10450688,
      "step": 18400
    },
    {
      "epoch": 322.8141592920354,
      "eval_loss": 0.5702996850013733,
      "eval_runtime": 1.0661,
      "eval_samples_per_second": 23.449,
      "eval_steps_per_second": 12.194,
      "num_input_tokens_seen": 10450688,
      "step": 18400
    },
    {
      "epoch": 322.9026548672566,
      "grad_norm": 7.948801794555038e-07,
      "learning_rate": 2.8125538633969183e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10453200,
      "step": 18405
    },
    {
      "epoch": 322.9911504424779,
      "grad_norm": 3.7601392932629096e-07,
      "learning_rate": 2.8115797943838677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10456016,
      "step": 18410
    },
    {
      "epoch": 323.070796460177,
      "grad_norm": 7.329235813813284e-07,
      "learning_rate": 2.810605677321298e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10458352,
      "step": 18415
    },
    {
      "epoch": 323.1592920353982,
      "grad_norm": 6.00692715124751e-07,
      "learning_rate": 2.809631512359428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10461456,
      "step": 18420
    },
    {
      "epoch": 323.24778761061947,
      "grad_norm": 1.1338267995597562e-06,
      "learning_rate": 2.8086572996484884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10464256,
      "step": 18425
    },
    {
      "epoch": 323.3362831858407,
      "grad_norm": 6.576790951839939e-07,
      "learning_rate": 2.8076830393387143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10467488,
      "step": 18430
    },
    {
      "epoch": 323.42477876106193,
      "grad_norm": 3.4354864055785583e-06,
      "learning_rate": 2.8067087315803497e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10470112,
      "step": 18435
    },
    {
      "epoch": 323.5132743362832,
      "grad_norm": 4.322396307543386e-07,
      "learning_rate": 2.8057343765236433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10473312,
      "step": 18440
    },
    {
      "epoch": 323.60176991150445,
      "grad_norm": 1.2872511661043973e-06,
      "learning_rate": 2.804759974318854e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10476192,
      "step": 18445
    },
    {
      "epoch": 323.69026548672565,
      "grad_norm": 1.0454757557454286e-06,
      "learning_rate": 2.8037855251162482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10478960,
      "step": 18450
    },
    {
      "epoch": 323.7787610619469,
      "grad_norm": 6.900335165482829e-07,
      "learning_rate": 2.802811029066096e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10481904,
      "step": 18455
    },
    {
      "epoch": 323.86725663716817,
      "grad_norm": 8.435360996372765e-07,
      "learning_rate": 2.8018364863186764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10484672,
      "step": 18460
    },
    {
      "epoch": 323.95575221238937,
      "grad_norm": 4.237367079440446e-07,
      "learning_rate": 2.800861897024279e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10487344,
      "step": 18465
    },
    {
      "epoch": 324.0353982300885,
      "grad_norm": 2.2091123810241697e-06,
      "learning_rate": 2.799887261333196e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10489664,
      "step": 18470
    },
    {
      "epoch": 324.12389380530976,
      "grad_norm": 9.362125865663984e-07,
      "learning_rate": 2.798912579395728e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10492464,
      "step": 18475
    },
    {
      "epoch": 324.21238938053096,
      "grad_norm": 8.846269565765397e-07,
      "learning_rate": 2.797937851362185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10495344,
      "step": 18480
    },
    {
      "epoch": 324.3008849557522,
      "grad_norm": 9.911196912071318e-07,
      "learning_rate": 2.7969630773828802e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10497984,
      "step": 18485
    },
    {
      "epoch": 324.3893805309734,
      "grad_norm": 9.890155752145802e-07,
      "learning_rate": 2.7959882576081382e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10501120,
      "step": 18490
    },
    {
      "epoch": 324.4778761061947,
      "grad_norm": 3.1501583634963026e-06,
      "learning_rate": 2.795013392188286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10503968,
      "step": 18495
    },
    {
      "epoch": 324.56637168141594,
      "grad_norm": 1.9013669998457772e-06,
      "learning_rate": 2.7940384812736614e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10506880,
      "step": 18500
    },
    {
      "epoch": 324.65486725663715,
      "grad_norm": 4.69071380848618e-07,
      "learning_rate": 2.7930635250146087e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10509520,
      "step": 18505
    },
    {
      "epoch": 324.7433628318584,
      "grad_norm": 2.5432684651605086e-06,
      "learning_rate": 2.792088523561477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10512192,
      "step": 18510
    },
    {
      "epoch": 324.83185840707966,
      "grad_norm": 4.565526126043551e-07,
      "learning_rate": 2.7911134770646246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10514640,
      "step": 18515
    },
    {
      "epoch": 324.92035398230087,
      "grad_norm": 5.470606652124843e-07,
      "learning_rate": 2.7901383856744157e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10517840,
      "step": 18520
    },
    {
      "epoch": 325.0,
      "grad_norm": 2.245176801807247e-06,
      "learning_rate": 2.7891632495412217e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10520464,
      "step": 18525
    },
    {
      "epoch": 325.08849557522126,
      "grad_norm": 2.8438130357244518e-06,
      "learning_rate": 2.7881880688154205e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10523584,
      "step": 18530
    },
    {
      "epoch": 325.17699115044246,
      "grad_norm": 1.024511789182725e-06,
      "learning_rate": 2.7872128436473977e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10526144,
      "step": 18535
    },
    {
      "epoch": 325.2654867256637,
      "grad_norm": 7.249998816405423e-06,
      "learning_rate": 2.7862375741875448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10528656,
      "step": 18540
    },
    {
      "epoch": 325.353982300885,
      "grad_norm": 3.3038804758689366e-07,
      "learning_rate": 2.785262260586261e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10531824,
      "step": 18545
    },
    {
      "epoch": 325.4424778761062,
      "grad_norm": 4.566566644825798e-07,
      "learning_rate": 2.7842869029939517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10534576,
      "step": 18550
    },
    {
      "epoch": 325.53097345132744,
      "grad_norm": 2.1584237401839346e-06,
      "learning_rate": 2.7833115015610296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10537520,
      "step": 18555
    },
    {
      "epoch": 325.6194690265487,
      "grad_norm": 7.983814498402353e-07,
      "learning_rate": 2.7823360564379136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10540640,
      "step": 18560
    },
    {
      "epoch": 325.7079646017699,
      "grad_norm": 5.64211575238005e-07,
      "learning_rate": 2.7813605677750297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10543808,
      "step": 18565
    },
    {
      "epoch": 325.79646017699116,
      "grad_norm": 8.94714673904673e-07,
      "learning_rate": 2.7803850357228102e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10546896,
      "step": 18570
    },
    {
      "epoch": 325.88495575221236,
      "grad_norm": 1.961642055903212e-06,
      "learning_rate": 2.779409460431695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10549776,
      "step": 18575
    },
    {
      "epoch": 325.9734513274336,
      "grad_norm": 7.040802643132338e-07,
      "learning_rate": 2.778433842052129e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10552192,
      "step": 18580
    },
    {
      "epoch": 326.05309734513276,
      "grad_norm": 2.457405344102881e-06,
      "learning_rate": 2.7774581807345664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10554360,
      "step": 18585
    },
    {
      "epoch": 326.14159292035396,
      "grad_norm": 8.892951655070647e-07,
      "learning_rate": 2.776482476629465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10557224,
      "step": 18590
    },
    {
      "epoch": 326.2300884955752,
      "grad_norm": 1.554854179630638e-06,
      "learning_rate": 2.7755067298872924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10559768,
      "step": 18595
    },
    {
      "epoch": 326.3185840707965,
      "grad_norm": 2.362822897339356e-06,
      "learning_rate": 2.774530940658518e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10563128,
      "step": 18600
    },
    {
      "epoch": 326.3185840707965,
      "eval_loss": 0.566409170627594,
      "eval_runtime": 1.0621,
      "eval_samples_per_second": 23.538,
      "eval_steps_per_second": 12.24,
      "num_input_tokens_seen": 10563128,
      "step": 18600
    },
    {
      "epoch": 326.4070796460177,
      "grad_norm": 2.1098646811879007e-06,
      "learning_rate": 2.7735551090936236e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10566088,
      "step": 18605
    },
    {
      "epoch": 326.49557522123894,
      "grad_norm": 1.8125701899407431e-06,
      "learning_rate": 2.7725792353430934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10569016,
      "step": 18610
    },
    {
      "epoch": 326.5840707964602,
      "grad_norm": 2.2031179469195195e-05,
      "learning_rate": 2.77160331955742e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10572008,
      "step": 18615
    },
    {
      "epoch": 326.6725663716814,
      "grad_norm": 1.983895572266192e-06,
      "learning_rate": 2.7706273618871008e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10574696,
      "step": 18620
    },
    {
      "epoch": 326.76106194690266,
      "grad_norm": 1.3495615576175624e-06,
      "learning_rate": 2.769651362482642e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10577720,
      "step": 18625
    },
    {
      "epoch": 326.8495575221239,
      "grad_norm": 4.510615781327942e-06,
      "learning_rate": 2.768675321494555e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10580696,
      "step": 18630
    },
    {
      "epoch": 326.9380530973451,
      "grad_norm": 8.885955935511447e-07,
      "learning_rate": 2.7676992390733565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10583208,
      "step": 18635
    },
    {
      "epoch": 327.01769911504425,
      "grad_norm": 5.193242600398662e-07,
      "learning_rate": 2.766723115369571e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10586016,
      "step": 18640
    },
    {
      "epoch": 327.1061946902655,
      "grad_norm": 1.8983391782967374e-06,
      "learning_rate": 2.765746950533729e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10588768,
      "step": 18645
    },
    {
      "epoch": 327.1946902654867,
      "grad_norm": 1.980057277251035e-06,
      "learning_rate": 2.7647707447163684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10592224,
      "step": 18650
    },
    {
      "epoch": 327.283185840708,
      "grad_norm": 7.126328114281932e-07,
      "learning_rate": 2.7637944980680315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10595264,
      "step": 18655
    },
    {
      "epoch": 327.37168141592923,
      "grad_norm": 2.2656909095530864e-06,
      "learning_rate": 2.762818210739268e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10597792,
      "step": 18660
    },
    {
      "epoch": 327.46017699115043,
      "grad_norm": 9.73410578808398e-07,
      "learning_rate": 2.7618418828806332e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10600688,
      "step": 18665
    },
    {
      "epoch": 327.5486725663717,
      "grad_norm": 2.245802079414716e-06,
      "learning_rate": 2.76086551464269e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10603504,
      "step": 18670
    },
    {
      "epoch": 327.6371681415929,
      "grad_norm": 8.761558660808078e-07,
      "learning_rate": 2.759889106176006e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10606112,
      "step": 18675
    },
    {
      "epoch": 327.72566371681415,
      "grad_norm": 9.204754860547837e-07,
      "learning_rate": 2.758912657631156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10609136,
      "step": 18680
    },
    {
      "epoch": 327.8141592920354,
      "grad_norm": 5.610632797470316e-06,
      "learning_rate": 2.7579361691587198e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10611664,
      "step": 18685
    },
    {
      "epoch": 327.9026548672566,
      "grad_norm": 5.664867330779089e-07,
      "learning_rate": 2.756959640909285e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10614624,
      "step": 18690
    },
    {
      "epoch": 327.9911504424779,
      "grad_norm": 5.445397164294263e-07,
      "learning_rate": 2.7559830730334452e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10618032,
      "step": 18695
    },
    {
      "epoch": 328.070796460177,
      "grad_norm": 1.9231727037549717e-06,
      "learning_rate": 2.7550064656817988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10619992,
      "step": 18700
    },
    {
      "epoch": 328.1592920353982,
      "grad_norm": 2.619645442791807e-07,
      "learning_rate": 2.7540298190049503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10622616,
      "step": 18705
    },
    {
      "epoch": 328.24778761061947,
      "grad_norm": 4.297215525639331e-07,
      "learning_rate": 2.7530531331535107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10625608,
      "step": 18710
    },
    {
      "epoch": 328.3362831858407,
      "grad_norm": 5.383823804550047e-07,
      "learning_rate": 2.752076408278099e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10629160,
      "step": 18715
    },
    {
      "epoch": 328.42477876106193,
      "grad_norm": 9.651066648075357e-07,
      "learning_rate": 2.751099644529337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10632072,
      "step": 18720
    },
    {
      "epoch": 328.5132743362832,
      "grad_norm": 3.0679038331982156e-07,
      "learning_rate": 2.7501228420578533e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10634872,
      "step": 18725
    },
    {
      "epoch": 328.60176991150445,
      "grad_norm": 4.886910573986825e-06,
      "learning_rate": 2.7491460010142857e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10637656,
      "step": 18730
    },
    {
      "epoch": 328.69026548672565,
      "grad_norm": 1.0623888329064357e-06,
      "learning_rate": 2.7481691215492727e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10640456,
      "step": 18735
    },
    {
      "epoch": 328.7787610619469,
      "grad_norm": 2.6547779725660803e-06,
      "learning_rate": 2.747192203813463e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10643512,
      "step": 18740
    },
    {
      "epoch": 328.86725663716817,
      "grad_norm": 6.583815661542758e-07,
      "learning_rate": 2.7462152479575087e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10646280,
      "step": 18745
    },
    {
      "epoch": 328.95575221238937,
      "grad_norm": 1.1257124015173758e-06,
      "learning_rate": 2.7452382541320697e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10649288,
      "step": 18750
    },
    {
      "epoch": 329.0353982300885,
      "grad_norm": 5.129854798724409e-07,
      "learning_rate": 2.7442612224878096e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10651880,
      "step": 18755
    },
    {
      "epoch": 329.12389380530976,
      "grad_norm": 7.802491381880827e-07,
      "learning_rate": 2.7432841531753994e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10654648,
      "step": 18760
    },
    {
      "epoch": 329.21238938053096,
      "grad_norm": 9.269394922739593e-07,
      "learning_rate": 2.7423070463455147e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10657496,
      "step": 18765
    },
    {
      "epoch": 329.3008849557522,
      "grad_norm": 2.727800392676727e-06,
      "learning_rate": 2.7413299021488397e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10660216,
      "step": 18770
    },
    {
      "epoch": 329.3893805309734,
      "grad_norm": 1.009297079690441e-06,
      "learning_rate": 2.7403527207360615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10663112,
      "step": 18775
    },
    {
      "epoch": 329.4778761061947,
      "grad_norm": 2.393344175288803e-06,
      "learning_rate": 2.7393755022578722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10665928,
      "step": 18780
    },
    {
      "epoch": 329.56637168141594,
      "grad_norm": 1.048073272613692e-06,
      "learning_rate": 2.7383982468649714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10669080,
      "step": 18785
    },
    {
      "epoch": 329.65486725663715,
      "grad_norm": 1.7688419120531762e-06,
      "learning_rate": 2.7374209547080665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10672296,
      "step": 18790
    },
    {
      "epoch": 329.7433628318584,
      "grad_norm": 7.876768108872056e-07,
      "learning_rate": 2.7364436259378663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10674904,
      "step": 18795
    },
    {
      "epoch": 329.83185840707966,
      "grad_norm": 1.8573233546703705e-06,
      "learning_rate": 2.735466260705088e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10677928,
      "step": 18800
    },
    {
      "epoch": 329.83185840707966,
      "eval_loss": 0.5949233770370483,
      "eval_runtime": 1.0677,
      "eval_samples_per_second": 23.414,
      "eval_steps_per_second": 12.175,
      "num_input_tokens_seen": 10677928,
      "step": 18800
    },
    {
      "epoch": 329.92035398230087,
      "grad_norm": 4.401902060635621e-06,
      "learning_rate": 2.7344888591604524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10681080,
      "step": 18805
    },
    {
      "epoch": 330.0,
      "grad_norm": 3.9845249375503045e-06,
      "learning_rate": 2.7335114214546893e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10683296,
      "step": 18810
    },
    {
      "epoch": 330.08849557522126,
      "grad_norm": 9.453914913137851e-07,
      "learning_rate": 2.7325339477385293e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10686096,
      "step": 18815
    },
    {
      "epoch": 330.17699115044246,
      "grad_norm": 9.219397156812192e-07,
      "learning_rate": 2.7315564381627128e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10689152,
      "step": 18820
    },
    {
      "epoch": 330.2654867256637,
      "grad_norm": 7.48140450923529e-07,
      "learning_rate": 2.7305788928779835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10691856,
      "step": 18825
    },
    {
      "epoch": 330.353982300885,
      "grad_norm": 6.360483553180529e-07,
      "learning_rate": 2.729601312035091e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10694768,
      "step": 18830
    },
    {
      "epoch": 330.4424778761062,
      "grad_norm": 2.570102651588968e-06,
      "learning_rate": 2.7286236957847915e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10697328,
      "step": 18835
    },
    {
      "epoch": 330.53097345132744,
      "grad_norm": 1.457780285818444e-06,
      "learning_rate": 2.7276460442778446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10699616,
      "step": 18840
    },
    {
      "epoch": 330.6194690265487,
      "grad_norm": 1.685001279838616e-06,
      "learning_rate": 2.726668357665017e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10702736,
      "step": 18845
    },
    {
      "epoch": 330.7079646017699,
      "grad_norm": 3.6303749766375404e-06,
      "learning_rate": 2.7256906360970808e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10705728,
      "step": 18850
    },
    {
      "epoch": 330.79646017699116,
      "grad_norm": 1.1613744845817564e-06,
      "learning_rate": 2.7247128797248117e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10708784,
      "step": 18855
    },
    {
      "epoch": 330.88495575221236,
      "grad_norm": 6.461030466198281e-07,
      "learning_rate": 2.7237350886989925e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10711296,
      "step": 18860
    },
    {
      "epoch": 330.9734513274336,
      "grad_norm": 6.111764037086687e-07,
      "learning_rate": 2.7227572631704107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10714576,
      "step": 18865
    },
    {
      "epoch": 331.05309734513276,
      "grad_norm": 1.2428100717443158e-06,
      "learning_rate": 2.7217794032898596e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10717048,
      "step": 18870
    },
    {
      "epoch": 331.14159292035396,
      "grad_norm": 1.0004786190620507e-06,
      "learning_rate": 2.7208015092081384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10719976,
      "step": 18875
    },
    {
      "epoch": 331.2300884955752,
      "grad_norm": 6.304333055595635e-07,
      "learning_rate": 2.719823581076049e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10722696,
      "step": 18880
    },
    {
      "epoch": 331.3185840707965,
      "grad_norm": 4.0166383996620425e-07,
      "learning_rate": 2.718845619044401e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10725816,
      "step": 18885
    },
    {
      "epoch": 331.4070796460177,
      "grad_norm": 2.1835082861798583e-06,
      "learning_rate": 2.7178676232640088e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10728728,
      "step": 18890
    },
    {
      "epoch": 331.49557522123894,
      "grad_norm": 7.9651852047391e-07,
      "learning_rate": 2.716889593885691e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10732424,
      "step": 18895
    },
    {
      "epoch": 331.5840707964602,
      "grad_norm": 2.0367655451991595e-05,
      "learning_rate": 2.7159115310602716e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10735064,
      "step": 18900
    },
    {
      "epoch": 331.6725663716814,
      "grad_norm": 5.432602847577073e-07,
      "learning_rate": 2.7149334349385814e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10737944,
      "step": 18905
    },
    {
      "epoch": 331.76106194690266,
      "grad_norm": 3.654848228507035e-07,
      "learning_rate": 2.713955305671454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10740568,
      "step": 18910
    },
    {
      "epoch": 331.8495575221239,
      "grad_norm": 3.8732659390916524e-07,
      "learning_rate": 2.71297714340973e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10743384,
      "step": 18915
    },
    {
      "epoch": 331.9380530973451,
      "grad_norm": 3.5080695397482486e-06,
      "learning_rate": 2.7119989483042545e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10745736,
      "step": 18920
    },
    {
      "epoch": 332.01769911504425,
      "grad_norm": 1.035004288496566e-06,
      "learning_rate": 2.7110207205058768e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10748000,
      "step": 18925
    },
    {
      "epoch": 332.1061946902655,
      "grad_norm": 5.857605742676242e-07,
      "learning_rate": 2.7100424601654517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10750432,
      "step": 18930
    },
    {
      "epoch": 332.1946902654867,
      "grad_norm": 4.702230853581568e-06,
      "learning_rate": 2.7090641674338403e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10753488,
      "step": 18935
    },
    {
      "epoch": 332.283185840708,
      "grad_norm": 8.699457225702645e-07,
      "learning_rate": 2.7080858424619072e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10756592,
      "step": 18940
    },
    {
      "epoch": 332.37168141592923,
      "grad_norm": 7.247735425153223e-07,
      "learning_rate": 2.707107485400521e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10759424,
      "step": 18945
    },
    {
      "epoch": 332.46017699115043,
      "grad_norm": 1.5912951312202495e-06,
      "learning_rate": 2.7061290964005586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10762192,
      "step": 18950
    },
    {
      "epoch": 332.5486725663717,
      "grad_norm": 1.1950821772188647e-06,
      "learning_rate": 2.7051506756129e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10764912,
      "step": 18955
    },
    {
      "epoch": 332.6371681415929,
      "grad_norm": 1.4468308791037998e-06,
      "learning_rate": 2.704172223188428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10767872,
      "step": 18960
    },
    {
      "epoch": 332.72566371681415,
      "grad_norm": 6.240411494218279e-07,
      "learning_rate": 2.7031937392780334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10771152,
      "step": 18965
    },
    {
      "epoch": 332.8141592920354,
      "grad_norm": 2.2736130631528795e-06,
      "learning_rate": 2.702215224032611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10774000,
      "step": 18970
    },
    {
      "epoch": 332.9026548672566,
      "grad_norm": 1.9171716303389985e-06,
      "learning_rate": 2.70123667760306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10776688,
      "step": 18975
    },
    {
      "epoch": 332.9911504424779,
      "grad_norm": 2.564399665061501e-06,
      "learning_rate": 2.7002581001402845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10779600,
      "step": 18980
    },
    {
      "epoch": 333.070796460177,
      "grad_norm": 4.117547632631613e-06,
      "learning_rate": 2.6992794917951923e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10781936,
      "step": 18985
    },
    {
      "epoch": 333.1592920353982,
      "grad_norm": 4.82935593026923e-06,
      "learning_rate": 2.6983008527187e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10784768,
      "step": 18990
    },
    {
      "epoch": 333.24778761061947,
      "grad_norm": 7.2776133492880035e-06,
      "learning_rate": 2.697322183061723e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10788048,
      "step": 18995
    },
    {
      "epoch": 333.3362831858407,
      "grad_norm": 1.5479922694794368e-06,
      "learning_rate": 2.696343482975186e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10790896,
      "step": 19000
    },
    {
      "epoch": 333.3362831858407,
      "eval_loss": 0.5917893648147583,
      "eval_runtime": 1.0733,
      "eval_samples_per_second": 23.293,
      "eval_steps_per_second": 12.112,
      "num_input_tokens_seen": 10790896,
      "step": 19000
    },
    {
      "epoch": 333.42477876106193,
      "grad_norm": 4.502729211708356e-07,
      "learning_rate": 2.695364752610016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10793696,
      "step": 19005
    },
    {
      "epoch": 333.5132743362832,
      "grad_norm": 5.966038543192553e-07,
      "learning_rate": 2.6943859921171467e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10796880,
      "step": 19010
    },
    {
      "epoch": 333.60176991150445,
      "grad_norm": 3.838677457679296e-06,
      "learning_rate": 2.6934072016475143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10799472,
      "step": 19015
    },
    {
      "epoch": 333.69026548672565,
      "grad_norm": 2.859986807379755e-06,
      "learning_rate": 2.6924283813520606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10801856,
      "step": 19020
    },
    {
      "epoch": 333.7787610619469,
      "grad_norm": 6.642692824243568e-06,
      "learning_rate": 2.691449531381733e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10804864,
      "step": 19025
    },
    {
      "epoch": 333.86725663716817,
      "grad_norm": 9.165086112261633e-07,
      "learning_rate": 2.6904706518874816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10807968,
      "step": 19030
    },
    {
      "epoch": 333.95575221238937,
      "grad_norm": 4.002824425697327e-06,
      "learning_rate": 2.6894917430202615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10811104,
      "step": 19035
    },
    {
      "epoch": 334.0353982300885,
      "grad_norm": 7.088035545166349e-07,
      "learning_rate": 2.6885128049310343e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10813792,
      "step": 19040
    },
    {
      "epoch": 334.12389380530976,
      "grad_norm": 1.3917489241066505e-06,
      "learning_rate": 2.687533837770762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10816576,
      "step": 19045
    },
    {
      "epoch": 334.21238938053096,
      "grad_norm": 7.489538234040083e-07,
      "learning_rate": 2.6865548416904162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10819632,
      "step": 19050
    },
    {
      "epoch": 334.3008849557522,
      "grad_norm": 2.6095872271980625e-06,
      "learning_rate": 2.68557581684097e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10822992,
      "step": 19055
    },
    {
      "epoch": 334.3893805309734,
      "grad_norm": 1.199726170852955e-06,
      "learning_rate": 2.6845967633733998e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10825696,
      "step": 19060
    },
    {
      "epoch": 334.4778761061947,
      "grad_norm": 2.724027353906422e-06,
      "learning_rate": 2.683617681438689e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10828432,
      "step": 19065
    },
    {
      "epoch": 334.56637168141594,
      "grad_norm": 2.410141632935847e-06,
      "learning_rate": 2.682638571187825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10831520,
      "step": 19070
    },
    {
      "epoch": 334.65486725663715,
      "grad_norm": 8.384960210605641e-07,
      "learning_rate": 2.6816594327717976e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10833904,
      "step": 19075
    },
    {
      "epoch": 334.7433628318584,
      "grad_norm": 1.663195348555746e-06,
      "learning_rate": 2.680680266341603e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10836848,
      "step": 19080
    },
    {
      "epoch": 334.83185840707966,
      "grad_norm": 6.955223170734826e-07,
      "learning_rate": 2.67970107204824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10839440,
      "step": 19085
    },
    {
      "epoch": 334.92035398230087,
      "grad_norm": 1.002252133730508e-06,
      "learning_rate": 2.6787218500427142e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10842144,
      "step": 19090
    },
    {
      "epoch": 335.0,
      "grad_norm": 1.540973607916385e-05,
      "learning_rate": 2.6777426004760332e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10844392,
      "step": 19095
    },
    {
      "epoch": 335.08849557522126,
      "grad_norm": 6.325814752017322e-07,
      "learning_rate": 2.6767633234992094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10847320,
      "step": 19100
    },
    {
      "epoch": 335.17699115044246,
      "grad_norm": 1.6452529507660074e-06,
      "learning_rate": 2.6757840192632598e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10850632,
      "step": 19105
    },
    {
      "epoch": 335.2654867256637,
      "grad_norm": 1.3760932233708445e-06,
      "learning_rate": 2.6748046879192052e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10853352,
      "step": 19110
    },
    {
      "epoch": 335.353982300885,
      "grad_norm": 6.545187716255896e-07,
      "learning_rate": 2.673825329618071e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10855960,
      "step": 19115
    },
    {
      "epoch": 335.4424778761062,
      "grad_norm": 2.8004617433907697e-06,
      "learning_rate": 2.6728459445108866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10858568,
      "step": 19120
    },
    {
      "epoch": 335.53097345132744,
      "grad_norm": 8.797383088676725e-07,
      "learning_rate": 2.6718665327486854e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10861432,
      "step": 19125
    },
    {
      "epoch": 335.6194690265487,
      "grad_norm": 3.34503170051903e-06,
      "learning_rate": 2.6708870944825048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10864664,
      "step": 19130
    },
    {
      "epoch": 335.7079646017699,
      "grad_norm": 1.1115128017991083e-06,
      "learning_rate": 2.6699076298633874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10867512,
      "step": 19135
    },
    {
      "epoch": 335.79646017699116,
      "grad_norm": 1.7409275869795238e-06,
      "learning_rate": 2.6689281390423788e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10870888,
      "step": 19140
    },
    {
      "epoch": 335.88495575221236,
      "grad_norm": 1.4457719771598931e-05,
      "learning_rate": 2.667948622170527e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10873768,
      "step": 19145
    },
    {
      "epoch": 335.9734513274336,
      "grad_norm": 5.881453830625105e-07,
      "learning_rate": 2.6669690793988873e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10876248,
      "step": 19150
    },
    {
      "epoch": 336.05309734513276,
      "grad_norm": 3.1464503535971744e-07,
      "learning_rate": 2.665989510878518e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10878952,
      "step": 19155
    },
    {
      "epoch": 336.14159292035396,
      "grad_norm": 4.329896512444975e-07,
      "learning_rate": 2.6650099167604793e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10881896,
      "step": 19160
    },
    {
      "epoch": 336.2300884955752,
      "grad_norm": 2.1098016134146746e-07,
      "learning_rate": 2.6640302971958376e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10885224,
      "step": 19165
    },
    {
      "epoch": 336.3185840707965,
      "grad_norm": 3.599247406782524e-07,
      "learning_rate": 2.6630506523356635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10888120,
      "step": 19170
    },
    {
      "epoch": 336.4070796460177,
      "grad_norm": 2.4826908884278964e-07,
      "learning_rate": 2.6620709823310297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10890808,
      "step": 19175
    },
    {
      "epoch": 336.49557522123894,
      "grad_norm": 3.860282902223844e-07,
      "learning_rate": 2.661091287333014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10893752,
      "step": 19180
    },
    {
      "epoch": 336.5840707964602,
      "grad_norm": 1.349076569567842e-06,
      "learning_rate": 2.660111567492696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10896536,
      "step": 19185
    },
    {
      "epoch": 336.6725663716814,
      "grad_norm": 5.648759042742313e-07,
      "learning_rate": 2.6591318229611635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10899192,
      "step": 19190
    },
    {
      "epoch": 336.76106194690266,
      "grad_norm": 4.37874092540369e-07,
      "learning_rate": 2.6581520538895037e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10902008,
      "step": 19195
    },
    {
      "epoch": 336.8495575221239,
      "grad_norm": 4.3506187807906826e-07,
      "learning_rate": 2.6571722604288102e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10904600,
      "step": 19200
    },
    {
      "epoch": 336.8495575221239,
      "eval_loss": 0.5861619710922241,
      "eval_runtime": 1.0683,
      "eval_samples_per_second": 23.403,
      "eval_steps_per_second": 12.169,
      "num_input_tokens_seen": 10904600,
      "step": 19200
    },
    {
      "epoch": 336.9380530973451,
      "grad_norm": 3.359095614996477e-07,
      "learning_rate": 2.656192442730179e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10907928,
      "step": 19205
    },
    {
      "epoch": 337.01769911504425,
      "grad_norm": 1.3729622878599912e-06,
      "learning_rate": 2.6552126009447098e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10910000,
      "step": 19210
    },
    {
      "epoch": 337.1061946902655,
      "grad_norm": 5.301912437971623e-07,
      "learning_rate": 2.654232735223507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10912960,
      "step": 19215
    },
    {
      "epoch": 337.1946902654867,
      "grad_norm": 4.942333475810301e-07,
      "learning_rate": 2.6532528457176787e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10915760,
      "step": 19220
    },
    {
      "epoch": 337.283185840708,
      "grad_norm": 5.672115548804868e-06,
      "learning_rate": 2.6522729325783348e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10919184,
      "step": 19225
    },
    {
      "epoch": 337.37168141592923,
      "grad_norm": 3.2172476949199336e-06,
      "learning_rate": 2.6512929959565914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10922128,
      "step": 19230
    },
    {
      "epoch": 337.46017699115043,
      "grad_norm": 7.631489893356047e-07,
      "learning_rate": 2.6503130360035673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10924864,
      "step": 19235
    },
    {
      "epoch": 337.5486725663717,
      "grad_norm": 5.852663775840483e-07,
      "learning_rate": 2.6493330528703835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10927936,
      "step": 19240
    },
    {
      "epoch": 337.6371681415929,
      "grad_norm": 9.648538252804428e-07,
      "learning_rate": 2.648353046708167e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10930928,
      "step": 19245
    },
    {
      "epoch": 337.72566371681415,
      "grad_norm": 7.86037844591192e-07,
      "learning_rate": 2.647373017668046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10933792,
      "step": 19250
    },
    {
      "epoch": 337.8141592920354,
      "grad_norm": 8.686964747539605e-07,
      "learning_rate": 2.6463929659011537e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10936432,
      "step": 19255
    },
    {
      "epoch": 337.9026548672566,
      "grad_norm": 7.199180913630698e-07,
      "learning_rate": 2.6454128915586262e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10939152,
      "step": 19260
    },
    {
      "epoch": 337.9911504424779,
      "grad_norm": 7.513874606956961e-06,
      "learning_rate": 2.6444327947916036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10941792,
      "step": 19265
    },
    {
      "epoch": 338.070796460177,
      "grad_norm": 8.369264037355606e-07,
      "learning_rate": 2.6434526757512292e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10944112,
      "step": 19270
    },
    {
      "epoch": 338.1592920353982,
      "grad_norm": 3.5465595829009544e-06,
      "learning_rate": 2.6424725345886486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10947408,
      "step": 19275
    },
    {
      "epoch": 338.24778761061947,
      "grad_norm": 4.266461928637e-06,
      "learning_rate": 2.641492371455014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10950048,
      "step": 19280
    },
    {
      "epoch": 338.3362831858407,
      "grad_norm": 1.1211950550205074e-06,
      "learning_rate": 2.640512186501477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10953424,
      "step": 19285
    },
    {
      "epoch": 338.42477876106193,
      "grad_norm": 1.3231473303676466e-06,
      "learning_rate": 2.639531979879195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10956016,
      "step": 19290
    },
    {
      "epoch": 338.5132743362832,
      "grad_norm": 2.5687249944894575e-06,
      "learning_rate": 2.638551751739328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10958960,
      "step": 19295
    },
    {
      "epoch": 338.60176991150445,
      "grad_norm": 1.3669398413185263e-06,
      "learning_rate": 2.6375715022330404e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10961824,
      "step": 19300
    },
    {
      "epoch": 338.69026548672565,
      "grad_norm": 3.357558398420224e-06,
      "learning_rate": 2.6365912315114976e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10964368,
      "step": 19305
    },
    {
      "epoch": 338.7787610619469,
      "grad_norm": 7.798950036885799e-07,
      "learning_rate": 2.6356109397258704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10967520,
      "step": 19310
    },
    {
      "epoch": 338.86725663716817,
      "grad_norm": 1.8390284139968571e-06,
      "learning_rate": 2.6346306270273325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10970288,
      "step": 19315
    },
    {
      "epoch": 338.95575221238937,
      "grad_norm": 1.2224110150782508e-06,
      "learning_rate": 2.6336502935670608e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10973344,
      "step": 19320
    },
    {
      "epoch": 339.0353982300885,
      "grad_norm": 1.0901776477112435e-06,
      "learning_rate": 2.6326699394962333e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10975544,
      "step": 19325
    },
    {
      "epoch": 339.12389380530976,
      "grad_norm": 1.9632514067779994e-06,
      "learning_rate": 2.6316895649660334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10978328,
      "step": 19330
    },
    {
      "epoch": 339.21238938053096,
      "grad_norm": 8.922364145291795e-07,
      "learning_rate": 2.6307091701276486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10980952,
      "step": 19335
    },
    {
      "epoch": 339.3008849557522,
      "grad_norm": 1.0628372137944098e-06,
      "learning_rate": 2.629728755132267e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10983928,
      "step": 19340
    },
    {
      "epoch": 339.3893805309734,
      "grad_norm": 9.72132738752407e-07,
      "learning_rate": 2.628748320131081e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10987000,
      "step": 19345
    },
    {
      "epoch": 339.4778761061947,
      "grad_norm": 6.278537966863951e-06,
      "learning_rate": 2.6277678652752856e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10990520,
      "step": 19350
    },
    {
      "epoch": 339.56637168141594,
      "grad_norm": 8.517238825334061e-07,
      "learning_rate": 2.6267873907160807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10993288,
      "step": 19355
    },
    {
      "epoch": 339.65486725663715,
      "grad_norm": 3.2390471460530534e-06,
      "learning_rate": 2.6258068966046668e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10996008,
      "step": 19360
    },
    {
      "epoch": 339.7433628318584,
      "grad_norm": 7.156519927775662e-07,
      "learning_rate": 2.6248263830922475e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 10999176,
      "step": 19365
    },
    {
      "epoch": 339.83185840707966,
      "grad_norm": 9.202912565342558e-07,
      "learning_rate": 2.6238458503300318e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11001864,
      "step": 19370
    },
    {
      "epoch": 339.92035398230087,
      "grad_norm": 2.331525820409297e-06,
      "learning_rate": 2.6228652984692292e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11004408,
      "step": 19375
    },
    {
      "epoch": 340.0,
      "grad_norm": 6.190380759107939e-07,
      "learning_rate": 2.621884727661054e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11006656,
      "step": 19380
    },
    {
      "epoch": 340.08849557522126,
      "grad_norm": 1.0507631031941855e-06,
      "learning_rate": 2.6209041380567222e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11009664,
      "step": 19385
    },
    {
      "epoch": 340.17699115044246,
      "grad_norm": 4.435707410266332e-07,
      "learning_rate": 2.6199235298074527e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11012592,
      "step": 19390
    },
    {
      "epoch": 340.2654867256637,
      "grad_norm": 1.4829624888079707e-06,
      "learning_rate": 2.618942903064468e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11015328,
      "step": 19395
    },
    {
      "epoch": 340.353982300885,
      "grad_norm": 1.4253884728532284e-05,
      "learning_rate": 2.6179622579789932e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11018112,
      "step": 19400
    },
    {
      "epoch": 340.353982300885,
      "eval_loss": 0.562694251537323,
      "eval_runtime": 1.064,
      "eval_samples_per_second": 23.497,
      "eval_steps_per_second": 12.219,
      "num_input_tokens_seen": 11018112,
      "step": 19400
    },
    {
      "epoch": 340.4424778761062,
      "grad_norm": 5.214469638303854e-07,
      "learning_rate": 2.6169815947022553e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11021200,
      "step": 19405
    },
    {
      "epoch": 340.53097345132744,
      "grad_norm": 1.4625978792537353e-06,
      "learning_rate": 2.6160009133854853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11024320,
      "step": 19410
    },
    {
      "epoch": 340.6194690265487,
      "grad_norm": 7.958688570397499e-07,
      "learning_rate": 2.6150202141799168e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11027104,
      "step": 19415
    },
    {
      "epoch": 340.7079646017699,
      "grad_norm": 6.630566531384829e-07,
      "learning_rate": 2.614039497236786e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11029856,
      "step": 19420
    },
    {
      "epoch": 340.79646017699116,
      "grad_norm": 2.740978061410715e-06,
      "learning_rate": 2.6130587627073315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11032640,
      "step": 19425
    },
    {
      "epoch": 340.88495575221236,
      "grad_norm": 1.7665129234956112e-06,
      "learning_rate": 2.6120780107427956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11035648,
      "step": 19430
    },
    {
      "epoch": 340.9734513274336,
      "grad_norm": 1.9356109532964183e-06,
      "learning_rate": 2.6110972414944214e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11038416,
      "step": 19435
    },
    {
      "epoch": 341.05309734513276,
      "grad_norm": 9.10253959318652e-07,
      "learning_rate": 2.6101164551134565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11040640,
      "step": 19440
    },
    {
      "epoch": 341.14159292035396,
      "grad_norm": 1.0424683978271787e-06,
      "learning_rate": 2.6091356517511505e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11043136,
      "step": 19445
    },
    {
      "epoch": 341.2300884955752,
      "grad_norm": 3.035582722077379e-06,
      "learning_rate": 2.608154831558755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11046432,
      "step": 19450
    },
    {
      "epoch": 341.3185840707965,
      "grad_norm": 5.993778131596628e-07,
      "learning_rate": 2.607173994687526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11049520,
      "step": 19455
    },
    {
      "epoch": 341.4070796460177,
      "grad_norm": 2.6251016151945805e-06,
      "learning_rate": 2.6061931412887196e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11052496,
      "step": 19460
    },
    {
      "epoch": 341.49557522123894,
      "grad_norm": 6.179114961923915e-07,
      "learning_rate": 2.6052122715135973e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11055472,
      "step": 19465
    },
    {
      "epoch": 341.5840707964602,
      "grad_norm": 2.3984841845958726e-06,
      "learning_rate": 2.60423138551342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11058000,
      "step": 19470
    },
    {
      "epoch": 341.6725663716814,
      "grad_norm": 6.886817232043541e-07,
      "learning_rate": 2.6032504834394527e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11060640,
      "step": 19475
    },
    {
      "epoch": 341.76106194690266,
      "grad_norm": 1.3689997331312043e-06,
      "learning_rate": 2.602269565442964e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11064000,
      "step": 19480
    },
    {
      "epoch": 341.8495575221239,
      "grad_norm": 2.928814183178474e-06,
      "learning_rate": 2.6012886316752227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11067120,
      "step": 19485
    },
    {
      "epoch": 341.9380530973451,
      "grad_norm": 1.4531691476804554e-06,
      "learning_rate": 2.6003076822875018e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11069984,
      "step": 19490
    },
    {
      "epoch": 342.01769911504425,
      "grad_norm": 7.904534413682995e-07,
      "learning_rate": 2.5993267174310755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11072296,
      "step": 19495
    },
    {
      "epoch": 342.1061946902655,
      "grad_norm": 4.2795306853804504e-07,
      "learning_rate": 2.5983457372572218e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11075384,
      "step": 19500
    },
    {
      "epoch": 342.1946902654867,
      "grad_norm": 8.608152711531147e-06,
      "learning_rate": 2.597364741917219e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11078440,
      "step": 19505
    },
    {
      "epoch": 342.283185840708,
      "grad_norm": 2.44549323724641e-06,
      "learning_rate": 2.5963837315623492e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11081256,
      "step": 19510
    },
    {
      "epoch": 342.37168141592923,
      "grad_norm": 1.3516206536223763e-06,
      "learning_rate": 2.595402706343897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11084200,
      "step": 19515
    },
    {
      "epoch": 342.46017699115043,
      "grad_norm": 6.260859208850889e-07,
      "learning_rate": 2.594421666413148e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11087176,
      "step": 19520
    },
    {
      "epoch": 342.5486725663717,
      "grad_norm": 1.7001278820316656e-06,
      "learning_rate": 2.5934406119213928e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11089848,
      "step": 19525
    },
    {
      "epoch": 342.6371681415929,
      "grad_norm": 5.337498691915243e-07,
      "learning_rate": 2.5924595430199193e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11092376,
      "step": 19530
    },
    {
      "epoch": 342.72566371681415,
      "grad_norm": 6.874461178085767e-07,
      "learning_rate": 2.5914784598600238e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11095128,
      "step": 19535
    },
    {
      "epoch": 342.8141592920354,
      "grad_norm": 1.7778911569621414e-06,
      "learning_rate": 2.5904973625930002e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11098024,
      "step": 19540
    },
    {
      "epoch": 342.9026548672566,
      "grad_norm": 1.6453022908535786e-06,
      "learning_rate": 2.5895162513701456e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11100968,
      "step": 19545
    },
    {
      "epoch": 342.9911504424779,
      "grad_norm": 1.0276188504576567e-06,
      "learning_rate": 2.5885351263427593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11103816,
      "step": 19550
    },
    {
      "epoch": 343.070796460177,
      "grad_norm": 4.4395454779078136e-07,
      "learning_rate": 2.5875539876621448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11106368,
      "step": 19555
    },
    {
      "epoch": 343.1592920353982,
      "grad_norm": 4.405899744597264e-06,
      "learning_rate": 2.586572835479605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11109088,
      "step": 19560
    },
    {
      "epoch": 343.24778761061947,
      "grad_norm": 8.752966209613078e-07,
      "learning_rate": 2.585591669946446e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11111712,
      "step": 19565
    },
    {
      "epoch": 343.3362831858407,
      "grad_norm": 3.1104980280360905e-07,
      "learning_rate": 2.5846104912139756e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11114560,
      "step": 19570
    },
    {
      "epoch": 343.42477876106193,
      "grad_norm": 2.3242030238179723e-06,
      "learning_rate": 2.583629299433505e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11117568,
      "step": 19575
    },
    {
      "epoch": 343.5132743362832,
      "grad_norm": 2.4353316803171765e-06,
      "learning_rate": 2.582648094756345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11120320,
      "step": 19580
    },
    {
      "epoch": 343.60176991150445,
      "grad_norm": 5.485100587065972e-07,
      "learning_rate": 2.5816668773338098e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11123248,
      "step": 19585
    },
    {
      "epoch": 343.69026548672565,
      "grad_norm": 3.5106342011204106e-07,
      "learning_rate": 2.580685647317216e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11125840,
      "step": 19590
    },
    {
      "epoch": 343.7787610619469,
      "grad_norm": 3.3400161214558466e-07,
      "learning_rate": 2.5797044048578818e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11128672,
      "step": 19595
    },
    {
      "epoch": 343.86725663716817,
      "grad_norm": 1.2676561027546995e-06,
      "learning_rate": 2.5787231501071262e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11131712,
      "step": 19600
    },
    {
      "epoch": 343.86725663716817,
      "eval_loss": 0.6011966466903687,
      "eval_runtime": 1.064,
      "eval_samples_per_second": 23.496,
      "eval_steps_per_second": 12.218,
      "num_input_tokens_seen": 11131712,
      "step": 19600
    },
    {
      "epoch": 343.95575221238937,
      "grad_norm": 3.453942838405055e-07,
      "learning_rate": 2.577741883216272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11134928,
      "step": 19605
    },
    {
      "epoch": 344.0353982300885,
      "grad_norm": 7.591860935463046e-07,
      "learning_rate": 2.576760604336642e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11137648,
      "step": 19610
    },
    {
      "epoch": 344.12389380530976,
      "grad_norm": 9.948334991349839e-07,
      "learning_rate": 2.575779313619563e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11140272,
      "step": 19615
    },
    {
      "epoch": 344.21238938053096,
      "grad_norm": 2.2423543555305514e-07,
      "learning_rate": 2.5747980112163605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11142896,
      "step": 19620
    },
    {
      "epoch": 344.3008849557522,
      "grad_norm": 1.436952857147844e-06,
      "learning_rate": 2.5738166972783656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11146208,
      "step": 19625
    },
    {
      "epoch": 344.3893805309734,
      "grad_norm": 6.004481178933929e-07,
      "learning_rate": 2.5728353719569075e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11149584,
      "step": 19630
    },
    {
      "epoch": 344.4778761061947,
      "grad_norm": 1.4127193708191044e-06,
      "learning_rate": 2.57185403540332e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11152160,
      "step": 19635
    },
    {
      "epoch": 344.56637168141594,
      "grad_norm": 1.3239592817626544e-06,
      "learning_rate": 2.5708726877689375e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11154768,
      "step": 19640
    },
    {
      "epoch": 344.65486725663715,
      "grad_norm": 3.264492761445581e-06,
      "learning_rate": 2.5698913292050964e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11157456,
      "step": 19645
    },
    {
      "epoch": 344.7433628318584,
      "grad_norm": 8.914717568586639e-07,
      "learning_rate": 2.568909959863133e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11160576,
      "step": 19650
    },
    {
      "epoch": 344.83185840707966,
      "grad_norm": 1.2258892638783436e-06,
      "learning_rate": 2.5679285798943887e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11163520,
      "step": 19655
    },
    {
      "epoch": 344.92035398230087,
      "grad_norm": 8.724882150090707e-07,
      "learning_rate": 2.5669471894502035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11166160,
      "step": 19660
    },
    {
      "epoch": 345.0,
      "grad_norm": 5.443516215564159e-07,
      "learning_rate": 2.56596578868192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11168448,
      "step": 19665
    },
    {
      "epoch": 345.08849557522126,
      "grad_norm": 4.57291008615357e-07,
      "learning_rate": 2.564984377740883e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11170944,
      "step": 19670
    },
    {
      "epoch": 345.17699115044246,
      "grad_norm": 1.6182278841370135e-06,
      "learning_rate": 2.564002956778438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11173616,
      "step": 19675
    },
    {
      "epoch": 345.2654867256637,
      "grad_norm": 7.952128839860961e-07,
      "learning_rate": 2.563021525945934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11176368,
      "step": 19680
    },
    {
      "epoch": 345.353982300885,
      "grad_norm": 1.040109850691806e-06,
      "learning_rate": 2.562040085394718e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11179408,
      "step": 19685
    },
    {
      "epoch": 345.4424778761062,
      "grad_norm": 2.174917199226911e-06,
      "learning_rate": 2.56105863527614e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11182304,
      "step": 19690
    },
    {
      "epoch": 345.53097345132744,
      "grad_norm": 3.4095414775947575e-06,
      "learning_rate": 2.5600771757415548e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11185616,
      "step": 19695
    },
    {
      "epoch": 345.6194690265487,
      "grad_norm": 1.4687100247101625e-06,
      "learning_rate": 2.5590957069423134e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11188656,
      "step": 19700
    },
    {
      "epoch": 345.7079646017699,
      "grad_norm": 1.99892247110256e-06,
      "learning_rate": 2.5581142290297716e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11191712,
      "step": 19705
    },
    {
      "epoch": 345.79646017699116,
      "grad_norm": 1.031002170748252e-06,
      "learning_rate": 2.557132742155285e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11194480,
      "step": 19710
    },
    {
      "epoch": 345.88495575221236,
      "grad_norm": 4.66521441921941e-07,
      "learning_rate": 2.556151246470212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11197456,
      "step": 19715
    },
    {
      "epoch": 345.9734513274336,
      "grad_norm": 1.1942695891775656e-05,
      "learning_rate": 2.5551697421259114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11200352,
      "step": 19720
    },
    {
      "epoch": 346.05309734513276,
      "grad_norm": 3.7404268482532643e-07,
      "learning_rate": 2.554188229273743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11202480,
      "step": 19725
    },
    {
      "epoch": 346.14159292035396,
      "grad_norm": 7.80052573645662e-07,
      "learning_rate": 2.5532067080650678e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11204992,
      "step": 19730
    },
    {
      "epoch": 346.2300884955752,
      "grad_norm": 1.0818944247148465e-06,
      "learning_rate": 2.55222517865125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11208208,
      "step": 19735
    },
    {
      "epoch": 346.3185840707965,
      "grad_norm": 3.2337372886104276e-06,
      "learning_rate": 2.5512436411836538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11211472,
      "step": 19740
    },
    {
      "epoch": 346.4070796460177,
      "grad_norm": 1.9460130715742707e-06,
      "learning_rate": 2.5502620958136443e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11214624,
      "step": 19745
    },
    {
      "epoch": 346.49557522123894,
      "grad_norm": 1.7239875660379766e-06,
      "learning_rate": 2.5492805426925874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11217072,
      "step": 19750
    },
    {
      "epoch": 346.5840707964602,
      "grad_norm": 8.806562732388556e-07,
      "learning_rate": 2.5482989819718523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11220288,
      "step": 19755
    },
    {
      "epoch": 346.6725663716814,
      "grad_norm": 1.3245163472674903e-06,
      "learning_rate": 2.5473174138028065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11222928,
      "step": 19760
    },
    {
      "epoch": 346.76106194690266,
      "grad_norm": 1.9721755961654708e-06,
      "learning_rate": 2.5463358383368212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11225856,
      "step": 19765
    },
    {
      "epoch": 346.8495575221239,
      "grad_norm": 6.992505632297252e-07,
      "learning_rate": 2.545354255725267e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11228656,
      "step": 19770
    },
    {
      "epoch": 346.9380530973451,
      "grad_norm": 3.173230197717203e-06,
      "learning_rate": 2.5443726661195165e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11231744,
      "step": 19775
    },
    {
      "epoch": 347.01769911504425,
      "grad_norm": 6.694136800433625e-07,
      "learning_rate": 2.543391069670944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11233888,
      "step": 19780
    },
    {
      "epoch": 347.1061946902655,
      "grad_norm": 1.7160898551082937e-06,
      "learning_rate": 2.5424094665309228e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11236544,
      "step": 19785
    },
    {
      "epoch": 347.1946902654867,
      "grad_norm": 5.388770659919828e-06,
      "learning_rate": 2.5414278568508292e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11239392,
      "step": 19790
    },
    {
      "epoch": 347.283185840708,
      "grad_norm": 1.3803681895296904e-06,
      "learning_rate": 2.540446240782039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11242960,
      "step": 19795
    },
    {
      "epoch": 347.37168141592923,
      "grad_norm": 1.7063587165466743e-06,
      "learning_rate": 2.5394646184759307e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11245728,
      "step": 19800
    },
    {
      "epoch": 347.37168141592923,
      "eval_loss": 0.5383446216583252,
      "eval_runtime": 1.0776,
      "eval_samples_per_second": 23.2,
      "eval_steps_per_second": 12.064,
      "num_input_tokens_seen": 11245728,
      "step": 19800
    },
    {
      "epoch": 347.46017699115043,
      "grad_norm": 5.203442015044857e-07,
      "learning_rate": 2.538482990083882e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11248576,
      "step": 19805
    },
    {
      "epoch": 347.5486725663717,
      "grad_norm": 5.09767801304406e-07,
      "learning_rate": 2.5375013557572725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11251264,
      "step": 19810
    },
    {
      "epoch": 347.6371681415929,
      "grad_norm": 3.024164811904484e-07,
      "learning_rate": 2.536519715647483e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11254272,
      "step": 19815
    },
    {
      "epoch": 347.72566371681415,
      "grad_norm": 2.183238848374458e-06,
      "learning_rate": 2.535538069905894e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11257008,
      "step": 19820
    },
    {
      "epoch": 347.8141592920354,
      "grad_norm": 8.136073574860347e-07,
      "learning_rate": 2.534556418683888e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11259760,
      "step": 19825
    },
    {
      "epoch": 347.9026548672566,
      "grad_norm": 2.4728615244384855e-06,
      "learning_rate": 2.5335747621328486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11262688,
      "step": 19830
    },
    {
      "epoch": 347.9911504424779,
      "grad_norm": 2.9456168704200536e-06,
      "learning_rate": 2.5325931004041586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11265312,
      "step": 19835
    },
    {
      "epoch": 348.070796460177,
      "grad_norm": 2.4786061203485588e-06,
      "learning_rate": 2.5316114336492032e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11267912,
      "step": 19840
    },
    {
      "epoch": 348.1592920353982,
      "grad_norm": 6.428653591683542e-07,
      "learning_rate": 2.530629762019367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11270536,
      "step": 19845
    },
    {
      "epoch": 348.24778761061947,
      "grad_norm": 6.594017349925707e-07,
      "learning_rate": 2.5296480856660364e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11273512,
      "step": 19850
    },
    {
      "epoch": 348.3362831858407,
      "grad_norm": 1.6226107391048572e-06,
      "learning_rate": 2.528666404740599e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11276392,
      "step": 19855
    },
    {
      "epoch": 348.42477876106193,
      "grad_norm": 1.682402739788813e-06,
      "learning_rate": 2.527684719394442e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11279480,
      "step": 19860
    },
    {
      "epoch": 348.5132743362832,
      "grad_norm": 1.0302066584699787e-05,
      "learning_rate": 2.526703029778953e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11282088,
      "step": 19865
    },
    {
      "epoch": 348.60176991150445,
      "grad_norm": 6.165341801533941e-07,
      "learning_rate": 2.5257213360455208e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11285400,
      "step": 19870
    },
    {
      "epoch": 348.69026548672565,
      "grad_norm": 2.651434670042363e-06,
      "learning_rate": 2.5247396383455353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11288392,
      "step": 19875
    },
    {
      "epoch": 348.7787610619469,
      "grad_norm": 4.0164113102036936e-07,
      "learning_rate": 2.523757936830387e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11291192,
      "step": 19880
    },
    {
      "epoch": 348.86725663716817,
      "grad_norm": 2.5594346197976847e-07,
      "learning_rate": 2.5227762316514662e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11293768,
      "step": 19885
    },
    {
      "epoch": 348.95575221238937,
      "grad_norm": 3.6626730093303195e-07,
      "learning_rate": 2.5217945229601648e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11296952,
      "step": 19890
    },
    {
      "epoch": 349.0353982300885,
      "grad_norm": 1.5411843605761533e-06,
      "learning_rate": 2.5208128109078738e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11299280,
      "step": 19895
    },
    {
      "epoch": 349.12389380530976,
      "grad_norm": 1.8853459096135339e-06,
      "learning_rate": 2.5198310956459853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11302400,
      "step": 19900
    },
    {
      "epoch": 349.21238938053096,
      "grad_norm": 1.2609532404894708e-06,
      "learning_rate": 2.518849377325893e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11305104,
      "step": 19905
    },
    {
      "epoch": 349.3008849557522,
      "grad_norm": 3.2927439406194026e-06,
      "learning_rate": 2.51786765609899e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11308352,
      "step": 19910
    },
    {
      "epoch": 349.3893805309734,
      "grad_norm": 2.21771279029781e-06,
      "learning_rate": 2.5168859321166694e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11311328,
      "step": 19915
    },
    {
      "epoch": 349.4778761061947,
      "grad_norm": 1.1039982155125472e-06,
      "learning_rate": 2.515904205530326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11313552,
      "step": 19920
    },
    {
      "epoch": 349.56637168141594,
      "grad_norm": 3.4499564662837656e-06,
      "learning_rate": 2.514922476491355e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11316496,
      "step": 19925
    },
    {
      "epoch": 349.65486725663715,
      "grad_norm": 4.155424448981648e-06,
      "learning_rate": 2.51394074515115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11318912,
      "step": 19930
    },
    {
      "epoch": 349.7433628318584,
      "grad_norm": 9.646155376685783e-07,
      "learning_rate": 2.5129590116611067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11322416,
      "step": 19935
    },
    {
      "epoch": 349.83185840707966,
      "grad_norm": 7.153366254897264e-07,
      "learning_rate": 2.5119772761726212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11325248,
      "step": 19940
    },
    {
      "epoch": 349.92035398230087,
      "grad_norm": 1.5529126358160283e-06,
      "learning_rate": 2.5109955388370893e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11327872,
      "step": 19945
    },
    {
      "epoch": 350.0,
      "grad_norm": 2.3443328700523125e-07,
      "learning_rate": 2.510013799805907e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11330576,
      "step": 19950
    },
    {
      "epoch": 350.08849557522126,
      "grad_norm": 9.362446462546359e-07,
      "learning_rate": 2.5090320592304706e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11333232,
      "step": 19955
    },
    {
      "epoch": 350.17699115044246,
      "grad_norm": 5.315982889442239e-07,
      "learning_rate": 2.5080503172621777e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11335952,
      "step": 19960
    },
    {
      "epoch": 350.2654867256637,
      "grad_norm": 1.3886873375668074e-06,
      "learning_rate": 2.5070685740524246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11339136,
      "step": 19965
    },
    {
      "epoch": 350.353982300885,
      "grad_norm": 2.4811467937979614e-06,
      "learning_rate": 2.5060868297526084e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11341648,
      "step": 19970
    },
    {
      "epoch": 350.4424778761062,
      "grad_norm": 5.832299621033599e-07,
      "learning_rate": 2.5051050845141267e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11344272,
      "step": 19975
    },
    {
      "epoch": 350.53097345132744,
      "grad_norm": 5.135362926012021e-07,
      "learning_rate": 2.5041233384883765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11346896,
      "step": 19980
    },
    {
      "epoch": 350.6194690265487,
      "grad_norm": 6.361878490679373e-07,
      "learning_rate": 2.5031415918267564e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11349904,
      "step": 19985
    },
    {
      "epoch": 350.7079646017699,
      "grad_norm": 9.079390110855456e-07,
      "learning_rate": 2.5021598446806626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11352720,
      "step": 19990
    },
    {
      "epoch": 350.79646017699116,
      "grad_norm": 1.024523157866497e-06,
      "learning_rate": 2.5011780972014937e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11356320,
      "step": 19995
    },
    {
      "epoch": 350.88495575221236,
      "grad_norm": 3.5419333244135487e-07,
      "learning_rate": 2.5001963495406478e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11358800,
      "step": 20000
    },
    {
      "epoch": 350.88495575221236,
      "eval_loss": 0.538743257522583,
      "eval_runtime": 1.0665,
      "eval_samples_per_second": 23.441,
      "eval_steps_per_second": 12.189,
      "num_input_tokens_seen": 11358800,
      "step": 20000
    },
    {
      "epoch": 350.9734513274336,
      "grad_norm": 2.678608325368259e-06,
      "learning_rate": 2.499214601849522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11362048,
      "step": 20005
    },
    {
      "epoch": 351.05309734513276,
      "grad_norm": 2.9675040877918946e-06,
      "learning_rate": 2.4982328542795148e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11364464,
      "step": 20010
    },
    {
      "epoch": 351.14159292035396,
      "grad_norm": 1.0454443781782174e-06,
      "learning_rate": 2.497251106982024e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11367328,
      "step": 20015
    },
    {
      "epoch": 351.2300884955752,
      "grad_norm": 2.6994332529284293e-06,
      "learning_rate": 2.4962693601084458e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11370336,
      "step": 20020
    },
    {
      "epoch": 351.3185840707965,
      "grad_norm": 1.301671204601007e-06,
      "learning_rate": 2.4952876138101794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11373200,
      "step": 20025
    },
    {
      "epoch": 351.4070796460177,
      "grad_norm": 1.4166254231895437e-06,
      "learning_rate": 2.4943058682386233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11375888,
      "step": 20030
    },
    {
      "epoch": 351.49557522123894,
      "grad_norm": 2.8099179871787783e-06,
      "learning_rate": 2.493324123545173e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11379152,
      "step": 20035
    },
    {
      "epoch": 351.5840707964602,
      "grad_norm": 3.215212473151041e-06,
      "learning_rate": 2.4923423798812272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11382288,
      "step": 20040
    },
    {
      "epoch": 351.6725663716814,
      "grad_norm": 8.099685260276601e-07,
      "learning_rate": 2.4913606373981825e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11385344,
      "step": 20045
    },
    {
      "epoch": 351.76106194690266,
      "grad_norm": 4.593863650370622e-06,
      "learning_rate": 2.4903788962474357e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11387920,
      "step": 20050
    },
    {
      "epoch": 351.8495575221239,
      "grad_norm": 8.260145477834158e-06,
      "learning_rate": 2.489397156580385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11390768,
      "step": 20055
    },
    {
      "epoch": 351.9380530973451,
      "grad_norm": 1.5194431171039469e-06,
      "learning_rate": 2.4884154185484246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11393536,
      "step": 20060
    },
    {
      "epoch": 352.01769911504425,
      "grad_norm": 6.132103862910299e-06,
      "learning_rate": 2.4874336823029526e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11396296,
      "step": 20065
    },
    {
      "epoch": 352.1061946902655,
      "grad_norm": 3.3378696571162436e-06,
      "learning_rate": 2.4864519479953656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11398824,
      "step": 20070
    },
    {
      "epoch": 352.1946902654867,
      "grad_norm": 9.468926123190613e-07,
      "learning_rate": 2.485470215777058e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11401512,
      "step": 20075
    },
    {
      "epoch": 352.283185840708,
      "grad_norm": 2.7757530460803537e-06,
      "learning_rate": 2.4844884857994258e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11404296,
      "step": 20080
    },
    {
      "epoch": 352.37168141592923,
      "grad_norm": 9.178084496852534e-07,
      "learning_rate": 2.4835067582138638e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11407624,
      "step": 20085
    },
    {
      "epoch": 352.46017699115043,
      "grad_norm": 1.584528035891708e-06,
      "learning_rate": 2.4825250331717666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11410984,
      "step": 20090
    },
    {
      "epoch": 352.5486725663717,
      "grad_norm": 2.9159639325371245e-06,
      "learning_rate": 2.4815433108245298e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11413432,
      "step": 20095
    },
    {
      "epoch": 352.6371681415929,
      "grad_norm": 1.2044347386108711e-06,
      "learning_rate": 2.4805615913235456e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11416440,
      "step": 20100
    },
    {
      "epoch": 352.72566371681415,
      "grad_norm": 2.2142057787277736e-06,
      "learning_rate": 2.479579874820208e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11419496,
      "step": 20105
    },
    {
      "epoch": 352.8141592920354,
      "grad_norm": 1.9192564195691375e-06,
      "learning_rate": 2.4785981614659115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11422424,
      "step": 20110
    },
    {
      "epoch": 352.9026548672566,
      "grad_norm": 5.964166120975278e-07,
      "learning_rate": 2.477616451412047e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11425432,
      "step": 20115
    },
    {
      "epoch": 352.9911504424779,
      "grad_norm": 5.013613986193377e-07,
      "learning_rate": 2.476634744810007e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11427976,
      "step": 20120
    },
    {
      "epoch": 353.070796460177,
      "grad_norm": 2.1273569927870994e-06,
      "learning_rate": 2.475653041811183e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11430200,
      "step": 20125
    },
    {
      "epoch": 353.1592920353982,
      "grad_norm": 5.6143771871575154e-06,
      "learning_rate": 2.4746713425669652e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11432936,
      "step": 20130
    },
    {
      "epoch": 353.24778761061947,
      "grad_norm": 1.249696538252465e-06,
      "learning_rate": 2.4736896472287458e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11435432,
      "step": 20135
    },
    {
      "epoch": 353.3362831858407,
      "grad_norm": 5.561143439081206e-07,
      "learning_rate": 2.4727079559479124e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11438088,
      "step": 20140
    },
    {
      "epoch": 353.42477876106193,
      "grad_norm": 5.324928338268364e-07,
      "learning_rate": 2.4717262688758557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11441128,
      "step": 20145
    },
    {
      "epoch": 353.5132743362832,
      "grad_norm": 3.5897127759199066e-07,
      "learning_rate": 2.4707445861639637e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11444248,
      "step": 20150
    },
    {
      "epoch": 353.60176991150445,
      "grad_norm": 6.58275723708357e-07,
      "learning_rate": 2.4697629079636244e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11447304,
      "step": 20155
    },
    {
      "epoch": 353.69026548672565,
      "grad_norm": 6.758768904546741e-06,
      "learning_rate": 2.4687812344262244e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11450680,
      "step": 20160
    },
    {
      "epoch": 353.7787610619469,
      "grad_norm": 2.7766479888668982e-06,
      "learning_rate": 2.46779956570315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11453080,
      "step": 20165
    },
    {
      "epoch": 353.86725663716817,
      "grad_norm": 6.602913344977424e-07,
      "learning_rate": 2.466817901945787e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11456312,
      "step": 20170
    },
    {
      "epoch": 353.95575221238937,
      "grad_norm": 1.0066027016364387e-06,
      "learning_rate": 2.4658362433055217e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11459368,
      "step": 20175
    },
    {
      "epoch": 354.0353982300885,
      "grad_norm": 1.7060167465388076e-06,
      "learning_rate": 2.4648545899337356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11461496,
      "step": 20180
    },
    {
      "epoch": 354.12389380530976,
      "grad_norm": 2.2434267066273605e-06,
      "learning_rate": 2.4638729419818143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11464104,
      "step": 20185
    },
    {
      "epoch": 354.21238938053096,
      "grad_norm": 1.0083840606966987e-06,
      "learning_rate": 2.46289129960114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11466936,
      "step": 20190
    },
    {
      "epoch": 354.3008849557522,
      "grad_norm": 4.853241080127191e-06,
      "learning_rate": 2.4619096629430924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11469336,
      "step": 20195
    },
    {
      "epoch": 354.3893805309734,
      "grad_norm": 5.050304935139138e-06,
      "learning_rate": 2.4609280321590543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11471832,
      "step": 20200
    },
    {
      "epoch": 354.3893805309734,
      "eval_loss": 0.5425193905830383,
      "eval_runtime": 1.0603,
      "eval_samples_per_second": 23.578,
      "eval_steps_per_second": 12.261,
      "num_input_tokens_seen": 11471832,
      "step": 20200
    },
    {
      "epoch": 354.4778761061947,
      "grad_norm": 2.805689291562885e-06,
      "learning_rate": 2.4599464074004037e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11475144,
      "step": 20205
    },
    {
      "epoch": 354.56637168141594,
      "grad_norm": 6.458562893385533e-06,
      "learning_rate": 2.4589647888185204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11478120,
      "step": 20210
    },
    {
      "epoch": 354.65486725663715,
      "grad_norm": 8.97220900242246e-07,
      "learning_rate": 2.4579831765647836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11481352,
      "step": 20215
    },
    {
      "epoch": 354.7433628318584,
      "grad_norm": 6.744413099113444e-07,
      "learning_rate": 2.4570015707905676e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11484456,
      "step": 20220
    },
    {
      "epoch": 354.83185840707966,
      "grad_norm": 6.626320896430116e-07,
      "learning_rate": 2.4560199716472508e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11487720,
      "step": 20225
    },
    {
      "epoch": 354.92035398230087,
      "grad_norm": 7.470493983419146e-07,
      "learning_rate": 2.455038379286207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11490536,
      "step": 20230
    },
    {
      "epoch": 355.0,
      "grad_norm": 6.768155458303227e-07,
      "learning_rate": 2.4540567938588095e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11492832,
      "step": 20235
    },
    {
      "epoch": 355.08849557522126,
      "grad_norm": 5.658997679347522e-07,
      "learning_rate": 2.4530752155164328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11496256,
      "step": 20240
    },
    {
      "epoch": 355.17699115044246,
      "grad_norm": 1.6697179034963483e-06,
      "learning_rate": 2.4520936444104463e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11499040,
      "step": 20245
    },
    {
      "epoch": 355.2654867256637,
      "grad_norm": 1.147346779362124e-06,
      "learning_rate": 2.4511120806922218e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11501456,
      "step": 20250
    },
    {
      "epoch": 355.353982300885,
      "grad_norm": 3.7650945614586817e-06,
      "learning_rate": 2.45013052451313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11504528,
      "step": 20255
    },
    {
      "epoch": 355.4424778761062,
      "grad_norm": 7.24539688690129e-07,
      "learning_rate": 2.4491489760245376e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11507200,
      "step": 20260
    },
    {
      "epoch": 355.53097345132744,
      "grad_norm": 6.648689918620221e-07,
      "learning_rate": 2.4481674353778115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11510096,
      "step": 20265
    },
    {
      "epoch": 355.6194690265487,
      "grad_norm": 1.6325691376550822e-06,
      "learning_rate": 2.447185902724319e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11513056,
      "step": 20270
    },
    {
      "epoch": 355.7079646017699,
      "grad_norm": 3.818691993728862e-07,
      "learning_rate": 2.4462043782154233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11515824,
      "step": 20275
    },
    {
      "epoch": 355.79646017699116,
      "grad_norm": 1.0364638001192361e-06,
      "learning_rate": 2.4452228620024895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11519072,
      "step": 20280
    },
    {
      "epoch": 355.88495575221236,
      "grad_norm": 6.946660050743958e-06,
      "learning_rate": 2.4442413542368776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11521392,
      "step": 20285
    },
    {
      "epoch": 355.9734513274336,
      "grad_norm": 1.1136124840049888e-06,
      "learning_rate": 2.4432598550699502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11524080,
      "step": 20290
    },
    {
      "epoch": 356.05309734513276,
      "grad_norm": 7.460705546691315e-07,
      "learning_rate": 2.4422783646530663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11527064,
      "step": 20295
    },
    {
      "epoch": 356.14159292035396,
      "grad_norm": 1.4560979479938396e-06,
      "learning_rate": 2.441296883137584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11529688,
      "step": 20300
    },
    {
      "epoch": 356.2300884955752,
      "grad_norm": 6.409367188098258e-07,
      "learning_rate": 2.4403154106748592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11532456,
      "step": 20305
    },
    {
      "epoch": 356.3185840707965,
      "grad_norm": 4.661336788558401e-06,
      "learning_rate": 2.4393339474162494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11535272,
      "step": 20310
    },
    {
      "epoch": 356.4070796460177,
      "grad_norm": 8.387846719415393e-07,
      "learning_rate": 2.4383524935131062e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11537848,
      "step": 20315
    },
    {
      "epoch": 356.49557522123894,
      "grad_norm": 1.508272589489934e-06,
      "learning_rate": 2.437371049116784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11541256,
      "step": 20320
    },
    {
      "epoch": 356.5840707964602,
      "grad_norm": 1.1326661706334562e-06,
      "learning_rate": 2.436389614378632e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11544008,
      "step": 20325
    },
    {
      "epoch": 356.6725663716814,
      "grad_norm": 1.7165688177556149e-06,
      "learning_rate": 2.435408189450002e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11546904,
      "step": 20330
    },
    {
      "epoch": 356.76106194690266,
      "grad_norm": 3.4980262171302456e-06,
      "learning_rate": 2.4344267744822406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11550104,
      "step": 20335
    },
    {
      "epoch": 356.8495575221239,
      "grad_norm": 1.4051653124624863e-05,
      "learning_rate": 2.4334453696266944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11553176,
      "step": 20340
    },
    {
      "epoch": 356.9380530973451,
      "grad_norm": 8.9001991909754e-07,
      "learning_rate": 2.432463975034708e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11556024,
      "step": 20345
    },
    {
      "epoch": 357.01769911504425,
      "grad_norm": 9.720565685711335e-07,
      "learning_rate": 2.4314825908576265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11558240,
      "step": 20350
    },
    {
      "epoch": 357.1061946902655,
      "grad_norm": 4.028946420930879e-07,
      "learning_rate": 2.4305012172467897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11561456,
      "step": 20355
    },
    {
      "epoch": 357.1946902654867,
      "grad_norm": 1.1198294487257954e-06,
      "learning_rate": 2.4295198543535393e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11563968,
      "step": 20360
    },
    {
      "epoch": 357.283185840708,
      "grad_norm": 6.222470005923242e-07,
      "learning_rate": 2.4285385023292124e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11566320,
      "step": 20365
    },
    {
      "epoch": 357.37168141592923,
      "grad_norm": 7.878855399212625e-07,
      "learning_rate": 2.427557161325147e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11569376,
      "step": 20370
    },
    {
      "epoch": 357.46017699115043,
      "grad_norm": 1.7035436030710116e-06,
      "learning_rate": 2.4265758314926778e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11572320,
      "step": 20375
    },
    {
      "epoch": 357.5486725663717,
      "grad_norm": 1.2754277349813492e-06,
      "learning_rate": 2.4255945129831373e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11575168,
      "step": 20380
    },
    {
      "epoch": 357.6371681415929,
      "grad_norm": 1.7574886896909447e-06,
      "learning_rate": 2.4246132059478578e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11577792,
      "step": 20385
    },
    {
      "epoch": 357.72566371681415,
      "grad_norm": 8.397619808420131e-07,
      "learning_rate": 2.4236319105381706e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11580224,
      "step": 20390
    },
    {
      "epoch": 357.8141592920354,
      "grad_norm": 1.1624396165643702e-06,
      "learning_rate": 2.422650626905401e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11583024,
      "step": 20395
    },
    {
      "epoch": 357.9026548672566,
      "grad_norm": 2.5798358365136664e-06,
      "learning_rate": 2.4216693552008785e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11586368,
      "step": 20400
    },
    {
      "epoch": 357.9026548672566,
      "eval_loss": 0.5417349338531494,
      "eval_runtime": 1.0651,
      "eval_samples_per_second": 23.471,
      "eval_steps_per_second": 12.205,
      "num_input_tokens_seen": 11586368,
      "step": 20400
    },
    {
      "epoch": 357.9911504424779,
      "grad_norm": 3.707933728946955e-06,
      "learning_rate": 2.4206880955759247e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11589504,
      "step": 20405
    },
    {
      "epoch": 358.070796460177,
      "grad_norm": 6.721905378981319e-07,
      "learning_rate": 2.419706848181863e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11592016,
      "step": 20410
    },
    {
      "epoch": 358.1592920353982,
      "grad_norm": 9.717160764921573e-07,
      "learning_rate": 2.4187256131700153e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11595216,
      "step": 20415
    },
    {
      "epoch": 358.24778761061947,
      "grad_norm": 2.9644947971974034e-06,
      "learning_rate": 2.4177443906916985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11597664,
      "step": 20420
    },
    {
      "epoch": 358.3362831858407,
      "grad_norm": 8.381256293432671e-07,
      "learning_rate": 2.4167631808982303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11600880,
      "step": 20425
    },
    {
      "epoch": 358.42477876106193,
      "grad_norm": 1.633994884286949e-06,
      "learning_rate": 2.4157819839409264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11603632,
      "step": 20430
    },
    {
      "epoch": 358.5132743362832,
      "grad_norm": 9.847376531979535e-07,
      "learning_rate": 2.414800799971098e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11606288,
      "step": 20435
    },
    {
      "epoch": 358.60176991150445,
      "grad_norm": 8.180912232091941e-07,
      "learning_rate": 2.4138196291400582e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11609392,
      "step": 20440
    },
    {
      "epoch": 358.69026548672565,
      "grad_norm": 1.9656404219858814e-06,
      "learning_rate": 2.412838471599114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11612576,
      "step": 20445
    },
    {
      "epoch": 358.7787610619469,
      "grad_norm": 2.422310672045569e-06,
      "learning_rate": 2.411857327499572e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11615040,
      "step": 20450
    },
    {
      "epoch": 358.86725663716817,
      "grad_norm": 8.448993753518153e-07,
      "learning_rate": 2.410876196992739e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11618288,
      "step": 20455
    },
    {
      "epoch": 358.95575221238937,
      "grad_norm": 7.363244662883517e-07,
      "learning_rate": 2.4098950802299156e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11621008,
      "step": 20460
    },
    {
      "epoch": 359.0353982300885,
      "grad_norm": 8.86401664956793e-07,
      "learning_rate": 2.4089139773624027e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11623472,
      "step": 20465
    },
    {
      "epoch": 359.12389380530976,
      "grad_norm": 8.393874395551393e-07,
      "learning_rate": 2.4079328885415007e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11626176,
      "step": 20470
    },
    {
      "epoch": 359.21238938053096,
      "grad_norm": 9.166756740341953e-07,
      "learning_rate": 2.4069518139185036e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11629216,
      "step": 20475
    },
    {
      "epoch": 359.3008849557522,
      "grad_norm": 1.2809917961931205e-06,
      "learning_rate": 2.405970753644706e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11632352,
      "step": 20480
    },
    {
      "epoch": 359.3893805309734,
      "grad_norm": 1.1682274134727777e-06,
      "learning_rate": 2.4049897078714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11635024,
      "step": 20485
    },
    {
      "epoch": 359.4778761061947,
      "grad_norm": 9.925541917255032e-07,
      "learning_rate": 2.404008676749874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11637376,
      "step": 20490
    },
    {
      "epoch": 359.56637168141594,
      "grad_norm": 4.568182134789822e-07,
      "learning_rate": 2.403027660431418e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11640704,
      "step": 20495
    },
    {
      "epoch": 359.65486725663715,
      "grad_norm": 7.878991823417891e-07,
      "learning_rate": 2.402046659067314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11643760,
      "step": 20500
    },
    {
      "epoch": 359.7433628318584,
      "grad_norm": 1.8191518620369607e-06,
      "learning_rate": 2.401065672808847e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11646656,
      "step": 20505
    },
    {
      "epoch": 359.83185840707966,
      "grad_norm": 2.875537575164344e-06,
      "learning_rate": 2.400084701807296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11649552,
      "step": 20510
    },
    {
      "epoch": 359.92035398230087,
      "grad_norm": 2.825664068950573e-06,
      "learning_rate": 2.39910374621394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11652464,
      "step": 20515
    },
    {
      "epoch": 360.0,
      "grad_norm": 1.1374106634320924e-06,
      "learning_rate": 2.3981228061800544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11655312,
      "step": 20520
    },
    {
      "epoch": 360.08849557522126,
      "grad_norm": 1.7682349380265805e-06,
      "learning_rate": 2.3971418818569115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11657760,
      "step": 20525
    },
    {
      "epoch": 360.17699115044246,
      "grad_norm": 9.361328352497367e-07,
      "learning_rate": 2.3961609733957832e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11660656,
      "step": 20530
    },
    {
      "epoch": 360.2654867256637,
      "grad_norm": 1.6079492297649267e-06,
      "learning_rate": 2.395180080947939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11663344,
      "step": 20535
    },
    {
      "epoch": 360.353982300885,
      "grad_norm": 1.366719402540184e-06,
      "learning_rate": 2.394199204664642e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11666432,
      "step": 20540
    },
    {
      "epoch": 360.4424778761062,
      "grad_norm": 6.053570587027934e-07,
      "learning_rate": 2.3932183446971583e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11669008,
      "step": 20545
    },
    {
      "epoch": 360.53097345132744,
      "grad_norm": 5.95162964600604e-06,
      "learning_rate": 2.3922375011967473e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11672096,
      "step": 20550
    },
    {
      "epoch": 360.6194690265487,
      "grad_norm": 2.8273832867853343e-05,
      "learning_rate": 2.3912566743146676e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11675552,
      "step": 20555
    },
    {
      "epoch": 360.7079646017699,
      "grad_norm": 6.411310550902272e-06,
      "learning_rate": 2.390275864202176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11678272,
      "step": 20560
    },
    {
      "epoch": 360.79646017699116,
      "grad_norm": 9.998345831263578e-07,
      "learning_rate": 2.3892950710105243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11681472,
      "step": 20565
    },
    {
      "epoch": 360.88495575221236,
      "grad_norm": 8.61824446474202e-06,
      "learning_rate": 2.3883142948909635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11684272,
      "step": 20570
    },
    {
      "epoch": 360.9734513274336,
      "grad_norm": 1.093520836548123e-06,
      "learning_rate": 2.3873335359947433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11687008,
      "step": 20575
    },
    {
      "epoch": 361.05309734513276,
      "grad_norm": 1.0593162187433336e-05,
      "learning_rate": 2.3863527944731066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11688944,
      "step": 20580
    },
    {
      "epoch": 361.14159292035396,
      "grad_norm": 3.829300965207949e-07,
      "learning_rate": 2.385372070477298e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11692240,
      "step": 20585
    },
    {
      "epoch": 361.2300884955752,
      "grad_norm": 5.899823349864164e-07,
      "learning_rate": 2.384391364158556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11694816,
      "step": 20590
    },
    {
      "epoch": 361.3185840707965,
      "grad_norm": 1.983256936455291e-07,
      "learning_rate": 2.3834106756681185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11697648,
      "step": 20595
    },
    {
      "epoch": 361.4070796460177,
      "grad_norm": 3.009957254107576e-06,
      "learning_rate": 2.3824300051572206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11700176,
      "step": 20600
    },
    {
      "epoch": 361.4070796460177,
      "eval_loss": 0.56800377368927,
      "eval_runtime": 1.0606,
      "eval_samples_per_second": 23.571,
      "eval_steps_per_second": 12.257,
      "num_input_tokens_seen": 11700176,
      "step": 20600
    },
    {
      "epoch": 361.49557522123894,
      "grad_norm": 3.0015709739927843e-07,
      "learning_rate": 2.3814493527770923e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11703296,
      "step": 20605
    },
    {
      "epoch": 361.5840707964602,
      "grad_norm": 2.8256181394681334e-07,
      "learning_rate": 2.3804687186789637e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11705904,
      "step": 20610
    },
    {
      "epoch": 361.6725663716814,
      "grad_norm": 5.687335260518012e-07,
      "learning_rate": 2.379488103014062e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11708496,
      "step": 20615
    },
    {
      "epoch": 361.76106194690266,
      "grad_norm": 8.943161446950398e-07,
      "learning_rate": 2.3785075059336086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11711728,
      "step": 20620
    },
    {
      "epoch": 361.8495575221239,
      "grad_norm": 5.862788157173782e-07,
      "learning_rate": 2.3775269275888248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11714848,
      "step": 20625
    },
    {
      "epoch": 361.9380530973451,
      "grad_norm": 4.5816696570000204e-07,
      "learning_rate": 2.3765463681309274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11718352,
      "step": 20630
    },
    {
      "epoch": 362.01769911504425,
      "grad_norm": 5.951436605755589e-07,
      "learning_rate": 2.3755658277111313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11720688,
      "step": 20635
    },
    {
      "epoch": 362.1061946902655,
      "grad_norm": 9.657509281169041e-07,
      "learning_rate": 2.374585306480649e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11723040,
      "step": 20640
    },
    {
      "epoch": 362.1946902654867,
      "grad_norm": 3.5728319858208124e-07,
      "learning_rate": 2.3736048045906877e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11725872,
      "step": 20645
    },
    {
      "epoch": 362.283185840708,
      "grad_norm": 6.190558110574784e-07,
      "learning_rate": 2.372624322192454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11728960,
      "step": 20650
    },
    {
      "epoch": 362.37168141592923,
      "grad_norm": 1.9863473426084965e-06,
      "learning_rate": 2.3716438594371516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11732144,
      "step": 20655
    },
    {
      "epoch": 362.46017699115043,
      "grad_norm": 9.059212970896624e-06,
      "learning_rate": 2.3706634164759784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11735424,
      "step": 20660
    },
    {
      "epoch": 362.5486725663717,
      "grad_norm": 3.144381253150641e-06,
      "learning_rate": 2.3696829934601323e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11738144,
      "step": 20665
    },
    {
      "epoch": 362.6371681415929,
      "grad_norm": 2.543997652537655e-06,
      "learning_rate": 2.3687025905408053e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11740528,
      "step": 20670
    },
    {
      "epoch": 362.72566371681415,
      "grad_norm": 2.6541767965682084e-06,
      "learning_rate": 2.3677222078691886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11743344,
      "step": 20675
    },
    {
      "epoch": 362.8141592920354,
      "grad_norm": 2.3421926016453654e-06,
      "learning_rate": 2.366741845596471e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11746624,
      "step": 20680
    },
    {
      "epoch": 362.9026548672566,
      "grad_norm": 9.924008281814167e-07,
      "learning_rate": 2.3657615038738343e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11749920,
      "step": 20685
    },
    {
      "epoch": 362.9911504424779,
      "grad_norm": 9.365300002173171e-07,
      "learning_rate": 2.3647811828524614e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11752432,
      "step": 20690
    },
    {
      "epoch": 363.070796460177,
      "grad_norm": 1.3623229051518138e-06,
      "learning_rate": 2.363800882683529e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11754384,
      "step": 20695
    },
    {
      "epoch": 363.1592920353982,
      "grad_norm": 1.1500287655508146e-06,
      "learning_rate": 2.3628206035182125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11756912,
      "step": 20700
    },
    {
      "epoch": 363.24778761061947,
      "grad_norm": 1.4711190488014836e-06,
      "learning_rate": 2.361840345507683e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11760048,
      "step": 20705
    },
    {
      "epoch": 363.3362831858407,
      "grad_norm": 1.2722005067189457e-06,
      "learning_rate": 2.3608601088031073e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11762960,
      "step": 20710
    },
    {
      "epoch": 363.42477876106193,
      "grad_norm": 1.5481889477086952e-06,
      "learning_rate": 2.3598798935556516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11766320,
      "step": 20715
    },
    {
      "epoch": 363.5132743362832,
      "grad_norm": 7.249523719110584e-07,
      "learning_rate": 2.3588996999164784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11769264,
      "step": 20720
    },
    {
      "epoch": 363.60176991150445,
      "grad_norm": 5.376574563342729e-07,
      "learning_rate": 2.3579195280367434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11771776,
      "step": 20725
    },
    {
      "epoch": 363.69026548672565,
      "grad_norm": 1.5328887457144447e-06,
      "learning_rate": 2.356939378067603e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11774432,
      "step": 20730
    },
    {
      "epoch": 363.7787610619469,
      "grad_norm": 1.2877717381343246e-06,
      "learning_rate": 2.3559592501602092e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11778000,
      "step": 20735
    },
    {
      "epoch": 363.86725663716817,
      "grad_norm": 2.4485946141794557e-06,
      "learning_rate": 2.3549791444657076e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11781072,
      "step": 20740
    },
    {
      "epoch": 363.95575221238937,
      "grad_norm": 5.68082214158494e-06,
      "learning_rate": 2.353999061135246e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11783888,
      "step": 20745
    },
    {
      "epoch": 364.0353982300885,
      "grad_norm": 1.1282812693025335e-06,
      "learning_rate": 2.3530190003199626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11786048,
      "step": 20750
    },
    {
      "epoch": 364.12389380530976,
      "grad_norm": 2.7978780963167083e-06,
      "learning_rate": 2.3520389621709965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11788784,
      "step": 20755
    },
    {
      "epoch": 364.21238938053096,
      "grad_norm": 6.613262939936249e-06,
      "learning_rate": 2.351058946839483e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11791712,
      "step": 20760
    },
    {
      "epoch": 364.3008849557522,
      "grad_norm": 8.267521707239212e-07,
      "learning_rate": 2.350078954476551e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11794320,
      "step": 20765
    },
    {
      "epoch": 364.3893805309734,
      "grad_norm": 4.764428013004363e-06,
      "learning_rate": 2.3490989852333272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11796768,
      "step": 20770
    },
    {
      "epoch": 364.4778761061947,
      "grad_norm": 1.4314637155621313e-06,
      "learning_rate": 2.3481190392609377e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11799904,
      "step": 20775
    },
    {
      "epoch": 364.56637168141594,
      "grad_norm": 3.0497512852889486e-06,
      "learning_rate": 2.3471391167105e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11802240,
      "step": 20780
    },
    {
      "epoch": 364.65486725663715,
      "grad_norm": 1.004252794700733e-06,
      "learning_rate": 2.3461592177331325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11805440,
      "step": 20785
    },
    {
      "epoch": 364.7433628318584,
      "grad_norm": 5.334029538062168e-07,
      "learning_rate": 2.345179342479946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11808352,
      "step": 20790
    },
    {
      "epoch": 364.83185840707966,
      "grad_norm": 2.0552906789816916e-06,
      "learning_rate": 2.3441994911020503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11811280,
      "step": 20795
    },
    {
      "epoch": 364.92035398230087,
      "grad_norm": 5.406017749010061e-07,
      "learning_rate": 2.3432196637505522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11814304,
      "step": 20800
    },
    {
      "epoch": 364.92035398230087,
      "eval_loss": 0.5215240716934204,
      "eval_runtime": 1.0594,
      "eval_samples_per_second": 23.598,
      "eval_steps_per_second": 12.271,
      "num_input_tokens_seen": 11814304,
      "step": 20800
    },
    {
      "epoch": 365.0,
      "grad_norm": 1.158676923296298e-06,
      "learning_rate": 2.3422398605765515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11817184,
      "step": 20805
    },
    {
      "epoch": 365.08849557522126,
      "grad_norm": 9.762577519722981e-07,
      "learning_rate": 2.3412600817311462e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11819888,
      "step": 20810
    },
    {
      "epoch": 365.17699115044246,
      "grad_norm": 8.590707807343279e-07,
      "learning_rate": 2.3402803273654326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11822624,
      "step": 20815
    },
    {
      "epoch": 365.2654867256637,
      "grad_norm": 2.68032226813375e-06,
      "learning_rate": 2.3393005976304983e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11825344,
      "step": 20820
    },
    {
      "epoch": 365.353982300885,
      "grad_norm": 3.860212018480524e-06,
      "learning_rate": 2.338320892677432e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11828384,
      "step": 20825
    },
    {
      "epoch": 365.4424778761062,
      "grad_norm": 3.9661922528466675e-06,
      "learning_rate": 2.3373412126573155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11831248,
      "step": 20830
    },
    {
      "epoch": 365.53097345132744,
      "grad_norm": 8.358947525266558e-07,
      "learning_rate": 2.3363615577212285e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11834224,
      "step": 20835
    },
    {
      "epoch": 365.6194690265487,
      "grad_norm": 3.501004357531201e-06,
      "learning_rate": 2.3353819280202455e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11837504,
      "step": 20840
    },
    {
      "epoch": 365.7079646017699,
      "grad_norm": 5.624727691611042e-06,
      "learning_rate": 2.334402323705438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11840288,
      "step": 20845
    },
    {
      "epoch": 365.79646017699116,
      "grad_norm": 8.977216907624097e-07,
      "learning_rate": 2.3334227449278725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11842832,
      "step": 20850
    },
    {
      "epoch": 365.88495575221236,
      "grad_norm": 7.927433216536883e-07,
      "learning_rate": 2.3324431918386143e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11846192,
      "step": 20855
    },
    {
      "epoch": 365.9734513274336,
      "grad_norm": 6.814393032072985e-07,
      "learning_rate": 2.3314636645887207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11849232,
      "step": 20860
    },
    {
      "epoch": 366.05309734513276,
      "grad_norm": 1.3933648688180256e-06,
      "learning_rate": 2.3304841633292487e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11851592,
      "step": 20865
    },
    {
      "epoch": 366.14159292035396,
      "grad_norm": 7.124255603230267e-07,
      "learning_rate": 2.329504688211248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11854664,
      "step": 20870
    },
    {
      "epoch": 366.2300884955752,
      "grad_norm": 7.690543952776352e-07,
      "learning_rate": 2.3285252393857677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11857544,
      "step": 20875
    },
    {
      "epoch": 366.3185840707965,
      "grad_norm": 1.3661409639098565e-06,
      "learning_rate": 2.327545817003851e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11860536,
      "step": 20880
    },
    {
      "epoch": 366.4070796460177,
      "grad_norm": 1.1695303783199051e-06,
      "learning_rate": 2.326566421216535e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11863272,
      "step": 20885
    },
    {
      "epoch": 366.49557522123894,
      "grad_norm": 3.058595893890015e-06,
      "learning_rate": 2.3255870521748565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11865944,
      "step": 20890
    },
    {
      "epoch": 366.5840707964602,
      "grad_norm": 4.816693603970634e-07,
      "learning_rate": 2.3246077100298474e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11868776,
      "step": 20895
    },
    {
      "epoch": 366.6725663716814,
      "grad_norm": 5.143189127920778e-07,
      "learning_rate": 2.3236283949325328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11871816,
      "step": 20900
    },
    {
      "epoch": 366.76106194690266,
      "grad_norm": 9.939622032106854e-07,
      "learning_rate": 2.3226491070339368e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11874520,
      "step": 20905
    },
    {
      "epoch": 366.8495575221239,
      "grad_norm": 5.737763331126189e-06,
      "learning_rate": 2.3216698464850762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11877288,
      "step": 20910
    },
    {
      "epoch": 366.9380530973451,
      "grad_norm": 1.686328573669016e-06,
      "learning_rate": 2.320690613436967e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11880328,
      "step": 20915
    },
    {
      "epoch": 367.01769911504425,
      "grad_norm": 4.282549639356148e-07,
      "learning_rate": 2.3197114080406192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11882952,
      "step": 20920
    },
    {
      "epoch": 367.1061946902655,
      "grad_norm": 2.6216014248348074e-06,
      "learning_rate": 2.3187322304470365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11885848,
      "step": 20925
    },
    {
      "epoch": 367.1946902654867,
      "grad_norm": 9.316580076301761e-07,
      "learning_rate": 2.3177530808072222e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11888424,
      "step": 20930
    },
    {
      "epoch": 367.283185840708,
      "grad_norm": 1.9269234599050833e-06,
      "learning_rate": 2.316773959272174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11891224,
      "step": 20935
    },
    {
      "epoch": 367.37168141592923,
      "grad_norm": 1.2506421853686334e-06,
      "learning_rate": 2.3157948659928823e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11894264,
      "step": 20940
    },
    {
      "epoch": 367.46017699115043,
      "grad_norm": 4.222601546644e-06,
      "learning_rate": 2.3148158011203388e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11896984,
      "step": 20945
    },
    {
      "epoch": 367.5486725663717,
      "grad_norm": 1.0799520850923727e-06,
      "learning_rate": 2.3138367648055253e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11900184,
      "step": 20950
    },
    {
      "epoch": 367.6371681415929,
      "grad_norm": 2.743264076343621e-06,
      "learning_rate": 2.312857757199422e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11903464,
      "step": 20955
    },
    {
      "epoch": 367.72566371681415,
      "grad_norm": 3.6000963063997915e-06,
      "learning_rate": 2.3118787784530048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11906056,
      "step": 20960
    },
    {
      "epoch": 367.8141592920354,
      "grad_norm": 1.1683303000609158e-06,
      "learning_rate": 2.310899828717243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11908888,
      "step": 20965
    },
    {
      "epoch": 367.9026548672566,
      "grad_norm": 1.0949950137728592e-06,
      "learning_rate": 2.309920908143104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11911704,
      "step": 20970
    },
    {
      "epoch": 367.9911504424779,
      "grad_norm": 2.7327350835548714e-06,
      "learning_rate": 2.308942016881551e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11914248,
      "step": 20975
    },
    {
      "epoch": 368.070796460177,
      "grad_norm": 2.74321723736648e-06,
      "learning_rate": 2.307963155083539e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11916344,
      "step": 20980
    },
    {
      "epoch": 368.1592920353982,
      "grad_norm": 6.323886623249564e-07,
      "learning_rate": 2.306984322900022e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11919304,
      "step": 20985
    },
    {
      "epoch": 368.24778761061947,
      "grad_norm": 6.972636583668645e-07,
      "learning_rate": 2.3060055204819482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11922376,
      "step": 20990
    },
    {
      "epoch": 368.3362831858407,
      "grad_norm": 8.178395205504785e-07,
      "learning_rate": 2.3050267479802604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11924888,
      "step": 20995
    },
    {
      "epoch": 368.42477876106193,
      "grad_norm": 4.305527170345158e-07,
      "learning_rate": 2.304048005545899e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11927464,
      "step": 21000
    },
    {
      "epoch": 368.42477876106193,
      "eval_loss": 0.5594981908798218,
      "eval_runtime": 1.0653,
      "eval_samples_per_second": 23.467,
      "eval_steps_per_second": 12.203,
      "num_input_tokens_seen": 11927464,
      "step": 21000
    },
    {
      "epoch": 368.5132743362832,
      "grad_norm": 1.918068164741271e-06,
      "learning_rate": 2.3030692933297972e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11930888,
      "step": 21005
    },
    {
      "epoch": 368.60176991150445,
      "grad_norm": 6.600645292564877e-07,
      "learning_rate": 2.3020906114828843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11933976,
      "step": 21010
    },
    {
      "epoch": 368.69026548672565,
      "grad_norm": 5.599097221420379e-07,
      "learning_rate": 2.301111960156088e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11936936,
      "step": 21015
    },
    {
      "epoch": 368.7787610619469,
      "grad_norm": 1.0419703357911203e-06,
      "learning_rate": 2.300133339500326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11939688,
      "step": 21020
    },
    {
      "epoch": 368.86725663716817,
      "grad_norm": 2.2283534235612024e-06,
      "learning_rate": 2.2991547496665148e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11942616,
      "step": 21025
    },
    {
      "epoch": 368.95575221238937,
      "grad_norm": 3.29319664160721e-06,
      "learning_rate": 2.298176190805565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11945304,
      "step": 21030
    },
    {
      "epoch": 369.0353982300885,
      "grad_norm": 5.286812665872276e-06,
      "learning_rate": 2.2971976630683826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11947520,
      "step": 21035
    },
    {
      "epoch": 369.12389380530976,
      "grad_norm": 3.1186337423605437e-07,
      "learning_rate": 2.29621916660587e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11951280,
      "step": 21040
    },
    {
      "epoch": 369.21238938053096,
      "grad_norm": 1.3880428468837636e-06,
      "learning_rate": 2.295240701568922e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11954384,
      "step": 21045
    },
    {
      "epoch": 369.3008849557522,
      "grad_norm": 6.055321136955172e-06,
      "learning_rate": 2.2942622681084312e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11956880,
      "step": 21050
    },
    {
      "epoch": 369.3893805309734,
      "grad_norm": 1.4630369378210162e-06,
      "learning_rate": 2.293283866375284e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11959744,
      "step": 21055
    },
    {
      "epoch": 369.4778761061947,
      "grad_norm": 2.5406761778867804e-06,
      "learning_rate": 2.2923054965203627e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11962656,
      "step": 21060
    },
    {
      "epoch": 369.56637168141594,
      "grad_norm": 7.130907988539548e-07,
      "learning_rate": 2.2913271586945443e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11965648,
      "step": 21065
    },
    {
      "epoch": 369.65486725663715,
      "grad_norm": 1.299432028645242e-06,
      "learning_rate": 2.290348853048699e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11968752,
      "step": 21070
    },
    {
      "epoch": 369.7433628318584,
      "grad_norm": 2.6964103199134115e-06,
      "learning_rate": 2.2893705797336956e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11971440,
      "step": 21075
    },
    {
      "epoch": 369.83185840707966,
      "grad_norm": 7.55708583710657e-07,
      "learning_rate": 2.288392338900397e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11974064,
      "step": 21080
    },
    {
      "epoch": 369.92035398230087,
      "grad_norm": 1.5348808801718405e-06,
      "learning_rate": 2.2874141306996576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11976816,
      "step": 21085
    },
    {
      "epoch": 370.0,
      "grad_norm": 7.681113629587344e-07,
      "learning_rate": 2.2864359552823312e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11978976,
      "step": 21090
    },
    {
      "epoch": 370.08849557522126,
      "grad_norm": 1.0131680028280243e-06,
      "learning_rate": 2.2854578127992648e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11981376,
      "step": 21095
    },
    {
      "epoch": 370.17699115044246,
      "grad_norm": 3.6844928672508104e-06,
      "learning_rate": 2.2844797034012988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11984576,
      "step": 21100
    },
    {
      "epoch": 370.2654867256637,
      "grad_norm": 4.096865268365946e-06,
      "learning_rate": 2.2835016272392722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11987328,
      "step": 21105
    },
    {
      "epoch": 370.353982300885,
      "grad_norm": 6.539739388244925e-06,
      "learning_rate": 2.2825235844640142e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11990720,
      "step": 21110
    },
    {
      "epoch": 370.4424778761062,
      "grad_norm": 1.5737595049358788e-06,
      "learning_rate": 2.2815455752263522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11993440,
      "step": 21115
    },
    {
      "epoch": 370.53097345132744,
      "grad_norm": 3.496280214676517e-06,
      "learning_rate": 2.2805675996771092e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11996112,
      "step": 21120
    },
    {
      "epoch": 370.6194690265487,
      "grad_norm": 1.1782320825659554e-06,
      "learning_rate": 2.2795896579670987e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 11998688,
      "step": 21125
    },
    {
      "epoch": 370.7079646017699,
      "grad_norm": 6.32781029707985e-06,
      "learning_rate": 2.2786117502471337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12001520,
      "step": 21130
    },
    {
      "epoch": 370.79646017699116,
      "grad_norm": 9.321270226791967e-07,
      "learning_rate": 2.2776338766680185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12004768,
      "step": 21135
    },
    {
      "epoch": 370.88495575221236,
      "grad_norm": 3.255277988500893e-06,
      "learning_rate": 2.2766560373805533e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12007552,
      "step": 21140
    },
    {
      "epoch": 370.9734513274336,
      "grad_norm": 1.3962079492557677e-06,
      "learning_rate": 2.2756782325355353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12010352,
      "step": 21145
    },
    {
      "epoch": 371.05309734513276,
      "grad_norm": 9.179412927551311e-07,
      "learning_rate": 2.2747004622837514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12012648,
      "step": 21150
    },
    {
      "epoch": 371.14159292035396,
      "grad_norm": 3.7757563404738903e-06,
      "learning_rate": 2.2737227267759878e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12016232,
      "step": 21155
    },
    {
      "epoch": 371.2300884955752,
      "grad_norm": 2.465255647621234e-06,
      "learning_rate": 2.272745026163024e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12018872,
      "step": 21160
    },
    {
      "epoch": 371.3185840707965,
      "grad_norm": 1.223811636918981e-06,
      "learning_rate": 2.271767360595633e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12021144,
      "step": 21165
    },
    {
      "epoch": 371.4070796460177,
      "grad_norm": 2.9200982680777088e-06,
      "learning_rate": 2.270789730224583e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12023752,
      "step": 21170
    },
    {
      "epoch": 371.49557522123894,
      "grad_norm": 2.421649924144731e-06,
      "learning_rate": 2.2698121352006367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12026648,
      "step": 21175
    },
    {
      "epoch": 371.5840707964602,
      "grad_norm": 8.222987162298523e-07,
      "learning_rate": 2.2688345756745517e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12029368,
      "step": 21180
    },
    {
      "epoch": 371.6725663716814,
      "grad_norm": 2.478017222529161e-06,
      "learning_rate": 2.267857051797081e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12032120,
      "step": 21185
    },
    {
      "epoch": 371.76106194690266,
      "grad_norm": 8.448973289887363e-07,
      "learning_rate": 2.2668795637189695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12035000,
      "step": 21190
    },
    {
      "epoch": 371.8495575221239,
      "grad_norm": 1.9207016066502547e-06,
      "learning_rate": 2.2659021115909586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12038024,
      "step": 21195
    },
    {
      "epoch": 371.9380530973451,
      "grad_norm": 5.878424417460337e-06,
      "learning_rate": 2.2649246955637847e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12041416,
      "step": 21200
    },
    {
      "epoch": 371.9380530973451,
      "eval_loss": 0.5175203680992126,
      "eval_runtime": 1.0614,
      "eval_samples_per_second": 23.553,
      "eval_steps_per_second": 12.248,
      "num_input_tokens_seen": 12041416,
      "step": 21200
    },
    {
      "epoch": 372.01769911504425,
      "grad_norm": 9.221263894687581e-07,
      "learning_rate": 2.2639473157881766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12044088,
      "step": 21205
    },
    {
      "epoch": 372.1061946902655,
      "grad_norm": 9.160070817415544e-07,
      "learning_rate": 2.2629699724148594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12046792,
      "step": 21210
    },
    {
      "epoch": 372.1946902654867,
      "grad_norm": 1.7043490743162693e-06,
      "learning_rate": 2.26199266559455e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12049480,
      "step": 21215
    },
    {
      "epoch": 372.283185840708,
      "grad_norm": 1.0251854291709606e-05,
      "learning_rate": 2.2610153954779625e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12052552,
      "step": 21220
    },
    {
      "epoch": 372.37168141592923,
      "grad_norm": 1.2903095694127842e-06,
      "learning_rate": 2.2600381622158056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12055480,
      "step": 21225
    },
    {
      "epoch": 372.46017699115043,
      "grad_norm": 1.5992286535038147e-06,
      "learning_rate": 2.2590609659587783e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12058232,
      "step": 21230
    },
    {
      "epoch": 372.5486725663717,
      "grad_norm": 6.487280643341364e-06,
      "learning_rate": 2.2580838068575787e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12060808,
      "step": 21235
    },
    {
      "epoch": 372.6371681415929,
      "grad_norm": 4.196440841042204e-06,
      "learning_rate": 2.257106685062896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12063896,
      "step": 21240
    },
    {
      "epoch": 372.72566371681415,
      "grad_norm": 2.8348792966426117e-06,
      "learning_rate": 2.256129600725415e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12066968,
      "step": 21245
    },
    {
      "epoch": 372.8141592920354,
      "grad_norm": 5.196215511205082e-07,
      "learning_rate": 2.2551525539958145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12070200,
      "step": 21250
    },
    {
      "epoch": 372.9026548672566,
      "grad_norm": 1.4670825976281776e-06,
      "learning_rate": 2.2541755450247663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12072728,
      "step": 21255
    },
    {
      "epoch": 372.9911504424779,
      "grad_norm": 3.006315182574326e-06,
      "learning_rate": 2.2531985739629382e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12075480,
      "step": 21260
    },
    {
      "epoch": 373.070796460177,
      "grad_norm": 6.283906941462192e-07,
      "learning_rate": 2.2522216409609924e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12077736,
      "step": 21265
    },
    {
      "epoch": 373.1592920353982,
      "grad_norm": 1.4503497141049593e-06,
      "learning_rate": 2.2512447461695826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12080472,
      "step": 21270
    },
    {
      "epoch": 373.24778761061947,
      "grad_norm": 5.391160016188223e-07,
      "learning_rate": 2.2502678897393593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12083224,
      "step": 21275
    },
    {
      "epoch": 373.3362831858407,
      "grad_norm": 9.662313686931157e-07,
      "learning_rate": 2.2492910718209665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12086216,
      "step": 21280
    },
    {
      "epoch": 373.42477876106193,
      "grad_norm": 7.92158334661508e-06,
      "learning_rate": 2.2483142925650398e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12088696,
      "step": 21285
    },
    {
      "epoch": 373.5132743362832,
      "grad_norm": 2.8797016966564115e-06,
      "learning_rate": 2.247337552122213e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12091384,
      "step": 21290
    },
    {
      "epoch": 373.60176991150445,
      "grad_norm": 1.043819224832987e-06,
      "learning_rate": 2.24636085064311e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12094376,
      "step": 21295
    },
    {
      "epoch": 373.69026548672565,
      "grad_norm": 9.985244560084539e-07,
      "learning_rate": 2.245384188278351e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12097176,
      "step": 21300
    },
    {
      "epoch": 373.7787610619469,
      "grad_norm": 8.691627044754568e-07,
      "learning_rate": 2.2444075651785513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12100488,
      "step": 21305
    },
    {
      "epoch": 373.86725663716817,
      "grad_norm": 9.007004564409726e-07,
      "learning_rate": 2.243430981494316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12103288,
      "step": 21310
    },
    {
      "epoch": 373.95575221238937,
      "grad_norm": 1.1625240858847974e-06,
      "learning_rate": 2.2424544373762475e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12106168,
      "step": 21315
    },
    {
      "epoch": 374.0353982300885,
      "grad_norm": 7.637248131686647e-07,
      "learning_rate": 2.2414779329749418e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12108248,
      "step": 21320
    },
    {
      "epoch": 374.12389380530976,
      "grad_norm": 6.553593721037032e-07,
      "learning_rate": 2.2405014684409873e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12111720,
      "step": 21325
    },
    {
      "epoch": 374.21238938053096,
      "grad_norm": 8.441554655291839e-07,
      "learning_rate": 2.239525043924968e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12115000,
      "step": 21330
    },
    {
      "epoch": 374.3008849557522,
      "grad_norm": 5.476483124766673e-07,
      "learning_rate": 2.2385486595774592e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12117720,
      "step": 21335
    },
    {
      "epoch": 374.3893805309734,
      "grad_norm": 9.358024612993177e-07,
      "learning_rate": 2.237572315549033e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12120152,
      "step": 21340
    },
    {
      "epoch": 374.4778761061947,
      "grad_norm": 1.0726495247581624e-06,
      "learning_rate": 2.2365960119902545e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12122616,
      "step": 21345
    },
    {
      "epoch": 374.56637168141594,
      "grad_norm": 1.2149938584116171e-06,
      "learning_rate": 2.2356197490516806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12125672,
      "step": 21350
    },
    {
      "epoch": 374.65486725663715,
      "grad_norm": 3.042925754925818e-06,
      "learning_rate": 2.234643526883863e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12128456,
      "step": 21355
    },
    {
      "epoch": 374.7433628318584,
      "grad_norm": 7.596732416459417e-07,
      "learning_rate": 2.2336673456373497e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12131288,
      "step": 21360
    },
    {
      "epoch": 374.83185840707966,
      "grad_norm": 9.215276008944784e-07,
      "learning_rate": 2.2326912054626772e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12133960,
      "step": 21365
    },
    {
      "epoch": 374.92035398230087,
      "grad_norm": 1.037150354932237e-06,
      "learning_rate": 2.2317151065103813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12136904,
      "step": 21370
    },
    {
      "epoch": 375.0,
      "grad_norm": 1.0429453141114209e-06,
      "learning_rate": 2.2307390489309865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12139256,
      "step": 21375
    },
    {
      "epoch": 375.08849557522126,
      "grad_norm": 2.604138444439741e-06,
      "learning_rate": 2.2297630328750146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12142568,
      "step": 21380
    },
    {
      "epoch": 375.17699115044246,
      "grad_norm": 4.094438281754265e-06,
      "learning_rate": 2.228787058492979e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12145368,
      "step": 21385
    },
    {
      "epoch": 375.2654867256637,
      "grad_norm": 1.0017129170591943e-06,
      "learning_rate": 2.2278111259353875e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12148312,
      "step": 21390
    },
    {
      "epoch": 375.353982300885,
      "grad_norm": 6.865586783533217e-07,
      "learning_rate": 2.2268352353527395e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12150776,
      "step": 21395
    },
    {
      "epoch": 375.4424778761062,
      "grad_norm": 7.254021170410851e-07,
      "learning_rate": 2.225859386895533e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12153176,
      "step": 21400
    },
    {
      "epoch": 375.4424778761062,
      "eval_loss": 0.5526835918426514,
      "eval_runtime": 1.0655,
      "eval_samples_per_second": 23.463,
      "eval_steps_per_second": 12.201,
      "num_input_tokens_seen": 12153176,
      "step": 21400
    },
    {
      "epoch": 375.53097345132744,
      "grad_norm": 7.551545309070207e-07,
      "learning_rate": 2.2248835807142525e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12155576,
      "step": 21405
    },
    {
      "epoch": 375.6194690265487,
      "grad_norm": 6.215484518179437e-07,
      "learning_rate": 2.2239078169593826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12158952,
      "step": 21410
    },
    {
      "epoch": 375.7079646017699,
      "grad_norm": 4.959079888067208e-06,
      "learning_rate": 2.222932095781396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12162232,
      "step": 21415
    },
    {
      "epoch": 375.79646017699116,
      "grad_norm": 1.0168014341616072e-06,
      "learning_rate": 2.221956417330762e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12165160,
      "step": 21420
    },
    {
      "epoch": 375.88495575221236,
      "grad_norm": 9.793219533094089e-07,
      "learning_rate": 2.2209807817579438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12167704,
      "step": 21425
    },
    {
      "epoch": 375.9734513274336,
      "grad_norm": 2.491276063665282e-06,
      "learning_rate": 2.220005189213394e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12170808,
      "step": 21430
    },
    {
      "epoch": 376.05309734513276,
      "grad_norm": 1.2844329830841161e-06,
      "learning_rate": 2.2190296398475624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12173112,
      "step": 21435
    },
    {
      "epoch": 376.14159292035396,
      "grad_norm": 2.5906570044753607e-06,
      "learning_rate": 2.2180541338108926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12175672,
      "step": 21440
    },
    {
      "epoch": 376.2300884955752,
      "grad_norm": 7.618032782374939e-07,
      "learning_rate": 2.2170786712538176e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12178584,
      "step": 21445
    },
    {
      "epoch": 376.3185840707965,
      "grad_norm": 6.422194019251037e-06,
      "learning_rate": 2.216103252326768e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12181048,
      "step": 21450
    },
    {
      "epoch": 376.4070796460177,
      "grad_norm": 1.3686362763110083e-06,
      "learning_rate": 2.2151278771801635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12183640,
      "step": 21455
    },
    {
      "epoch": 376.49557522123894,
      "grad_norm": 2.9021505270065973e-06,
      "learning_rate": 2.21415254596442e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12187160,
      "step": 21460
    },
    {
      "epoch": 376.5840707964602,
      "grad_norm": 1.0821594287335756e-06,
      "learning_rate": 2.213177258829947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12190152,
      "step": 21465
    },
    {
      "epoch": 376.6725663716814,
      "grad_norm": 1.8249106688017491e-06,
      "learning_rate": 2.2122020159271445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12193528,
      "step": 21470
    },
    {
      "epoch": 376.76106194690266,
      "grad_norm": 2.370110905758338e-06,
      "learning_rate": 2.2112268174064075e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12196360,
      "step": 21475
    },
    {
      "epoch": 376.8495575221239,
      "grad_norm": 3.559541710274061e-06,
      "learning_rate": 2.2102516634181253e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12199016,
      "step": 21480
    },
    {
      "epoch": 376.9380530973451,
      "grad_norm": 2.891407802962931e-06,
      "learning_rate": 2.209276554112677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12202232,
      "step": 21485
    },
    {
      "epoch": 377.01769911504425,
      "grad_norm": 1.1034750286853523e-06,
      "learning_rate": 2.2083014896404384e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12204616,
      "step": 21490
    },
    {
      "epoch": 377.1061946902655,
      "grad_norm": 1.1490767519717338e-06,
      "learning_rate": 2.207326470151775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12207736,
      "step": 21495
    },
    {
      "epoch": 377.1946902654867,
      "grad_norm": 5.357616146284272e-07,
      "learning_rate": 2.2063514957970477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12210952,
      "step": 21500
    },
    {
      "epoch": 377.283185840708,
      "grad_norm": 9.992473906095256e-07,
      "learning_rate": 2.205376566726611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12214168,
      "step": 21505
    },
    {
      "epoch": 377.37168141592923,
      "grad_norm": 2.6078319024236407e-06,
      "learning_rate": 2.204401683090809e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12216920,
      "step": 21510
    },
    {
      "epoch": 377.46017699115043,
      "grad_norm": 3.2744476357038366e-06,
      "learning_rate": 2.203426845039982e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12219864,
      "step": 21515
    },
    {
      "epoch": 377.5486725663717,
      "grad_norm": 9.190293894789647e-06,
      "learning_rate": 2.202452052724464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12223000,
      "step": 21520
    },
    {
      "epoch": 377.6371681415929,
      "grad_norm": 2.3366286768577993e-05,
      "learning_rate": 2.2014773062945777e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12225368,
      "step": 21525
    },
    {
      "epoch": 377.72566371681415,
      "grad_norm": 4.058972081111278e-06,
      "learning_rate": 2.2005026059006427e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12228056,
      "step": 21530
    },
    {
      "epoch": 377.8141592920354,
      "grad_norm": 8.910369615477975e-07,
      "learning_rate": 2.1995279516929695e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12230952,
      "step": 21535
    },
    {
      "epoch": 377.9026548672566,
      "grad_norm": 4.781220468430547e-07,
      "learning_rate": 2.1985533438218613e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12233944,
      "step": 21540
    },
    {
      "epoch": 377.9911504424779,
      "grad_norm": 6.729962933604838e-07,
      "learning_rate": 2.197578782437617e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12236552,
      "step": 21545
    },
    {
      "epoch": 378.070796460177,
      "grad_norm": 4.21429405150775e-07,
      "learning_rate": 2.196604267690524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12238784,
      "step": 21550
    },
    {
      "epoch": 378.1592920353982,
      "grad_norm": 3.3516535040689632e-06,
      "learning_rate": 2.195629799730865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12241248,
      "step": 21555
    },
    {
      "epoch": 378.24778761061947,
      "grad_norm": 4.4186560899106553e-07,
      "learning_rate": 2.1946553787089173e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12244688,
      "step": 21560
    },
    {
      "epoch": 378.3362831858407,
      "grad_norm": 1.0268296364301932e-06,
      "learning_rate": 2.193681004774947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12247472,
      "step": 21565
    },
    {
      "epoch": 378.42477876106193,
      "grad_norm": 7.172095592977712e-07,
      "learning_rate": 2.1927066780792154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12250176,
      "step": 21570
    },
    {
      "epoch": 378.5132743362832,
      "grad_norm": 2.677989641597378e-06,
      "learning_rate": 2.191732398771975e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12253104,
      "step": 21575
    },
    {
      "epoch": 378.60176991150445,
      "grad_norm": 2.519862391636707e-06,
      "learning_rate": 2.1907581670034725e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12256016,
      "step": 21580
    },
    {
      "epoch": 378.69026548672565,
      "grad_norm": 2.725584863583208e-06,
      "learning_rate": 2.189783982923948e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12259088,
      "step": 21585
    },
    {
      "epoch": 378.7787610619469,
      "grad_norm": 7.596447630930925e-06,
      "learning_rate": 2.1888098466836303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12261824,
      "step": 21590
    },
    {
      "epoch": 378.86725663716817,
      "grad_norm": 1.0504209058126435e-06,
      "learning_rate": 2.1878357584327457e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12265008,
      "step": 21595
    },
    {
      "epoch": 378.95575221238937,
      "grad_norm": 3.2690370517229894e-06,
      "learning_rate": 2.1868617183215103e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12267984,
      "step": 21600
    },
    {
      "epoch": 378.95575221238937,
      "eval_loss": 0.534430742263794,
      "eval_runtime": 1.0706,
      "eval_samples_per_second": 23.352,
      "eval_steps_per_second": 12.143,
      "num_input_tokens_seen": 12267984,
      "step": 21600
    },
    {
      "epoch": 379.0353982300885,
      "grad_norm": 9.343909255221661e-07,
      "learning_rate": 2.1858877265001327e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12270560,
      "step": 21605
    },
    {
      "epoch": 379.12389380530976,
      "grad_norm": 3.1165952805167763e-06,
      "learning_rate": 2.184913783118816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12273808,
      "step": 21610
    },
    {
      "epoch": 379.21238938053096,
      "grad_norm": 1.755456287355628e-05,
      "learning_rate": 2.1839398883277522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12276736,
      "step": 21615
    },
    {
      "epoch": 379.3008849557522,
      "grad_norm": 9.486803946856526e-07,
      "learning_rate": 2.182966042277129e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12279600,
      "step": 21620
    },
    {
      "epoch": 379.3893805309734,
      "grad_norm": 7.899968181845907e-07,
      "learning_rate": 2.181992245117128e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12282304,
      "step": 21625
    },
    {
      "epoch": 379.4778761061947,
      "grad_norm": 1.3208660902819247e-06,
      "learning_rate": 2.181018496997918e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12285040,
      "step": 21630
    },
    {
      "epoch": 379.56637168141594,
      "grad_norm": 5.946527039668581e-07,
      "learning_rate": 2.1800447980696648e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12287648,
      "step": 21635
    },
    {
      "epoch": 379.65486725663715,
      "grad_norm": 6.573999371539685e-07,
      "learning_rate": 2.1790711484825248e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12290288,
      "step": 21640
    },
    {
      "epoch": 379.7433628318584,
      "grad_norm": 8.492938263771066e-07,
      "learning_rate": 2.178097548386646e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12293376,
      "step": 21645
    },
    {
      "epoch": 379.83185840707966,
      "grad_norm": 3.4458457776054274e-06,
      "learning_rate": 2.1771239979321712e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12295840,
      "step": 21650
    },
    {
      "epoch": 379.92035398230087,
      "grad_norm": 5.21875904269109e-07,
      "learning_rate": 2.1761504972692327e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12298528,
      "step": 21655
    },
    {
      "epoch": 380.0,
      "grad_norm": 2.0253141883586068e-06,
      "learning_rate": 2.1751770465479572e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12301360,
      "step": 21660
    },
    {
      "epoch": 380.08849557522126,
      "grad_norm": 1.370788140775403e-06,
      "learning_rate": 2.174203645918464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12304800,
      "step": 21665
    },
    {
      "epoch": 380.17699115044246,
      "grad_norm": 4.4929761315870564e-06,
      "learning_rate": 2.1732302955308624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12307664,
      "step": 21670
    },
    {
      "epoch": 380.2654867256637,
      "grad_norm": 1.6436019905086141e-06,
      "learning_rate": 2.172256995535255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12310624,
      "step": 21675
    },
    {
      "epoch": 380.353982300885,
      "grad_norm": 9.812871212488972e-06,
      "learning_rate": 2.171283746081739e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12313376,
      "step": 21680
    },
    {
      "epoch": 380.4424778761062,
      "grad_norm": 2.776813516902621e-06,
      "learning_rate": 2.1703105473203988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12316112,
      "step": 21685
    },
    {
      "epoch": 380.53097345132744,
      "grad_norm": 2.177850319640129e-06,
      "learning_rate": 2.1693373994013168e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12318848,
      "step": 21690
    },
    {
      "epoch": 380.6194690265487,
      "grad_norm": 8.968583529167518e-07,
      "learning_rate": 2.168364302474562e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12321536,
      "step": 21695
    },
    {
      "epoch": 380.7079646017699,
      "grad_norm": 7.700948572164634e-07,
      "learning_rate": 2.167391256690199e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12324336,
      "step": 21700
    },
    {
      "epoch": 380.79646017699116,
      "grad_norm": 8.083222269306134e-07,
      "learning_rate": 2.1664182621982855e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12327280,
      "step": 21705
    },
    {
      "epoch": 380.88495575221236,
      "grad_norm": 8.963651225712965e-07,
      "learning_rate": 2.1654453191488673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12330336,
      "step": 21710
    },
    {
      "epoch": 380.9734513274336,
      "grad_norm": 2.6218779112241464e-06,
      "learning_rate": 2.1644724276919846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12333232,
      "step": 21715
    },
    {
      "epoch": 381.05309734513276,
      "grad_norm": 7.57736586365354e-07,
      "learning_rate": 2.1634995879776715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12335352,
      "step": 21720
    },
    {
      "epoch": 381.14159292035396,
      "grad_norm": 1.2232867447892204e-06,
      "learning_rate": 2.162526800155949e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12338376,
      "step": 21725
    },
    {
      "epoch": 381.2300884955752,
      "grad_norm": 5.559452802117448e-06,
      "learning_rate": 2.1615540643768363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12341416,
      "step": 21730
    },
    {
      "epoch": 381.3185840707965,
      "grad_norm": 2.792222176140058e-06,
      "learning_rate": 2.160581380790339e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12344568,
      "step": 21735
    },
    {
      "epoch": 381.4070796460177,
      "grad_norm": 8.160388915712247e-07,
      "learning_rate": 2.1596087495464586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12347592,
      "step": 21740
    },
    {
      "epoch": 381.49557522123894,
      "grad_norm": 1.2177274584246334e-06,
      "learning_rate": 2.1586361707951866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12350808,
      "step": 21745
    },
    {
      "epoch": 381.5840707964602,
      "grad_norm": 1.467937181587331e-06,
      "learning_rate": 2.157663644686507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12353304,
      "step": 21750
    },
    {
      "epoch": 381.6725663716814,
      "grad_norm": 5.547462023969274e-06,
      "learning_rate": 2.156691171370396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12356184,
      "step": 21755
    },
    {
      "epoch": 381.76106194690266,
      "grad_norm": 1.0893658100030734e-06,
      "learning_rate": 2.1557187509968195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12359080,
      "step": 21760
    },
    {
      "epoch": 381.8495575221239,
      "grad_norm": 1.7116964272645419e-06,
      "learning_rate": 2.1547463837157382e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12361832,
      "step": 21765
    },
    {
      "epoch": 381.9380530973451,
      "grad_norm": 2.6874818104261067e-06,
      "learning_rate": 2.1537740696771045e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12364424,
      "step": 21770
    },
    {
      "epoch": 382.01769911504425,
      "grad_norm": 3.783483407460153e-06,
      "learning_rate": 2.1528018090308587e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12366928,
      "step": 21775
    },
    {
      "epoch": 382.1061946902655,
      "grad_norm": 2.571741106294212e-06,
      "learning_rate": 2.151829601926938e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12369616,
      "step": 21780
    },
    {
      "epoch": 382.1946902654867,
      "grad_norm": 1.4446227396547329e-06,
      "learning_rate": 2.1508574485152684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12372720,
      "step": 21785
    },
    {
      "epoch": 382.283185840708,
      "grad_norm": 7.701748927502194e-07,
      "learning_rate": 2.1498853489457667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12375616,
      "step": 21790
    },
    {
      "epoch": 382.37168141592923,
      "grad_norm": 3.6220303627487738e-06,
      "learning_rate": 2.1489133033683455e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12378528,
      "step": 21795
    },
    {
      "epoch": 382.46017699115043,
      "grad_norm": 1.0460762496222742e-06,
      "learning_rate": 2.1479413119329038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12381424,
      "step": 21800
    },
    {
      "epoch": 382.46017699115043,
      "eval_loss": 0.5042070746421814,
      "eval_runtime": 1.0672,
      "eval_samples_per_second": 23.426,
      "eval_steps_per_second": 12.181,
      "num_input_tokens_seen": 12381424,
      "step": 21800
    },
    {
      "epoch": 382.5486725663717,
      "grad_norm": 5.979391062282957e-06,
      "learning_rate": 2.1469693747893355e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12384224,
      "step": 21805
    },
    {
      "epoch": 382.6371681415929,
      "grad_norm": 5.943648829997983e-06,
      "learning_rate": 2.1459974920875274e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12387072,
      "step": 21810
    },
    {
      "epoch": 382.72566371681415,
      "grad_norm": 1.1110225841548527e-06,
      "learning_rate": 2.145025663977354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12389856,
      "step": 21815
    },
    {
      "epoch": 382.8141592920354,
      "grad_norm": 2.242713435407495e-06,
      "learning_rate": 2.1440538906086844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12392464,
      "step": 21820
    },
    {
      "epoch": 382.9026548672566,
      "grad_norm": 3.0122871521598427e-06,
      "learning_rate": 2.1430821721313782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12395520,
      "step": 21825
    },
    {
      "epoch": 382.9911504424779,
      "grad_norm": 3.27056091009581e-06,
      "learning_rate": 2.142110508695286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12398432,
      "step": 21830
    },
    {
      "epoch": 383.070796460177,
      "grad_norm": 7.317706263165746e-07,
      "learning_rate": 2.1411389004502515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12400704,
      "step": 21835
    },
    {
      "epoch": 383.1592920353982,
      "grad_norm": 1.339597588412289e-06,
      "learning_rate": 2.140167347546107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12403776,
      "step": 21840
    },
    {
      "epoch": 383.24778761061947,
      "grad_norm": 2.810952082654694e-06,
      "learning_rate": 2.1391958501326793e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12406576,
      "step": 21845
    },
    {
      "epoch": 383.3362831858407,
      "grad_norm": 1.0474037708263495e-06,
      "learning_rate": 2.1382244083597873e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12409088,
      "step": 21850
    },
    {
      "epoch": 383.42477876106193,
      "grad_norm": 4.930574959871592e-06,
      "learning_rate": 2.137253022377237e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12411952,
      "step": 21855
    },
    {
      "epoch": 383.5132743362832,
      "grad_norm": 1.1719299436663277e-05,
      "learning_rate": 2.136281692334829e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12415440,
      "step": 21860
    },
    {
      "epoch": 383.60176991150445,
      "grad_norm": 1.3905824971516267e-06,
      "learning_rate": 2.135310418382356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12418272,
      "step": 21865
    },
    {
      "epoch": 383.69026548672565,
      "grad_norm": 5.552957645704737e-06,
      "learning_rate": 2.134339200669598e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12421120,
      "step": 21870
    },
    {
      "epoch": 383.7787610619469,
      "grad_norm": 1.3074317166683613e-06,
      "learning_rate": 2.133368039346331e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12424272,
      "step": 21875
    },
    {
      "epoch": 383.86725663716817,
      "grad_norm": 1.3437413599604042e-06,
      "learning_rate": 2.1323969345623195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12426848,
      "step": 21880
    },
    {
      "epoch": 383.95575221238937,
      "grad_norm": 5.780814831268799e-07,
      "learning_rate": 2.1314258864673207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12429424,
      "step": 21885
    },
    {
      "epoch": 384.0353982300885,
      "grad_norm": 6.434535748667258e-07,
      "learning_rate": 2.130454895211082e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12431960,
      "step": 21890
    },
    {
      "epoch": 384.12389380530976,
      "grad_norm": 7.470909622497857e-06,
      "learning_rate": 2.129483960943342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12434840,
      "step": 21895
    },
    {
      "epoch": 384.21238938053096,
      "grad_norm": 5.692270406143507e-07,
      "learning_rate": 2.128513083813831e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12437944,
      "step": 21900
    },
    {
      "epoch": 384.3008849557522,
      "grad_norm": 1.3437286270345794e-06,
      "learning_rate": 2.1275422639722724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12440568,
      "step": 21905
    },
    {
      "epoch": 384.3893805309734,
      "grad_norm": 2.1707883206545375e-06,
      "learning_rate": 2.126571501568376e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12443912,
      "step": 21910
    },
    {
      "epoch": 384.4778761061947,
      "grad_norm": 1.671969812377938e-06,
      "learning_rate": 2.1256007967518478e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12447032,
      "step": 21915
    },
    {
      "epoch": 384.56637168141594,
      "grad_norm": 3.6765529785043327e-06,
      "learning_rate": 2.124630149672381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12449544,
      "step": 21920
    },
    {
      "epoch": 384.65486725663715,
      "grad_norm": 3.487109552224865e-06,
      "learning_rate": 2.1236595604796624e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12452504,
      "step": 21925
    },
    {
      "epoch": 384.7433628318584,
      "grad_norm": 1.561529188620625e-06,
      "learning_rate": 2.1226890293233693e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12455560,
      "step": 21930
    },
    {
      "epoch": 384.83185840707966,
      "grad_norm": 1.1266497494943906e-06,
      "learning_rate": 2.1217185563531694e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12458104,
      "step": 21935
    },
    {
      "epoch": 384.92035398230087,
      "grad_norm": 5.98282838382147e-07,
      "learning_rate": 2.120748141718721e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12460680,
      "step": 21940
    },
    {
      "epoch": 385.0,
      "grad_norm": 3.0685914680361748e-06,
      "learning_rate": 2.1197777855696765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12462904,
      "step": 21945
    },
    {
      "epoch": 385.08849557522126,
      "grad_norm": 3.3846845326479524e-06,
      "learning_rate": 2.1188074880556746e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12466184,
      "step": 21950
    },
    {
      "epoch": 385.17699115044246,
      "grad_norm": 1.8076514152198797e-06,
      "learning_rate": 2.1178372493263495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12468536,
      "step": 21955
    },
    {
      "epoch": 385.2654867256637,
      "grad_norm": 2.394117700532661e-06,
      "learning_rate": 2.116867069531322e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12471016,
      "step": 21960
    },
    {
      "epoch": 385.353982300885,
      "grad_norm": 3.03965634884662e-06,
      "learning_rate": 2.1158969488202073e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12474040,
      "step": 21965
    },
    {
      "epoch": 385.4424778761062,
      "grad_norm": 2.8666272555710748e-06,
      "learning_rate": 2.114926887342611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12476904,
      "step": 21970
    },
    {
      "epoch": 385.53097345132744,
      "grad_norm": 7.274057338690909e-07,
      "learning_rate": 2.113956885248127e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12479768,
      "step": 21975
    },
    {
      "epoch": 385.6194690265487,
      "grad_norm": 1.4780218862142647e-06,
      "learning_rate": 2.112986942686342e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12482552,
      "step": 21980
    },
    {
      "epoch": 385.7079646017699,
      "grad_norm": 6.260303507588105e-06,
      "learning_rate": 2.112017059806835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12485864,
      "step": 21985
    },
    {
      "epoch": 385.79646017699116,
      "grad_norm": 1.854432184700272e-06,
      "learning_rate": 2.1110472367591724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12488568,
      "step": 21990
    },
    {
      "epoch": 385.88495575221236,
      "grad_norm": 1.5479066632906324e-06,
      "learning_rate": 2.1100774736929145e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12491368,
      "step": 21995
    },
    {
      "epoch": 385.9734513274336,
      "grad_norm": 3.0580565635318635e-06,
      "learning_rate": 2.10910777075761e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12494280,
      "step": 22000
    },
    {
      "epoch": 385.9734513274336,
      "eval_loss": 0.5429548621177673,
      "eval_runtime": 1.0655,
      "eval_samples_per_second": 23.462,
      "eval_steps_per_second": 12.2,
      "num_input_tokens_seen": 12494280,
      "step": 22000
    },
    {
      "epoch": 386.05309734513276,
      "grad_norm": 1.0653167237251182e-06,
      "learning_rate": 2.108138128102799e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12496928,
      "step": 22005
    },
    {
      "epoch": 386.14159292035396,
      "grad_norm": 8.65803713168134e-07,
      "learning_rate": 2.107168545878014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12499904,
      "step": 22010
    },
    {
      "epoch": 386.2300884955752,
      "grad_norm": 5.1437878028082196e-06,
      "learning_rate": 2.106199024232775e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12502752,
      "step": 22015
    },
    {
      "epoch": 386.3185840707965,
      "grad_norm": 9.282526320930629e-07,
      "learning_rate": 2.105229563316595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12505552,
      "step": 22020
    },
    {
      "epoch": 386.4070796460177,
      "grad_norm": 2.189568022004096e-06,
      "learning_rate": 2.1042601632789784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12508000,
      "step": 22025
    },
    {
      "epoch": 386.49557522123894,
      "grad_norm": 8.286052093353646e-07,
      "learning_rate": 2.103290824269417e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12510784,
      "step": 22030
    },
    {
      "epoch": 386.5840707964602,
      "grad_norm": 6.975812993914587e-07,
      "learning_rate": 2.1023215464373965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12513920,
      "step": 22035
    },
    {
      "epoch": 386.6725663716814,
      "grad_norm": 3.224999773010495e-06,
      "learning_rate": 2.1013523299323908e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12516448,
      "step": 22040
    },
    {
      "epoch": 386.76106194690266,
      "grad_norm": 6.198373284860281e-06,
      "learning_rate": 2.1003831749038654e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12519440,
      "step": 22045
    },
    {
      "epoch": 386.8495575221239,
      "grad_norm": 1.4572865438822191e-06,
      "learning_rate": 2.099414081501277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12522544,
      "step": 22050
    },
    {
      "epoch": 386.9380530973451,
      "grad_norm": 2.7620976652542595e-06,
      "learning_rate": 2.09844504987407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12525408,
      "step": 22055
    },
    {
      "epoch": 387.01769911504425,
      "grad_norm": 1.4425525023398222e-06,
      "learning_rate": 2.097476080171683e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12527936,
      "step": 22060
    },
    {
      "epoch": 387.1061946902655,
      "grad_norm": 1.336400828222395e-06,
      "learning_rate": 2.0965071725435436e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12530976,
      "step": 22065
    },
    {
      "epoch": 387.1946902654867,
      "grad_norm": 6.946791586415202e-07,
      "learning_rate": 2.0955383271390684e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12533952,
      "step": 22070
    },
    {
      "epoch": 387.283185840708,
      "grad_norm": 1.3493262258634786e-06,
      "learning_rate": 2.094569544107666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12536656,
      "step": 22075
    },
    {
      "epoch": 387.37168141592923,
      "grad_norm": 8.619950335742033e-07,
      "learning_rate": 2.093600823598735e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12539104,
      "step": 22080
    },
    {
      "epoch": 387.46017699115043,
      "grad_norm": 8.274053016066318e-07,
      "learning_rate": 2.092632165761663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12541920,
      "step": 22085
    },
    {
      "epoch": 387.5486725663717,
      "grad_norm": 2.7363284971215762e-06,
      "learning_rate": 2.091663570745832e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12545024,
      "step": 22090
    },
    {
      "epoch": 387.6371681415929,
      "grad_norm": 8.88974852841784e-07,
      "learning_rate": 2.0906950387006086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12547920,
      "step": 22095
    },
    {
      "epoch": 387.72566371681415,
      "grad_norm": 2.1087093955429737e-06,
      "learning_rate": 2.0897265697753543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12550848,
      "step": 22100
    },
    {
      "epoch": 387.8141592920354,
      "grad_norm": 1.1309815590720973e-06,
      "learning_rate": 2.088758164119419e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12553952,
      "step": 22105
    },
    {
      "epoch": 387.9026548672566,
      "grad_norm": 1.137355297942122e-06,
      "learning_rate": 2.0877898218821428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12556880,
      "step": 22110
    },
    {
      "epoch": 387.9911504424779,
      "grad_norm": 2.067702098429436e-06,
      "learning_rate": 2.0868215432128565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12559952,
      "step": 22115
    },
    {
      "epoch": 388.070796460177,
      "grad_norm": 1.313181428486132e-06,
      "learning_rate": 2.0858533282608796e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12562832,
      "step": 22120
    },
    {
      "epoch": 388.1592920353982,
      "grad_norm": 9.755927976584644e-07,
      "learning_rate": 2.084885177175524e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12566016,
      "step": 22125
    },
    {
      "epoch": 388.24778761061947,
      "grad_norm": 2.6181639896094566e-06,
      "learning_rate": 2.0839170901060917e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12568544,
      "step": 22130
    },
    {
      "epoch": 388.3362831858407,
      "grad_norm": 5.892808530916227e-06,
      "learning_rate": 2.082949067201872e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12571184,
      "step": 22135
    },
    {
      "epoch": 388.42477876106193,
      "grad_norm": 2.5243468826374738e-06,
      "learning_rate": 2.0819811086121475e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12573904,
      "step": 22140
    },
    {
      "epoch": 388.5132743362832,
      "grad_norm": 2.8896902222186327e-06,
      "learning_rate": 2.08101321448619e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12576800,
      "step": 22145
    },
    {
      "epoch": 388.60176991150445,
      "grad_norm": 1.3974298553875997e-06,
      "learning_rate": 2.080045384973259e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12579664,
      "step": 22150
    },
    {
      "epoch": 388.69026548672565,
      "grad_norm": 2.352926230742014e-06,
      "learning_rate": 2.0790776202226082e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12582288,
      "step": 22155
    },
    {
      "epoch": 388.7787610619469,
      "grad_norm": 2.732533857852104e-06,
      "learning_rate": 2.078109920383477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12585184,
      "step": 22160
    },
    {
      "epoch": 388.86725663716817,
      "grad_norm": 8.165276312865899e-07,
      "learning_rate": 2.0771422856050978e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12587952,
      "step": 22165
    },
    {
      "epoch": 388.95575221238937,
      "grad_norm": 1.947784994627e-06,
      "learning_rate": 2.076174716036693e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12590960,
      "step": 22170
    },
    {
      "epoch": 389.0353982300885,
      "grad_norm": 4.189382707409095e-06,
      "learning_rate": 2.075207211827472e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12593288,
      "step": 22175
    },
    {
      "epoch": 389.12389380530976,
      "grad_norm": 1.1378350563973072e-06,
      "learning_rate": 2.074239773126638e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12596200,
      "step": 22180
    },
    {
      "epoch": 389.21238938053096,
      "grad_norm": 1.2966969507033355e-06,
      "learning_rate": 2.073272400083382e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12599256,
      "step": 22185
    },
    {
      "epoch": 389.3008849557522,
      "grad_norm": 2.679401404748205e-06,
      "learning_rate": 2.072305092846883e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12601800,
      "step": 22190
    },
    {
      "epoch": 389.3893805309734,
      "grad_norm": 1.1283218555036e-06,
      "learning_rate": 2.0713378515663152e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12604984,
      "step": 22195
    },
    {
      "epoch": 389.4778761061947,
      "grad_norm": 7.808696864231024e-06,
      "learning_rate": 2.070370676390836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12608008,
      "step": 22200
    },
    {
      "epoch": 389.4778761061947,
      "eval_loss": 0.5208003520965576,
      "eval_runtime": 1.061,
      "eval_samples_per_second": 23.563,
      "eval_steps_per_second": 12.253,
      "num_input_tokens_seen": 12608008,
      "step": 22200
    },
    {
      "epoch": 389.56637168141594,
      "grad_norm": 1.3926513702244847e-06,
      "learning_rate": 2.0694035674695974e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12610984,
      "step": 22205
    },
    {
      "epoch": 389.65486725663715,
      "grad_norm": 6.439798653445905e-06,
      "learning_rate": 2.0684365249517416e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12613400,
      "step": 22210
    },
    {
      "epoch": 389.7433628318584,
      "grad_norm": 3.303094672446605e-06,
      "learning_rate": 2.067469548986396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12615992,
      "step": 22215
    },
    {
      "epoch": 389.83185840707966,
      "grad_norm": 1.3509002201317344e-06,
      "learning_rate": 2.066502639722681e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12619112,
      "step": 22220
    },
    {
      "epoch": 389.92035398230087,
      "grad_norm": 2.422016677883221e-06,
      "learning_rate": 2.065535797309708e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12622136,
      "step": 22225
    },
    {
      "epoch": 390.0,
      "grad_norm": 2.0230543213983765e-06,
      "learning_rate": 2.0645690218965736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12624536,
      "step": 22230
    },
    {
      "epoch": 390.08849557522126,
      "grad_norm": 1.1366385024302872e-06,
      "learning_rate": 2.063602313632369e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12627944,
      "step": 22235
    },
    {
      "epoch": 390.17699115044246,
      "grad_norm": 9.04352475572523e-07,
      "learning_rate": 2.0626356726661704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12631048,
      "step": 22240
    },
    {
      "epoch": 390.2654867256637,
      "grad_norm": 3.19768037115864e-06,
      "learning_rate": 2.0616690991470477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12633560,
      "step": 22245
    },
    {
      "epoch": 390.353982300885,
      "grad_norm": 8.1511950611457e-07,
      "learning_rate": 2.0607025932240595e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12636600,
      "step": 22250
    },
    {
      "epoch": 390.4424778761062,
      "grad_norm": 7.567504667349567e-07,
      "learning_rate": 2.059736155046251e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12639320,
      "step": 22255
    },
    {
      "epoch": 390.53097345132744,
      "grad_norm": 8.669204021316546e-07,
      "learning_rate": 2.0587697847626603e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12642328,
      "step": 22260
    },
    {
      "epoch": 390.6194690265487,
      "grad_norm": 5.107588094688253e-07,
      "learning_rate": 2.057803482522314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12644936,
      "step": 22265
    },
    {
      "epoch": 390.7079646017699,
      "grad_norm": 8.454485396214295e-06,
      "learning_rate": 2.056837248474227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12647432,
      "step": 22270
    },
    {
      "epoch": 390.79646017699116,
      "grad_norm": 4.4589137360162567e-07,
      "learning_rate": 2.0558710827674064e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12650072,
      "step": 22275
    },
    {
      "epoch": 390.88495575221236,
      "grad_norm": 4.4358549189382757e-07,
      "learning_rate": 2.054904985550845e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12652728,
      "step": 22280
    },
    {
      "epoch": 390.9734513274336,
      "grad_norm": 1.1992964346063673e-06,
      "learning_rate": 2.0539389569735287e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12656328,
      "step": 22285
    },
    {
      "epoch": 391.05309734513276,
      "grad_norm": 6.373555265781761e-07,
      "learning_rate": 2.052972997184431e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12658760,
      "step": 22290
    },
    {
      "epoch": 391.14159292035396,
      "grad_norm": 8.286022534775839e-07,
      "learning_rate": 2.0520071063325146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12661448,
      "step": 22295
    },
    {
      "epoch": 391.2300884955752,
      "grad_norm": 5.2304553719295654e-06,
      "learning_rate": 2.051041284566732e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12664920,
      "step": 22300
    },
    {
      "epoch": 391.3185840707965,
      "grad_norm": 1.7996169390244177e-06,
      "learning_rate": 2.050075532036026e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12667864,
      "step": 22305
    },
    {
      "epoch": 391.4070796460177,
      "grad_norm": 9.204442221744102e-07,
      "learning_rate": 2.0491098488893264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12671160,
      "step": 22310
    },
    {
      "epoch": 391.49557522123894,
      "grad_norm": 1.0748786962722079e-06,
      "learning_rate": 2.0481442352755546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12673672,
      "step": 22315
    },
    {
      "epoch": 391.5840707964602,
      "grad_norm": 2.470658500897116e-06,
      "learning_rate": 2.0471786913436198e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12676552,
      "step": 22320
    },
    {
      "epoch": 391.6725663716814,
      "grad_norm": 2.875483687603264e-06,
      "learning_rate": 2.0462132172424218e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12679160,
      "step": 22325
    },
    {
      "epoch": 391.76106194690266,
      "grad_norm": 1.2837786016461905e-06,
      "learning_rate": 2.0452478131208484e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12681640,
      "step": 22330
    },
    {
      "epoch": 391.8495575221239,
      "grad_norm": 5.278599928715266e-06,
      "learning_rate": 2.0442824791277765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12684328,
      "step": 22335
    },
    {
      "epoch": 391.9380530973451,
      "grad_norm": 1.057239842339186e-06,
      "learning_rate": 2.0433172154120727e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12687224,
      "step": 22340
    },
    {
      "epoch": 392.01769911504425,
      "grad_norm": 7.193516466941219e-07,
      "learning_rate": 2.0423520221225947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12689680,
      "step": 22345
    },
    {
      "epoch": 392.1061946902655,
      "grad_norm": 2.7858168323291466e-06,
      "learning_rate": 2.0413868994081848e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12692400,
      "step": 22350
    },
    {
      "epoch": 392.1946902654867,
      "grad_norm": 1.0755187531685806e-06,
      "learning_rate": 2.0404218474176795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12695712,
      "step": 22355
    },
    {
      "epoch": 392.283185840708,
      "grad_norm": 3.214060143363895e-06,
      "learning_rate": 2.0394568662999002e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12698448,
      "step": 22360
    },
    {
      "epoch": 392.37168141592923,
      "grad_norm": 2.8815577479690546e-06,
      "learning_rate": 2.0384919562036593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12701216,
      "step": 22365
    },
    {
      "epoch": 392.46017699115043,
      "grad_norm": 2.548455540818395e-06,
      "learning_rate": 2.0375271172777593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12703984,
      "step": 22370
    },
    {
      "epoch": 392.5486725663717,
      "grad_norm": 8.459231821689173e-07,
      "learning_rate": 2.0365623496709885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12707440,
      "step": 22375
    },
    {
      "epoch": 392.6371681415929,
      "grad_norm": 2.4898428819142282e-05,
      "learning_rate": 2.0355976535321283e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12710336,
      "step": 22380
    },
    {
      "epoch": 392.72566371681415,
      "grad_norm": 6.8739491325686686e-06,
      "learning_rate": 2.034633029009945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12713024,
      "step": 22385
    },
    {
      "epoch": 392.8141592920354,
      "grad_norm": 3.251032467233017e-06,
      "learning_rate": 2.0336684762531972e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12715952,
      "step": 22390
    },
    {
      "epoch": 392.9026548672566,
      "grad_norm": 3.0909232009435073e-06,
      "learning_rate": 2.032703995410631e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12718672,
      "step": 22395
    },
    {
      "epoch": 392.9911504424779,
      "grad_norm": 9.071108024727437e-07,
      "learning_rate": 2.031739586630981e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12721456,
      "step": 22400
    },
    {
      "epoch": 392.9911504424779,
      "eval_loss": 0.5806827545166016,
      "eval_runtime": 1.0742,
      "eval_samples_per_second": 23.272,
      "eval_steps_per_second": 12.102,
      "num_input_tokens_seen": 12721456,
      "step": 22400
    },
    {
      "epoch": 393.070796460177,
      "grad_norm": 1.5938976503093727e-06,
      "learning_rate": 2.0307752500629707e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12724032,
      "step": 22405
    },
    {
      "epoch": 393.1592920353982,
      "grad_norm": 1.2271839295863174e-05,
      "learning_rate": 2.0298109858553144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12727008,
      "step": 22410
    },
    {
      "epoch": 393.24778761061947,
      "grad_norm": 1.5522954299740377e-06,
      "learning_rate": 2.028846794156712e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12730048,
      "step": 22415
    },
    {
      "epoch": 393.3362831858407,
      "grad_norm": 3.6950436879124027e-06,
      "learning_rate": 2.027882675115856e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12732960,
      "step": 22420
    },
    {
      "epoch": 393.42477876106193,
      "grad_norm": 2.709500449782354e-06,
      "learning_rate": 2.026918628881423e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12735632,
      "step": 22425
    },
    {
      "epoch": 393.5132743362832,
      "grad_norm": 1.1756615094782319e-06,
      "learning_rate": 2.0259546556020833e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12738240,
      "step": 22430
    },
    {
      "epoch": 393.60176991150445,
      "grad_norm": 1.479616457800148e-06,
      "learning_rate": 2.024990755426493e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12740544,
      "step": 22435
    },
    {
      "epoch": 393.69026548672565,
      "grad_norm": 1.5350798321378534e-06,
      "learning_rate": 2.0240269285032975e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12743872,
      "step": 22440
    },
    {
      "epoch": 393.7787610619469,
      "grad_norm": 1.5674933138143388e-06,
      "learning_rate": 2.0230631749811306e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12746496,
      "step": 22445
    },
    {
      "epoch": 393.86725663716817,
      "grad_norm": 2.9715474738623016e-06,
      "learning_rate": 2.0220994950086162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12749568,
      "step": 22450
    },
    {
      "epoch": 393.95575221238937,
      "grad_norm": 5.973275278847723e-07,
      "learning_rate": 2.021135888734365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12753008,
      "step": 22455
    },
    {
      "epoch": 394.0353982300885,
      "grad_norm": 1.599264919605048e-06,
      "learning_rate": 2.0201723563069783e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12755224,
      "step": 22460
    },
    {
      "epoch": 394.12389380530976,
      "grad_norm": 8.049447615121608e-07,
      "learning_rate": 2.0192088978750433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12757896,
      "step": 22465
    },
    {
      "epoch": 394.21238938053096,
      "grad_norm": 1.0228166047454579e-06,
      "learning_rate": 2.0182455135871385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12760840,
      "step": 22470
    },
    {
      "epoch": 394.3008849557522,
      "grad_norm": 9.553883728585788e-07,
      "learning_rate": 2.0172822035918305e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12763656,
      "step": 22475
    },
    {
      "epoch": 394.3893805309734,
      "grad_norm": 3.313043634989299e-06,
      "learning_rate": 2.016318968037671e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12766712,
      "step": 22480
    },
    {
      "epoch": 394.4778761061947,
      "grad_norm": 4.683042789110914e-06,
      "learning_rate": 2.015355807073206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12769368,
      "step": 22485
    },
    {
      "epoch": 394.56637168141594,
      "grad_norm": 1.496212689744425e-06,
      "learning_rate": 2.0143927208469664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12772568,
      "step": 22490
    },
    {
      "epoch": 394.65486725663715,
      "grad_norm": 2.4189450869016582e-06,
      "learning_rate": 2.0134297095074708e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12775304,
      "step": 22495
    },
    {
      "epoch": 394.7433628318584,
      "grad_norm": 6.141799531178549e-06,
      "learning_rate": 2.0124667732032297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12778184,
      "step": 22500
    },
    {
      "epoch": 394.83185840707966,
      "grad_norm": 4.162082404945977e-06,
      "learning_rate": 2.011503912082738e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12781112,
      "step": 22505
    },
    {
      "epoch": 394.92035398230087,
      "grad_norm": 8.625567602393858e-07,
      "learning_rate": 2.0105411262944823e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12783848,
      "step": 22510
    },
    {
      "epoch": 395.0,
      "grad_norm": 6.76136664878868e-07,
      "learning_rate": 2.0095784159869366e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12786624,
      "step": 22515
    },
    {
      "epoch": 395.08849557522126,
      "grad_norm": 1.0211043672825326e-06,
      "learning_rate": 2.0086157813085608e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12789248,
      "step": 22520
    },
    {
      "epoch": 395.17699115044246,
      "grad_norm": 1.1907644648090354e-06,
      "learning_rate": 2.0076532224078068e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12792016,
      "step": 22525
    },
    {
      "epoch": 395.2654867256637,
      "grad_norm": 2.651514705576119e-06,
      "learning_rate": 2.0066907394331142e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12795568,
      "step": 22530
    },
    {
      "epoch": 395.353982300885,
      "grad_norm": 8.511844953318359e-07,
      "learning_rate": 2.0057283325329077e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12798624,
      "step": 22535
    },
    {
      "epoch": 395.4424778761062,
      "grad_norm": 2.821830094035249e-06,
      "learning_rate": 2.0047660018556047e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12801184,
      "step": 22540
    },
    {
      "epoch": 395.53097345132744,
      "grad_norm": 7.478166025975952e-06,
      "learning_rate": 2.0038037475496075e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12803872,
      "step": 22545
    },
    {
      "epoch": 395.6194690265487,
      "grad_norm": 1.4049538776816917e-06,
      "learning_rate": 2.0028415697633073e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12806544,
      "step": 22550
    },
    {
      "epoch": 395.7079646017699,
      "grad_norm": 2.581631633802317e-06,
      "learning_rate": 2.0018794686450858e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12809376,
      "step": 22555
    },
    {
      "epoch": 395.79646017699116,
      "grad_norm": 1.227521806868026e-05,
      "learning_rate": 2.0009174443433088e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12812368,
      "step": 22560
    },
    {
      "epoch": 395.88495575221236,
      "grad_norm": 1.3702227761314134e-06,
      "learning_rate": 1.999955497006334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12815424,
      "step": 22565
    },
    {
      "epoch": 395.9734513274336,
      "grad_norm": 1.3503924947144696e-06,
      "learning_rate": 1.9989936267825067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12818208,
      "step": 22570
    },
    {
      "epoch": 396.05309734513276,
      "grad_norm": 9.350955224363133e-06,
      "learning_rate": 1.9980318338201572e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12820568,
      "step": 22575
    },
    {
      "epoch": 396.14159292035396,
      "grad_norm": 9.24546100122825e-07,
      "learning_rate": 1.997070118267607e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12823480,
      "step": 22580
    },
    {
      "epoch": 396.2300884955752,
      "grad_norm": 1.1352510682627326e-06,
      "learning_rate": 1.9961084802731654e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12826072,
      "step": 22585
    },
    {
      "epoch": 396.3185840707965,
      "grad_norm": 1.6293561202473938e-06,
      "learning_rate": 1.9951469199851273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12829112,
      "step": 22590
    },
    {
      "epoch": 396.4070796460177,
      "grad_norm": 4.931821422360372e-06,
      "learning_rate": 1.99418543755178e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12831832,
      "step": 22595
    },
    {
      "epoch": 396.49557522123894,
      "grad_norm": 1.5265690308297053e-05,
      "learning_rate": 1.9932240331213936e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12835240,
      "step": 22600
    },
    {
      "epoch": 396.49557522123894,
      "eval_loss": 0.5170691013336182,
      "eval_runtime": 1.0591,
      "eval_samples_per_second": 23.606,
      "eval_steps_per_second": 12.275,
      "num_input_tokens_seen": 12835240,
      "step": 22600
    },
    {
      "epoch": 396.5840707964602,
      "grad_norm": 1.5720520423201378e-06,
      "learning_rate": 1.9922627068422297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12838904,
      "step": 22605
    },
    {
      "epoch": 396.6725663716814,
      "grad_norm": 1.187587031381554e-06,
      "learning_rate": 1.991301458862538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12841800,
      "step": 22610
    },
    {
      "epoch": 396.76106194690266,
      "grad_norm": 1.2809764484700281e-06,
      "learning_rate": 1.9903402893305536e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12844472,
      "step": 22615
    },
    {
      "epoch": 396.8495575221239,
      "grad_norm": 8.629625654066331e-07,
      "learning_rate": 1.9893791983945016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12847128,
      "step": 22620
    },
    {
      "epoch": 396.9380530973451,
      "grad_norm": 1.0540310313444934e-06,
      "learning_rate": 1.988418186202594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12849640,
      "step": 22625
    },
    {
      "epoch": 397.01769911504425,
      "grad_norm": 1.542366362627945e-06,
      "learning_rate": 1.98745725290303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12851704,
      "step": 22630
    },
    {
      "epoch": 397.1061946902655,
      "grad_norm": 1.7452066458645277e-06,
      "learning_rate": 1.986496398644e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12854392,
      "step": 22635
    },
    {
      "epoch": 397.1946902654867,
      "grad_norm": 2.6706823064159835e-06,
      "learning_rate": 1.9855356235736777e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12857016,
      "step": 22640
    },
    {
      "epoch": 397.283185840708,
      "grad_norm": 1.2752612974509248e-06,
      "learning_rate": 1.9845749278402277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12859896,
      "step": 22645
    },
    {
      "epoch": 397.37168141592923,
      "grad_norm": 5.525728738575708e-06,
      "learning_rate": 1.9836143115918006e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12862744,
      "step": 22650
    },
    {
      "epoch": 397.46017699115043,
      "grad_norm": 1.7788578361432883e-06,
      "learning_rate": 1.9826537749765367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12865464,
      "step": 22655
    },
    {
      "epoch": 397.5486725663717,
      "grad_norm": 2.8218933039170224e-06,
      "learning_rate": 1.9816933181425625e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12868120,
      "step": 22660
    },
    {
      "epoch": 397.6371681415929,
      "grad_norm": 3.221314955226262e-06,
      "learning_rate": 1.9807329412379903e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12871112,
      "step": 22665
    },
    {
      "epoch": 397.72566371681415,
      "grad_norm": 2.537071623009979e-06,
      "learning_rate": 1.9797726444109247e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12874008,
      "step": 22670
    },
    {
      "epoch": 397.8141592920354,
      "grad_norm": 2.9698069283767836e-06,
      "learning_rate": 1.9788124278094557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12877112,
      "step": 22675
    },
    {
      "epoch": 397.9026548672566,
      "grad_norm": 9.96615995063621e-07,
      "learning_rate": 1.9778522915816594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12879944,
      "step": 22680
    },
    {
      "epoch": 397.9911504424779,
      "grad_norm": 1.2279087968636304e-06,
      "learning_rate": 1.9768922358756014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12883672,
      "step": 22685
    },
    {
      "epoch": 398.070796460177,
      "grad_norm": 2.228287485195324e-06,
      "learning_rate": 1.9759322608393353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12886360,
      "step": 22690
    },
    {
      "epoch": 398.1592920353982,
      "grad_norm": 2.6358991362940287e-06,
      "learning_rate": 1.9749723666208992e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12888904,
      "step": 22695
    },
    {
      "epoch": 398.24778761061947,
      "grad_norm": 7.661322456442576e-07,
      "learning_rate": 1.9740125533683235e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12891864,
      "step": 22700
    },
    {
      "epoch": 398.3362831858407,
      "grad_norm": 1.719147689982492e-06,
      "learning_rate": 1.9730528212296208e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12894888,
      "step": 22705
    },
    {
      "epoch": 398.42477876106193,
      "grad_norm": 1.3772282727586571e-06,
      "learning_rate": 1.9720931703527945e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12897768,
      "step": 22710
    },
    {
      "epoch": 398.5132743362832,
      "grad_norm": 2.9721722967224196e-06,
      "learning_rate": 1.9711336008858373e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12900392,
      "step": 22715
    },
    {
      "epoch": 398.60176991150445,
      "grad_norm": 2.8746519546984928e-06,
      "learning_rate": 1.9701741129767233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12903304,
      "step": 22720
    },
    {
      "epoch": 398.69026548672565,
      "grad_norm": 4.532182174443733e-06,
      "learning_rate": 1.9692147067734202e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12906360,
      "step": 22725
    },
    {
      "epoch": 398.7787610619469,
      "grad_norm": 6.679447778878966e-06,
      "learning_rate": 1.96825538242388e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12909448,
      "step": 22730
    },
    {
      "epoch": 398.86725663716817,
      "grad_norm": 1.520146270195255e-06,
      "learning_rate": 1.967296140076041e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12912376,
      "step": 22735
    },
    {
      "epoch": 398.95575221238937,
      "grad_norm": 2.948310338979354e-06,
      "learning_rate": 1.966336979877833e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12914872,
      "step": 22740
    },
    {
      "epoch": 399.0353982300885,
      "grad_norm": 2.0330012375779916e-06,
      "learning_rate": 1.9653779019771678e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12917160,
      "step": 22745
    },
    {
      "epoch": 399.12389380530976,
      "grad_norm": 1.7187107914651278e-06,
      "learning_rate": 1.9644189065219488e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12919800,
      "step": 22750
    },
    {
      "epoch": 399.21238938053096,
      "grad_norm": 1.6643709841446253e-06,
      "learning_rate": 1.9634599936600655e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12922920,
      "step": 22755
    },
    {
      "epoch": 399.3008849557522,
      "grad_norm": 4.698299562733155e-06,
      "learning_rate": 1.9625011635393935e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12925656,
      "step": 22760
    },
    {
      "epoch": 399.3893805309734,
      "grad_norm": 3.8944090192671865e-06,
      "learning_rate": 1.9615424163077963e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12928568,
      "step": 22765
    },
    {
      "epoch": 399.4778761061947,
      "grad_norm": 1.2694976021521143e-06,
      "learning_rate": 1.9605837521131263e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12931720,
      "step": 22770
    },
    {
      "epoch": 399.56637168141594,
      "grad_norm": 3.0080657325015636e-06,
      "learning_rate": 1.9596251711032192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12934264,
      "step": 22775
    },
    {
      "epoch": 399.65486725663715,
      "grad_norm": 1.2525771353466553e-06,
      "learning_rate": 1.958666673425903e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12937848,
      "step": 22780
    },
    {
      "epoch": 399.7433628318584,
      "grad_norm": 1.8970237078974606e-06,
      "learning_rate": 1.957708259228987e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12940632,
      "step": 22785
    },
    {
      "epoch": 399.83185840707966,
      "grad_norm": 3.2642826681694714e-06,
      "learning_rate": 1.956749928660273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12943512,
      "step": 22790
    },
    {
      "epoch": 399.92035398230087,
      "grad_norm": 2.9749419354629936e-06,
      "learning_rate": 1.955791681867547e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12946456,
      "step": 22795
    },
    {
      "epoch": 400.0,
      "grad_norm": 1.842039068833401e-06,
      "learning_rate": 1.9548335189985824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12948416,
      "step": 22800
    },
    {
      "epoch": 400.0,
      "eval_loss": 0.5288233160972595,
      "eval_runtime": 1.0683,
      "eval_samples_per_second": 23.402,
      "eval_steps_per_second": 12.169,
      "num_input_tokens_seen": 12948416,
      "step": 22800
    },
    {
      "epoch": 400.08849557522126,
      "grad_norm": 3.424512897254317e-06,
      "learning_rate": 1.9538754402011396e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12951568,
      "step": 22805
    },
    {
      "epoch": 400.17699115044246,
      "grad_norm": 1.5195364539977163e-06,
      "learning_rate": 1.952917445622968e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12954160,
      "step": 22810
    },
    {
      "epoch": 400.2654867256637,
      "grad_norm": 5.5758137023076415e-06,
      "learning_rate": 1.9519595354118005e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12957136,
      "step": 22815
    },
    {
      "epoch": 400.353982300885,
      "grad_norm": 1.3369473208513227e-06,
      "learning_rate": 1.951001709715361e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12959712,
      "step": 22820
    },
    {
      "epoch": 400.4424778761062,
      "grad_norm": 7.412505169668293e-07,
      "learning_rate": 1.9500439686813556e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12962416,
      "step": 22825
    },
    {
      "epoch": 400.53097345132744,
      "grad_norm": 7.023148782536737e-07,
      "learning_rate": 1.949086312457482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12965456,
      "step": 22830
    },
    {
      "epoch": 400.6194690265487,
      "grad_norm": 2.516169843147509e-06,
      "learning_rate": 1.9481287411914223e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12968480,
      "step": 22835
    },
    {
      "epoch": 400.7079646017699,
      "grad_norm": 2.4290106921398547e-06,
      "learning_rate": 1.9471712550308457e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12971376,
      "step": 22840
    },
    {
      "epoch": 400.79646017699116,
      "grad_norm": 7.832499022697448e-07,
      "learning_rate": 1.946213854123409e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12974288,
      "step": 22845
    },
    {
      "epoch": 400.88495575221236,
      "grad_norm": 2.338689682801487e-06,
      "learning_rate": 1.9452565386167554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12976928,
      "step": 22850
    },
    {
      "epoch": 400.9734513274336,
      "grad_norm": 2.2833044113212964e-06,
      "learning_rate": 1.9442993086585142e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12979712,
      "step": 22855
    },
    {
      "epoch": 401.05309734513276,
      "grad_norm": 1.1440932894402067e-06,
      "learning_rate": 1.9433421643963043e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12982208,
      "step": 22860
    },
    {
      "epoch": 401.14159292035396,
      "grad_norm": 1.2569374803206301e-06,
      "learning_rate": 1.942385105977727e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12985184,
      "step": 22865
    },
    {
      "epoch": 401.2300884955752,
      "grad_norm": 1.215040015267732e-06,
      "learning_rate": 1.9414281335503743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12987984,
      "step": 22870
    },
    {
      "epoch": 401.3185840707965,
      "grad_norm": 1.119659714277077e-06,
      "learning_rate": 1.9404712472618232e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12990400,
      "step": 22875
    },
    {
      "epoch": 401.4070796460177,
      "grad_norm": 1.0299402219970943e-06,
      "learning_rate": 1.939514447259636e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12993296,
      "step": 22880
    },
    {
      "epoch": 401.49557522123894,
      "grad_norm": 2.7551184302865295e-06,
      "learning_rate": 1.938557733691365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12996224,
      "step": 22885
    },
    {
      "epoch": 401.5840707964602,
      "grad_norm": 1.181260813609697e-06,
      "learning_rate": 1.9376011067045476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 12999248,
      "step": 22890
    },
    {
      "epoch": 401.6725663716814,
      "grad_norm": 3.2867092158994637e-06,
      "learning_rate": 1.9366445664467065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13002672,
      "step": 22895
    },
    {
      "epoch": 401.76106194690266,
      "grad_norm": 1.9784574760706164e-06,
      "learning_rate": 1.9356881130653533e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13005984,
      "step": 22900
    },
    {
      "epoch": 401.8495575221239,
      "grad_norm": 1.1793762269007857e-06,
      "learning_rate": 1.9347317467079846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13008480,
      "step": 22905
    },
    {
      "epoch": 401.9380530973451,
      "grad_norm": 5.39537904842291e-06,
      "learning_rate": 1.9337754675220836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13011088,
      "step": 22910
    },
    {
      "epoch": 402.01769911504425,
      "grad_norm": 1.4435997854889138e-06,
      "learning_rate": 1.9328192756551218e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13013504,
      "step": 22915
    },
    {
      "epoch": 402.1061946902655,
      "grad_norm": 1.467191054871364e-06,
      "learning_rate": 1.931863171254555e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13016416,
      "step": 22920
    },
    {
      "epoch": 402.1946902654867,
      "grad_norm": 1.1072162351410952e-06,
      "learning_rate": 1.930907154467826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13018992,
      "step": 22925
    },
    {
      "epoch": 402.283185840708,
      "grad_norm": 2.236157797597116e-06,
      "learning_rate": 1.9299512254423673e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13022208,
      "step": 22930
    },
    {
      "epoch": 402.37168141592923,
      "grad_norm": 7.056347612888203e-07,
      "learning_rate": 1.9289953843255914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13024912,
      "step": 22935
    },
    {
      "epoch": 402.46017699115043,
      "grad_norm": 1.4916997770342277e-06,
      "learning_rate": 1.9280396312649048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13027872,
      "step": 22940
    },
    {
      "epoch": 402.5486725663717,
      "grad_norm": 7.969629791659827e-07,
      "learning_rate": 1.9270839664076936e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13030640,
      "step": 22945
    },
    {
      "epoch": 402.6371681415929,
      "grad_norm": 9.132013474300038e-07,
      "learning_rate": 1.9261283899013345e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13033904,
      "step": 22950
    },
    {
      "epoch": 402.72566371681415,
      "grad_norm": 9.015235491460771e-07,
      "learning_rate": 1.92517290189319e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13036784,
      "step": 22955
    },
    {
      "epoch": 402.8141592920354,
      "grad_norm": 7.145451945689274e-07,
      "learning_rate": 1.924217502530607e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13039744,
      "step": 22960
    },
    {
      "epoch": 402.9026548672566,
      "grad_norm": 1.5303959344237228e-06,
      "learning_rate": 1.9232621919609207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13042384,
      "step": 22965
    },
    {
      "epoch": 402.9911504424779,
      "grad_norm": 3.25386054100818e-06,
      "learning_rate": 1.9223069703314534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13045216,
      "step": 22970
    },
    {
      "epoch": 403.070796460177,
      "grad_norm": 5.641563348035561e-06,
      "learning_rate": 1.92135183778951e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13047600,
      "step": 22975
    },
    {
      "epoch": 403.1592920353982,
      "grad_norm": 1.3683463748748181e-06,
      "learning_rate": 1.9203967944823857e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13049888,
      "step": 22980
    },
    {
      "epoch": 403.24778761061947,
      "grad_norm": 1.3687075579582597e-06,
      "learning_rate": 1.9194418405573588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13052912,
      "step": 22985
    },
    {
      "epoch": 403.3362831858407,
      "grad_norm": 2.9597179036500165e-06,
      "learning_rate": 1.9184869761616954e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13055584,
      "step": 22990
    },
    {
      "epoch": 403.42477876106193,
      "grad_norm": 2.0326883714005817e-06,
      "learning_rate": 1.9175322014426495e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13058736,
      "step": 22995
    },
    {
      "epoch": 403.5132743362832,
      "grad_norm": 1.8000246200244874e-06,
      "learning_rate": 1.9165775165474565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13061472,
      "step": 23000
    },
    {
      "epoch": 403.5132743362832,
      "eval_loss": 0.5603658556938171,
      "eval_runtime": 1.0628,
      "eval_samples_per_second": 23.523,
      "eval_steps_per_second": 12.232,
      "num_input_tokens_seen": 13061472,
      "step": 23000
    },
    {
      "epoch": 403.60176991150445,
      "grad_norm": 4.008453288406599e-06,
      "learning_rate": 1.9156229216233434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13064528,
      "step": 23005
    },
    {
      "epoch": 403.69026548672565,
      "grad_norm": 1.940057927640737e-06,
      "learning_rate": 1.9146684168175184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13067712,
      "step": 23010
    },
    {
      "epoch": 403.7787610619469,
      "grad_norm": 1.6691873270247015e-06,
      "learning_rate": 1.9137140022771796e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13070528,
      "step": 23015
    },
    {
      "epoch": 403.86725663716817,
      "grad_norm": 1.6742375237299711e-06,
      "learning_rate": 1.9127596781495103e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13073328,
      "step": 23020
    },
    {
      "epoch": 403.95575221238937,
      "grad_norm": 2.00845647668757e-06,
      "learning_rate": 1.9118054445816767e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13076688,
      "step": 23025
    },
    {
      "epoch": 404.0353982300885,
      "grad_norm": 6.350591775117209e-06,
      "learning_rate": 1.9108513017208356e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13079304,
      "step": 23030
    },
    {
      "epoch": 404.12389380530976,
      "grad_norm": 1.2631574008992175e-06,
      "learning_rate": 1.9098972497141287e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13082024,
      "step": 23035
    },
    {
      "epoch": 404.21238938053096,
      "grad_norm": 1.1128901178381057e-06,
      "learning_rate": 1.9089432887086806e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13084680,
      "step": 23040
    },
    {
      "epoch": 404.3008849557522,
      "grad_norm": 1.3615065199701348e-06,
      "learning_rate": 1.9079894188516056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13087848,
      "step": 23045
    },
    {
      "epoch": 404.3893805309734,
      "grad_norm": 2.4418175144091947e-06,
      "learning_rate": 1.907035640290002e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13090840,
      "step": 23050
    },
    {
      "epoch": 404.4778761061947,
      "grad_norm": 1.3954667110738228e-06,
      "learning_rate": 1.9060819531709534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13093688,
      "step": 23055
    },
    {
      "epoch": 404.56637168141594,
      "grad_norm": 2.8261163151910296e-06,
      "learning_rate": 1.9051283576415325e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13096616,
      "step": 23060
    },
    {
      "epoch": 404.65486725663715,
      "grad_norm": 7.188902486632287e-07,
      "learning_rate": 1.904174853848793e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13099816,
      "step": 23065
    },
    {
      "epoch": 404.7433628318584,
      "grad_norm": 2.5073163669731002e-06,
      "learning_rate": 1.903221441939779e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13102680,
      "step": 23070
    },
    {
      "epoch": 404.83185840707966,
      "grad_norm": 4.751813378334191e-07,
      "learning_rate": 1.9022681220615194e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13105320,
      "step": 23075
    },
    {
      "epoch": 404.92035398230087,
      "grad_norm": 2.1519185793295037e-06,
      "learning_rate": 1.9013148943610255e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13107912,
      "step": 23080
    },
    {
      "epoch": 405.0,
      "grad_norm": 1.3966462120151846e-06,
      "learning_rate": 1.9003617589852998e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13110120,
      "step": 23085
    },
    {
      "epoch": 405.08849557522126,
      "grad_norm": 3.6242113310436253e-06,
      "learning_rate": 1.899408716081326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13112888,
      "step": 23090
    },
    {
      "epoch": 405.17699115044246,
      "grad_norm": 1.564597368997056e-06,
      "learning_rate": 1.898455765796075e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13115752,
      "step": 23095
    },
    {
      "epoch": 405.2654867256637,
      "grad_norm": 3.525114834701526e-06,
      "learning_rate": 1.8975029082765053e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13118424,
      "step": 23100
    },
    {
      "epoch": 405.353982300885,
      "grad_norm": 1.0712037692428567e-06,
      "learning_rate": 1.8965501436695577e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13121432,
      "step": 23105
    },
    {
      "epoch": 405.4424778761062,
      "grad_norm": 2.524237743273261e-06,
      "learning_rate": 1.895597472122161e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13124024,
      "step": 23110
    },
    {
      "epoch": 405.53097345132744,
      "grad_norm": 5.822887487738626e-06,
      "learning_rate": 1.894644893781231e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13127096,
      "step": 23115
    },
    {
      "epoch": 405.6194690265487,
      "grad_norm": 9.794308652999462e-07,
      "learning_rate": 1.893692408793665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13130136,
      "step": 23120
    },
    {
      "epoch": 405.7079646017699,
      "grad_norm": 2.894787030527368e-06,
      "learning_rate": 1.8927400173063493e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13133576,
      "step": 23125
    },
    {
      "epoch": 405.79646017699116,
      "grad_norm": 1.833697865549766e-06,
      "learning_rate": 1.891787719466154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13136472,
      "step": 23130
    },
    {
      "epoch": 405.88495575221236,
      "grad_norm": 6.275775831454666e-06,
      "learning_rate": 1.8908355154199346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13139368,
      "step": 23135
    },
    {
      "epoch": 405.9734513274336,
      "grad_norm": 2.2797180463385303e-06,
      "learning_rate": 1.8898834053145357e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13142104,
      "step": 23140
    },
    {
      "epoch": 406.05309734513276,
      "grad_norm": 5.490751391334925e-07,
      "learning_rate": 1.8889313892967813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13144904,
      "step": 23145
    },
    {
      "epoch": 406.14159292035396,
      "grad_norm": 9.537599225950544e-07,
      "learning_rate": 1.8879794675134863e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13148008,
      "step": 23150
    },
    {
      "epoch": 406.2300884955752,
      "grad_norm": 5.29515091329813e-06,
      "learning_rate": 1.8870276401114494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13151032,
      "step": 23155
    },
    {
      "epoch": 406.3185840707965,
      "grad_norm": 8.460069125248992e-07,
      "learning_rate": 1.886075907237453e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13154152,
      "step": 23160
    },
    {
      "epoch": 406.4070796460177,
      "grad_norm": 2.9363279736571712e-06,
      "learning_rate": 1.8851242690382672e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13156744,
      "step": 23165
    },
    {
      "epoch": 406.49557522123894,
      "grad_norm": 3.0752003112866078e-06,
      "learning_rate": 1.884172725660645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13159512,
      "step": 23170
    },
    {
      "epoch": 406.5840707964602,
      "grad_norm": 3.024867964995792e-06,
      "learning_rate": 1.8832212772513277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13162248,
      "step": 23175
    },
    {
      "epoch": 406.6725663716814,
      "grad_norm": 7.756100899314333e-07,
      "learning_rate": 1.8822699239570414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13165464,
      "step": 23180
    },
    {
      "epoch": 406.76106194690266,
      "grad_norm": 2.2424180770030944e-06,
      "learning_rate": 1.8813186659244943e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13167848,
      "step": 23185
    },
    {
      "epoch": 406.8495575221239,
      "grad_norm": 2.605062945804093e-06,
      "learning_rate": 1.880367503300385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13170648,
      "step": 23190
    },
    {
      "epoch": 406.9380530973451,
      "grad_norm": 7.378465056717687e-07,
      "learning_rate": 1.8794164362313927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13173528,
      "step": 23195
    },
    {
      "epoch": 407.01769911504425,
      "grad_norm": 4.6880455784048536e-07,
      "learning_rate": 1.878465464864185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13175888,
      "step": 23200
    },
    {
      "epoch": 407.01769911504425,
      "eval_loss": 0.5697670578956604,
      "eval_runtime": 1.0637,
      "eval_samples_per_second": 23.504,
      "eval_steps_per_second": 12.222,
      "num_input_tokens_seen": 13175888,
      "step": 23200
    },
    {
      "epoch": 407.1061946902655,
      "grad_norm": 3.1133240554481745e-06,
      "learning_rate": 1.877514589345414e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13178864,
      "step": 23205
    },
    {
      "epoch": 407.1946902654867,
      "grad_norm": 2.87196758108621e-06,
      "learning_rate": 1.876563809821715e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13181568,
      "step": 23210
    },
    {
      "epoch": 407.283185840708,
      "grad_norm": 1.1485855111459387e-06,
      "learning_rate": 1.8756131264397106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13184592,
      "step": 23215
    },
    {
      "epoch": 407.37168141592923,
      "grad_norm": 4.817694048142585e-07,
      "learning_rate": 1.87466253934601e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13187648,
      "step": 23220
    },
    {
      "epoch": 407.46017699115043,
      "grad_norm": 1.1026334050256992e-06,
      "learning_rate": 1.8737120486872033e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13190576,
      "step": 23225
    },
    {
      "epoch": 407.5486725663717,
      "grad_norm": 6.433141379602603e-07,
      "learning_rate": 1.8727616546098696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13193664,
      "step": 23230
    },
    {
      "epoch": 407.6371681415929,
      "grad_norm": 5.841007464368886e-07,
      "learning_rate": 1.8718113572605716e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13196592,
      "step": 23235
    },
    {
      "epoch": 407.72566371681415,
      "grad_norm": 1.428530822522589e-06,
      "learning_rate": 1.8708611567858554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13199264,
      "step": 23240
    },
    {
      "epoch": 407.8141592920354,
      "grad_norm": 1.6435515135526657e-06,
      "learning_rate": 1.8699110533322565e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13202096,
      "step": 23245
    },
    {
      "epoch": 407.9026548672566,
      "grad_norm": 8.943666216509882e-07,
      "learning_rate": 1.8689610470462897e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13205088,
      "step": 23250
    },
    {
      "epoch": 407.9911504424779,
      "grad_norm": 1.0726641903602285e-06,
      "learning_rate": 1.8680111380744604e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13207872,
      "step": 23255
    },
    {
      "epoch": 408.070796460177,
      "grad_norm": 1.8243431441078428e-06,
      "learning_rate": 1.8670613265632564e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13210088,
      "step": 23260
    },
    {
      "epoch": 408.1592920353982,
      "grad_norm": 4.007057668786729e-06,
      "learning_rate": 1.866111612659149e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13213192,
      "step": 23265
    },
    {
      "epoch": 408.24778761061947,
      "grad_norm": 1.1245861060160678e-06,
      "learning_rate": 1.8651619965085967e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13215784,
      "step": 23270
    },
    {
      "epoch": 408.3362831858407,
      "grad_norm": 2.6836896722670645e-06,
      "learning_rate": 1.8642124782580433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13219048,
      "step": 23275
    },
    {
      "epoch": 408.42477876106193,
      "grad_norm": 4.403801995067624e-06,
      "learning_rate": 1.8632630580539144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13221960,
      "step": 23280
    },
    {
      "epoch": 408.5132743362832,
      "grad_norm": 2.30466457651346e-06,
      "learning_rate": 1.862313736042625e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13224824,
      "step": 23285
    },
    {
      "epoch": 408.60176991150445,
      "grad_norm": 7.389261895696109e-07,
      "learning_rate": 1.8613645123705703e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13227688,
      "step": 23290
    },
    {
      "epoch": 408.69026548672565,
      "grad_norm": 7.864285294090223e-07,
      "learning_rate": 1.8604153871841328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13230552,
      "step": 23295
    },
    {
      "epoch": 408.7787610619469,
      "grad_norm": 6.977564908083878e-07,
      "learning_rate": 1.859466360629682e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13233128,
      "step": 23300
    },
    {
      "epoch": 408.86725663716817,
      "grad_norm": 7.549086831204477e-07,
      "learning_rate": 1.8585174328535666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13236200,
      "step": 23305
    },
    {
      "epoch": 408.95575221238937,
      "grad_norm": 1.1852471288875677e-06,
      "learning_rate": 1.857568604002124e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13239048,
      "step": 23310
    },
    {
      "epoch": 409.0353982300885,
      "grad_norm": 2.917766778409714e-06,
      "learning_rate": 1.8566198742216774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13241352,
      "step": 23315
    },
    {
      "epoch": 409.12389380530976,
      "grad_norm": 2.263724354634178e-06,
      "learning_rate": 1.85567124365853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13244296,
      "step": 23320
    },
    {
      "epoch": 409.21238938053096,
      "grad_norm": 2.5166225441353163e-06,
      "learning_rate": 1.854722712458975e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13247288,
      "step": 23325
    },
    {
      "epoch": 409.3008849557522,
      "grad_norm": 1.8447452703185263e-06,
      "learning_rate": 1.853774280769286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13250040,
      "step": 23330
    },
    {
      "epoch": 409.3893805309734,
      "grad_norm": 1.7839985275713843e-06,
      "learning_rate": 1.852825948735724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13252664,
      "step": 23335
    },
    {
      "epoch": 409.4778761061947,
      "grad_norm": 3.613811941249878e-06,
      "learning_rate": 1.851877716504534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13255512,
      "step": 23340
    },
    {
      "epoch": 409.56637168141594,
      "grad_norm": 2.0383461105666356e-06,
      "learning_rate": 1.8509295842219448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13258952,
      "step": 23345
    },
    {
      "epoch": 409.65486725663715,
      "grad_norm": 4.989328317606123e-06,
      "learning_rate": 1.8499815520341697e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13262264,
      "step": 23350
    },
    {
      "epoch": 409.7433628318584,
      "grad_norm": 1.1939608839384164e-06,
      "learning_rate": 1.8490336200874094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13264920,
      "step": 23355
    },
    {
      "epoch": 409.83185840707966,
      "grad_norm": 1.0794798299684771e-06,
      "learning_rate": 1.848085788527844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13267336,
      "step": 23360
    },
    {
      "epoch": 409.92035398230087,
      "grad_norm": 1.4465806543739745e-06,
      "learning_rate": 1.847138057501644e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13270216,
      "step": 23365
    },
    {
      "epoch": 410.0,
      "grad_norm": 5.686160307050159e-07,
      "learning_rate": 1.8461904271549582e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13272776,
      "step": 23370
    },
    {
      "epoch": 410.08849557522126,
      "grad_norm": 1.8424811969453003e-06,
      "learning_rate": 1.845242897633926e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13275736,
      "step": 23375
    },
    {
      "epoch": 410.17699115044246,
      "grad_norm": 1.0475798717379803e-06,
      "learning_rate": 1.844295469084667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13278712,
      "step": 23380
    },
    {
      "epoch": 410.2654867256637,
      "grad_norm": 1.4002534953760915e-06,
      "learning_rate": 1.843348141653286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13281368,
      "step": 23385
    },
    {
      "epoch": 410.353982300885,
      "grad_norm": 1.0629413509377628e-06,
      "learning_rate": 1.842400915485874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13284088,
      "step": 23390
    },
    {
      "epoch": 410.4424778761062,
      "grad_norm": 8.669269959682424e-07,
      "learning_rate": 1.8414537907285053e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13287048,
      "step": 23395
    },
    {
      "epoch": 410.53097345132744,
      "grad_norm": 1.2439638794603525e-06,
      "learning_rate": 1.840506767527237e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13289752,
      "step": 23400
    },
    {
      "epoch": 410.53097345132744,
      "eval_loss": 0.5085816979408264,
      "eval_runtime": 1.0657,
      "eval_samples_per_second": 23.458,
      "eval_steps_per_second": 12.198,
      "num_input_tokens_seen": 13289752,
      "step": 23400
    },
    {
      "epoch": 410.6194690265487,
      "grad_norm": 3.61501588486135e-06,
      "learning_rate": 1.8395598460281137e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13292584,
      "step": 23405
    },
    {
      "epoch": 410.7079646017699,
      "grad_norm": 7.967166766320588e-07,
      "learning_rate": 1.838613026377161e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13295704,
      "step": 23410
    },
    {
      "epoch": 410.79646017699116,
      "grad_norm": 3.438787871345994e-06,
      "learning_rate": 1.8376663087203917e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13298600,
      "step": 23415
    },
    {
      "epoch": 410.88495575221236,
      "grad_norm": 2.5190897758875508e-06,
      "learning_rate": 1.8367196932038014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13301912,
      "step": 23420
    },
    {
      "epoch": 410.9734513274336,
      "grad_norm": 7.74333511799341e-06,
      "learning_rate": 1.8357731799733686e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13304472,
      "step": 23425
    },
    {
      "epoch": 411.05309734513276,
      "grad_norm": 5.801222755508206e-07,
      "learning_rate": 1.8348267691750586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13307080,
      "step": 23430
    },
    {
      "epoch": 411.14159292035396,
      "grad_norm": 1.30218700178375e-06,
      "learning_rate": 1.833880460954821e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13310152,
      "step": 23435
    },
    {
      "epoch": 411.2300884955752,
      "grad_norm": 1.2947436971444404e-06,
      "learning_rate": 1.8329342554585866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13312840,
      "step": 23440
    },
    {
      "epoch": 411.3185840707965,
      "grad_norm": 5.34400305696181e-06,
      "learning_rate": 1.8319881528322735e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13316168,
      "step": 23445
    },
    {
      "epoch": 411.4070796460177,
      "grad_norm": 9.54574375100492e-07,
      "learning_rate": 1.8310421532217815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13319304,
      "step": 23450
    },
    {
      "epoch": 411.49557522123894,
      "grad_norm": 1.3224064332462149e-06,
      "learning_rate": 1.8300962567729958e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13321912,
      "step": 23455
    },
    {
      "epoch": 411.5840707964602,
      "grad_norm": 4.417475338414079e-06,
      "learning_rate": 1.8291504636317866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13324232,
      "step": 23460
    },
    {
      "epoch": 411.6725663716814,
      "grad_norm": 2.854363856386044e-06,
      "learning_rate": 1.8282047739440055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13327016,
      "step": 23465
    },
    {
      "epoch": 411.76106194690266,
      "grad_norm": 3.173348886775784e-06,
      "learning_rate": 1.8272591878554903e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13329784,
      "step": 23470
    },
    {
      "epoch": 411.8495575221239,
      "grad_norm": 8.481652571390441e-07,
      "learning_rate": 1.8263137055120638e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13332616,
      "step": 23475
    },
    {
      "epoch": 411.9380530973451,
      "grad_norm": 5.065410277893534e-07,
      "learning_rate": 1.8253683270595295e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13335736,
      "step": 23480
    },
    {
      "epoch": 412.01769911504425,
      "grad_norm": 1.0313268603567849e-06,
      "learning_rate": 1.824423052643677e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13338224,
      "step": 23485
    },
    {
      "epoch": 412.1061946902655,
      "grad_norm": 1.7834886421042029e-06,
      "learning_rate": 1.82347788241028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13341264,
      "step": 23490
    },
    {
      "epoch": 412.1946902654867,
      "grad_norm": 1.760229679348413e-06,
      "learning_rate": 1.8225328165050942e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13344048,
      "step": 23495
    },
    {
      "epoch": 412.283185840708,
      "grad_norm": 8.950841561272682e-07,
      "learning_rate": 1.821587855073863e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13346416,
      "step": 23500
    },
    {
      "epoch": 412.37168141592923,
      "grad_norm": 5.851482001162367e-06,
      "learning_rate": 1.8206429982623086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13349984,
      "step": 23505
    },
    {
      "epoch": 412.46017699115043,
      "grad_norm": 1.908619879031903e-06,
      "learning_rate": 1.8196982462161416e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13352736,
      "step": 23510
    },
    {
      "epoch": 412.5486725663717,
      "grad_norm": 1.5992902717698598e-06,
      "learning_rate": 1.818753599081055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13355136,
      "step": 23515
    },
    {
      "epoch": 412.6371681415929,
      "grad_norm": 1.065754190676671e-06,
      "learning_rate": 1.817809057002724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13358032,
      "step": 23520
    },
    {
      "epoch": 412.72566371681415,
      "grad_norm": 1.1876481949002482e-06,
      "learning_rate": 1.8168646201268096e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13360528,
      "step": 23525
    },
    {
      "epoch": 412.8141592920354,
      "grad_norm": 2.347320787521312e-06,
      "learning_rate": 1.8159202885989557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13363904,
      "step": 23530
    },
    {
      "epoch": 412.9026548672566,
      "grad_norm": 2.8202557587064803e-06,
      "learning_rate": 1.814976062564789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13366576,
      "step": 23535
    },
    {
      "epoch": 412.9911504424779,
      "grad_norm": 3.838842076220317e-06,
      "learning_rate": 1.8140319421699234e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13369584,
      "step": 23540
    },
    {
      "epoch": 413.070796460177,
      "grad_norm": 3.0590451842726907e-06,
      "learning_rate": 1.8130879275599515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13371928,
      "step": 23545
    },
    {
      "epoch": 413.1592920353982,
      "grad_norm": 1.8705159163801e-06,
      "learning_rate": 1.8121440188804544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13374888,
      "step": 23550
    },
    {
      "epoch": 413.24778761061947,
      "grad_norm": 3.922223640984157e-06,
      "learning_rate": 1.811200216276993e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13377848,
      "step": 23555
    },
    {
      "epoch": 413.3362831858407,
      "grad_norm": 2.0233237592037767e-06,
      "learning_rate": 1.810256519895115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13380408,
      "step": 23560
    },
    {
      "epoch": 413.42477876106193,
      "grad_norm": 5.977739874651888e-06,
      "learning_rate": 1.8093129298803494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13383128,
      "step": 23565
    },
    {
      "epoch": 413.5132743362832,
      "grad_norm": 1.3068647604086436e-06,
      "learning_rate": 1.808369446378209e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13386104,
      "step": 23570
    },
    {
      "epoch": 413.60176991150445,
      "grad_norm": 1.0755246648841421e-06,
      "learning_rate": 1.8074260695341914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13389144,
      "step": 23575
    },
    {
      "epoch": 413.69026548672565,
      "grad_norm": 7.753723139103386e-07,
      "learning_rate": 1.8064827994937782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13392312,
      "step": 23580
    },
    {
      "epoch": 413.7787610619469,
      "grad_norm": 1.2985356079298072e-06,
      "learning_rate": 1.8055396364024317e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13395096,
      "step": 23585
    },
    {
      "epoch": 413.86725663716817,
      "grad_norm": 1.9835347302432638e-06,
      "learning_rate": 1.804596580405601e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13398008,
      "step": 23590
    },
    {
      "epoch": 413.95575221238937,
      "grad_norm": 3.0213807349355193e-06,
      "learning_rate": 1.8036536316487174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13401112,
      "step": 23595
    },
    {
      "epoch": 414.0353982300885,
      "grad_norm": 1.1468927141322638e-06,
      "learning_rate": 1.802710790277193e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13403848,
      "step": 23600
    },
    {
      "epoch": 414.0353982300885,
      "eval_loss": 0.4857552647590637,
      "eval_runtime": 1.0588,
      "eval_samples_per_second": 23.611,
      "eval_steps_per_second": 12.278,
      "num_input_tokens_seen": 13403848,
      "step": 23600
    },
    {
      "epoch": 414.12389380530976,
      "grad_norm": 1.4702327462146059e-06,
      "learning_rate": 1.801768056436429e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13406680,
      "step": 23605
    },
    {
      "epoch": 414.21238938053096,
      "grad_norm": 2.7363189474272076e-06,
      "learning_rate": 1.8008254302718035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13409160,
      "step": 23610
    },
    {
      "epoch": 414.3008849557522,
      "grad_norm": 2.349073156437953e-06,
      "learning_rate": 1.7998829119286837e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13411752,
      "step": 23615
    },
    {
      "epoch": 414.3893805309734,
      "grad_norm": 1.2500671573434374e-06,
      "learning_rate": 1.798940501552418e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13413896,
      "step": 23620
    },
    {
      "epoch": 414.4778761061947,
      "grad_norm": 5.316167516866699e-06,
      "learning_rate": 1.797998199288336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13417048,
      "step": 23625
    },
    {
      "epoch": 414.56637168141594,
      "grad_norm": 2.0722507088066777e-06,
      "learning_rate": 1.7970560052817543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13419896,
      "step": 23630
    },
    {
      "epoch": 414.65486725663715,
      "grad_norm": 4.038126462546643e-06,
      "learning_rate": 1.7961139196779702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13422920,
      "step": 23635
    },
    {
      "epoch": 414.7433628318584,
      "grad_norm": 1.777973011485301e-06,
      "learning_rate": 1.7951719426222647e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13426424,
      "step": 23640
    },
    {
      "epoch": 414.83185840707966,
      "grad_norm": 2.984578259201953e-06,
      "learning_rate": 1.794230074259904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13429304,
      "step": 23645
    },
    {
      "epoch": 414.92035398230087,
      "grad_norm": 3.6621311210183194e-06,
      "learning_rate": 1.7932883147361336e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13432536,
      "step": 23650
    },
    {
      "epoch": 415.0,
      "grad_norm": 2.736116584856063e-06,
      "learning_rate": 1.7923466641961865e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13435104,
      "step": 23655
    },
    {
      "epoch": 415.08849557522126,
      "grad_norm": 9.776815204531886e-06,
      "learning_rate": 1.791405122785278e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13437776,
      "step": 23660
    },
    {
      "epoch": 415.17699115044246,
      "grad_norm": 1.584068968440988e-06,
      "learning_rate": 1.7904636906486037e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13440528,
      "step": 23665
    },
    {
      "epoch": 415.2654867256637,
      "grad_norm": 2.0428676634764997e-06,
      "learning_rate": 1.7895223679313448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13442944,
      "step": 23670
    },
    {
      "epoch": 415.353982300885,
      "grad_norm": 2.589305950095877e-06,
      "learning_rate": 1.7885811547786653e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13445856,
      "step": 23675
    },
    {
      "epoch": 415.4424778761062,
      "grad_norm": 9.413490715814987e-07,
      "learning_rate": 1.7876400513357115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13449424,
      "step": 23680
    },
    {
      "epoch": 415.53097345132744,
      "grad_norm": 6.820781663918751e-07,
      "learning_rate": 1.7866990577476146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13452608,
      "step": 23685
    },
    {
      "epoch": 415.6194690265487,
      "grad_norm": 9.24022970139049e-07,
      "learning_rate": 1.7857581741594863e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13455776,
      "step": 23690
    },
    {
      "epoch": 415.7079646017699,
      "grad_norm": 8.599495799899159e-07,
      "learning_rate": 1.7848174007164237e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13459024,
      "step": 23695
    },
    {
      "epoch": 415.79646017699116,
      "grad_norm": 1.745924464557902e-06,
      "learning_rate": 1.7838767375635052e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13461776,
      "step": 23700
    },
    {
      "epoch": 415.88495575221236,
      "grad_norm": 1.3890613672629115e-06,
      "learning_rate": 1.782936184845793e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13464416,
      "step": 23705
    },
    {
      "epoch": 415.9734513274336,
      "grad_norm": 2.7534174478205387e-06,
      "learning_rate": 1.7819957427083334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13467072,
      "step": 23710
    },
    {
      "epoch": 416.05309734513276,
      "grad_norm": 7.761572305753361e-06,
      "learning_rate": 1.7810554112961516e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13469936,
      "step": 23715
    },
    {
      "epoch": 416.14159292035396,
      "grad_norm": 1.0326749588784878e-06,
      "learning_rate": 1.7801151907542607e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13472528,
      "step": 23720
    },
    {
      "epoch": 416.2300884955752,
      "grad_norm": 1.4454839174504741e-06,
      "learning_rate": 1.7791750812276547e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13475264,
      "step": 23725
    },
    {
      "epoch": 416.3185840707965,
      "grad_norm": 1.2660036645684158e-06,
      "learning_rate": 1.778235082861309e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13478288,
      "step": 23730
    },
    {
      "epoch": 416.4070796460177,
      "grad_norm": 8.937573170442192e-07,
      "learning_rate": 1.777295195800184e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13481312,
      "step": 23735
    },
    {
      "epoch": 416.49557522123894,
      "grad_norm": 1.7386607851221925e-06,
      "learning_rate": 1.7763554201892215e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13484352,
      "step": 23740
    },
    {
      "epoch": 416.5840707964602,
      "grad_norm": 6.406043553397467e-07,
      "learning_rate": 1.7754157561733476e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13487712,
      "step": 23745
    },
    {
      "epoch": 416.6725663716814,
      "grad_norm": 1.1096309435743024e-06,
      "learning_rate": 1.7744762038974702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13490576,
      "step": 23750
    },
    {
      "epoch": 416.76106194690266,
      "grad_norm": 1.2957416402059607e-06,
      "learning_rate": 1.7735367635064788e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13493104,
      "step": 23755
    },
    {
      "epoch": 416.8495575221239,
      "grad_norm": 4.042884029331617e-06,
      "learning_rate": 1.7725974351452474e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13495792,
      "step": 23760
    },
    {
      "epoch": 416.9380530973451,
      "grad_norm": 1.0686250107028172e-06,
      "learning_rate": 1.771658218958634e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13498688,
      "step": 23765
    },
    {
      "epoch": 417.01769911504425,
      "grad_norm": 2.326136836927617e-06,
      "learning_rate": 1.770719115091475e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13500624,
      "step": 23770
    },
    {
      "epoch": 417.1061946902655,
      "grad_norm": 1.0341671440983191e-05,
      "learning_rate": 1.7697801236885935e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13503808,
      "step": 23775
    },
    {
      "epoch": 417.1946902654867,
      "grad_norm": 1.1650794476736337e-05,
      "learning_rate": 1.7688412448947944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13507136,
      "step": 23780
    },
    {
      "epoch": 417.283185840708,
      "grad_norm": 4.136436473345384e-07,
      "learning_rate": 1.767902478854862e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13509728,
      "step": 23785
    },
    {
      "epoch": 417.37168141592923,
      "grad_norm": 6.370703431457514e-07,
      "learning_rate": 1.766963825713569e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13512400,
      "step": 23790
    },
    {
      "epoch": 417.46017699115043,
      "grad_norm": 2.331347786821425e-06,
      "learning_rate": 1.766025285615665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13515904,
      "step": 23795
    },
    {
      "epoch": 417.5486725663717,
      "grad_norm": 6.143058044472127e-07,
      "learning_rate": 1.7650868587058854e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13518496,
      "step": 23800
    },
    {
      "epoch": 417.5486725663717,
      "eval_loss": 0.5352886915206909,
      "eval_runtime": 1.0676,
      "eval_samples_per_second": 23.416,
      "eval_steps_per_second": 12.176,
      "num_input_tokens_seen": 13518496,
      "step": 23800
    },
    {
      "epoch": 417.6371681415929,
      "grad_norm": 9.586101441527717e-07,
      "learning_rate": 1.7641485451289484e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13521296,
      "step": 23805
    },
    {
      "epoch": 417.72566371681415,
      "grad_norm": 9.03472823665652e-07,
      "learning_rate": 1.7632103450295534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13523728,
      "step": 23810
    },
    {
      "epoch": 417.8141592920354,
      "grad_norm": 7.322683472921199e-07,
      "learning_rate": 1.762272258552381e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13526624,
      "step": 23815
    },
    {
      "epoch": 417.9026548672566,
      "grad_norm": 1.049432398758654e-06,
      "learning_rate": 1.7613342858420988e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13529888,
      "step": 23820
    },
    {
      "epoch": 417.9911504424779,
      "grad_norm": 1.234100409419625e-06,
      "learning_rate": 1.760396427043351e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13532272,
      "step": 23825
    },
    {
      "epoch": 418.070796460177,
      "grad_norm": 1.063686568159028e-06,
      "learning_rate": 1.7594586823007696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13535136,
      "step": 23830
    },
    {
      "epoch": 418.1592920353982,
      "grad_norm": 8.850796007209283e-07,
      "learning_rate": 1.7585210517589646e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13538032,
      "step": 23835
    },
    {
      "epoch": 418.24778761061947,
      "grad_norm": 1.4844956695014844e-06,
      "learning_rate": 1.7575835355625314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13540816,
      "step": 23840
    },
    {
      "epoch": 418.3362831858407,
      "grad_norm": 1.5663404155930039e-06,
      "learning_rate": 1.756646133856048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13543776,
      "step": 23845
    },
    {
      "epoch": 418.42477876106193,
      "grad_norm": 7.596941031806637e-07,
      "learning_rate": 1.7557088467840714e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13546464,
      "step": 23850
    },
    {
      "epoch": 418.5132743362832,
      "grad_norm": 1.7474661945016123e-05,
      "learning_rate": 1.7547716744911438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13549136,
      "step": 23855
    },
    {
      "epoch": 418.60176991150445,
      "grad_norm": 8.410667646785441e-07,
      "learning_rate": 1.7538346171217902e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13551552,
      "step": 23860
    },
    {
      "epoch": 418.69026548672565,
      "grad_norm": 5.49734863852791e-07,
      "learning_rate": 1.7528976748205146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13554240,
      "step": 23865
    },
    {
      "epoch": 418.7787610619469,
      "grad_norm": 1.0287782288287417e-06,
      "learning_rate": 1.751960847731807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13557056,
      "step": 23870
    },
    {
      "epoch": 418.86725663716817,
      "grad_norm": 5.80174798869848e-07,
      "learning_rate": 1.7510241360001362e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13560880,
      "step": 23875
    },
    {
      "epoch": 418.95575221238937,
      "grad_norm": 2.593761337266187e-06,
      "learning_rate": 1.7500875397699562e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13563792,
      "step": 23880
    },
    {
      "epoch": 419.0353982300885,
      "grad_norm": 1.4792015008424642e-06,
      "learning_rate": 1.7491510591857015e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13566536,
      "step": 23885
    },
    {
      "epoch": 419.12389380530976,
      "grad_norm": 1.2081727618351579e-06,
      "learning_rate": 1.7482146943917896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13569256,
      "step": 23890
    },
    {
      "epoch": 419.21238938053096,
      "grad_norm": 1.915943812491605e-06,
      "learning_rate": 1.7472784455326185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13572216,
      "step": 23895
    },
    {
      "epoch": 419.3008849557522,
      "grad_norm": 9.63474485615734e-06,
      "learning_rate": 1.746342312752572e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13574984,
      "step": 23900
    },
    {
      "epoch": 419.3893805309734,
      "grad_norm": 7.35669289042562e-07,
      "learning_rate": 1.74540629619601e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13578040,
      "step": 23905
    },
    {
      "epoch": 419.4778761061947,
      "grad_norm": 1.1026037327610538e-06,
      "learning_rate": 1.7444703960072815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13581080,
      "step": 23910
    },
    {
      "epoch": 419.56637168141594,
      "grad_norm": 9.74721501734166e-07,
      "learning_rate": 1.7435346123307118e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13584104,
      "step": 23915
    },
    {
      "epoch": 419.65486725663715,
      "grad_norm": 5.39653569830989e-07,
      "learning_rate": 1.742598945310611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13586840,
      "step": 23920
    },
    {
      "epoch": 419.7433628318584,
      "grad_norm": 8.942301406023034e-07,
      "learning_rate": 1.741663395091272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13589448,
      "step": 23925
    },
    {
      "epoch": 419.83185840707966,
      "grad_norm": 1.3534656773117604e-06,
      "learning_rate": 1.7407279618169657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13592264,
      "step": 23930
    },
    {
      "epoch": 419.92035398230087,
      "grad_norm": 6.922055945324246e-06,
      "learning_rate": 1.73979264563195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13595080,
      "step": 23935
    },
    {
      "epoch": 420.0,
      "grad_norm": 2.0247045995347435e-06,
      "learning_rate": 1.7388574466804625e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13597592,
      "step": 23940
    },
    {
      "epoch": 420.08849557522126,
      "grad_norm": 1.1908844044228317e-06,
      "learning_rate": 1.7379223651067207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13600248,
      "step": 23945
    },
    {
      "epoch": 420.17699115044246,
      "grad_norm": 8.358568265975919e-06,
      "learning_rate": 1.736987401054928e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13603144,
      "step": 23950
    },
    {
      "epoch": 420.2654867256637,
      "grad_norm": 1.5122247987164883e-06,
      "learning_rate": 1.736052554669266e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13606296,
      "step": 23955
    },
    {
      "epoch": 420.353982300885,
      "grad_norm": 1.1291514283584547e-06,
      "learning_rate": 1.7351178260939007e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13609592,
      "step": 23960
    },
    {
      "epoch": 420.4424778761062,
      "grad_norm": 1.2396754982546554e-06,
      "learning_rate": 1.7341832154729794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13612024,
      "step": 23965
    },
    {
      "epoch": 420.53097345132744,
      "grad_norm": 1.7911884242494125e-06,
      "learning_rate": 1.7332487229506286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13615128,
      "step": 23970
    },
    {
      "epoch": 420.6194690265487,
      "grad_norm": 1.7784844885682105e-06,
      "learning_rate": 1.732314348670961e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13618024,
      "step": 23975
    },
    {
      "epoch": 420.7079646017699,
      "grad_norm": 6.04233628109796e-06,
      "learning_rate": 1.7313800927780686e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13621128,
      "step": 23980
    },
    {
      "epoch": 420.79646017699116,
      "grad_norm": 4.269141300028423e-06,
      "learning_rate": 1.7304459554160245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13624168,
      "step": 23985
    },
    {
      "epoch": 420.88495575221236,
      "grad_norm": 2.6619800337357447e-06,
      "learning_rate": 1.7295119367288853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13626616,
      "step": 23990
    },
    {
      "epoch": 420.9734513274336,
      "grad_norm": 1.3246168464320363e-06,
      "learning_rate": 1.728578036860688e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13629272,
      "step": 23995
    },
    {
      "epoch": 421.05309734513276,
      "grad_norm": 3.985278453910723e-06,
      "learning_rate": 1.7276442559554513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13631704,
      "step": 24000
    },
    {
      "epoch": 421.05309734513276,
      "eval_loss": 0.49583497643470764,
      "eval_runtime": 1.0716,
      "eval_samples_per_second": 23.329,
      "eval_steps_per_second": 12.131,
      "num_input_tokens_seen": 13631704,
      "step": 24000
    },
    {
      "epoch": 421.14159292035396,
      "grad_norm": 8.406973392993677e-07,
      "learning_rate": 1.726710594157177e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13634808,
      "step": 24005
    },
    {
      "epoch": 421.2300884955752,
      "grad_norm": 2.6902180252363905e-06,
      "learning_rate": 1.725777051609846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13637448,
      "step": 24010
    },
    {
      "epoch": 421.3185840707965,
      "grad_norm": 9.167666576104239e-06,
      "learning_rate": 1.7248436284574228e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13640520,
      "step": 24015
    },
    {
      "epoch": 421.4070796460177,
      "grad_norm": 2.00452700482856e-06,
      "learning_rate": 1.723910324843855e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13643064,
      "step": 24020
    },
    {
      "epoch": 421.49557522123894,
      "grad_norm": 1.4062749187360168e-06,
      "learning_rate": 1.722977140913067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13646296,
      "step": 24025
    },
    {
      "epoch": 421.5840707964602,
      "grad_norm": 1.189801650980371e-06,
      "learning_rate": 1.7220440768089688e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13649128,
      "step": 24030
    },
    {
      "epoch": 421.6725663716814,
      "grad_norm": 2.5092192572628846e-06,
      "learning_rate": 1.7211111326754505e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13651688,
      "step": 24035
    },
    {
      "epoch": 421.76106194690266,
      "grad_norm": 2.3390741716866614e-06,
      "learning_rate": 1.720178308656383e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13654744,
      "step": 24040
    },
    {
      "epoch": 421.8495575221239,
      "grad_norm": 6.643195774813648e-06,
      "learning_rate": 1.719245604895621e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13657544,
      "step": 24045
    },
    {
      "epoch": 421.9380530973451,
      "grad_norm": 8.497192425238609e-07,
      "learning_rate": 1.7183130215369972e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13660504,
      "step": 24050
    },
    {
      "epoch": 422.01769911504425,
      "grad_norm": 1.9094609342573676e-06,
      "learning_rate": 1.7173805587243292e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13662744,
      "step": 24055
    },
    {
      "epoch": 422.1061946902655,
      "grad_norm": 6.956496690690983e-06,
      "learning_rate": 1.7164482166014147e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13665576,
      "step": 24060
    },
    {
      "epoch": 422.1946902654867,
      "grad_norm": 1.292658225793275e-06,
      "learning_rate": 1.7155159953120313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13668056,
      "step": 24065
    },
    {
      "epoch": 422.283185840708,
      "grad_norm": 2.1519692836591275e-06,
      "learning_rate": 1.714583894999941e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13670872,
      "step": 24070
    },
    {
      "epoch": 422.37168141592923,
      "grad_norm": 1.4788835187573568e-06,
      "learning_rate": 1.7136519158088826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13674632,
      "step": 24075
    },
    {
      "epoch": 422.46017699115043,
      "grad_norm": 8.012464718376577e-07,
      "learning_rate": 1.712720057882581e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13677272,
      "step": 24080
    },
    {
      "epoch": 422.5486725663717,
      "grad_norm": 1.6286288655464887e-06,
      "learning_rate": 1.7117883213647413e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13680264,
      "step": 24085
    },
    {
      "epoch": 422.6371681415929,
      "grad_norm": 2.6893044378084596e-06,
      "learning_rate": 1.710856706399046e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13683432,
      "step": 24090
    },
    {
      "epoch": 422.72566371681415,
      "grad_norm": 3.2359077977162087e-06,
      "learning_rate": 1.7099252131291648e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13686168,
      "step": 24095
    },
    {
      "epoch": 422.8141592920354,
      "grad_norm": 1.6132710243255133e-06,
      "learning_rate": 1.708993841698744e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13688984,
      "step": 24100
    },
    {
      "epoch": 422.9026548672566,
      "grad_norm": 1.2670464002439985e-06,
      "learning_rate": 1.7080625922514132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13691880,
      "step": 24105
    },
    {
      "epoch": 422.9911504424779,
      "grad_norm": 1.7501914044260047e-05,
      "learning_rate": 1.7071314649307836e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13694440,
      "step": 24110
    },
    {
      "epoch": 423.070796460177,
      "grad_norm": 2.4492726424796274e-06,
      "learning_rate": 1.7062004598804448e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13696576,
      "step": 24115
    },
    {
      "epoch": 423.1592920353982,
      "grad_norm": 4.256809006619733e-06,
      "learning_rate": 1.7052695772439702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13699392,
      "step": 24120
    },
    {
      "epoch": 423.24778761061947,
      "grad_norm": 3.7660634006897453e-06,
      "learning_rate": 1.7043388171649154e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13702160,
      "step": 24125
    },
    {
      "epoch": 423.3362831858407,
      "grad_norm": 8.910963629205071e-07,
      "learning_rate": 1.7034081797868127e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13705200,
      "step": 24130
    },
    {
      "epoch": 423.42477876106193,
      "grad_norm": 2.9555703804362565e-06,
      "learning_rate": 1.70247766525318e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13707872,
      "step": 24135
    },
    {
      "epoch": 423.5132743362832,
      "grad_norm": 3.4121705994039075e-06,
      "learning_rate": 1.701547273707514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13710896,
      "step": 24140
    },
    {
      "epoch": 423.60176991150445,
      "grad_norm": 1.0119325679625035e-06,
      "learning_rate": 1.7006170052932916e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13713632,
      "step": 24145
    },
    {
      "epoch": 423.69026548672565,
      "grad_norm": 7.58254827815108e-07,
      "learning_rate": 1.6996868601539735e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13716848,
      "step": 24150
    },
    {
      "epoch": 423.7787610619469,
      "grad_norm": 9.319676905761298e-07,
      "learning_rate": 1.6987568384329977e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13719632,
      "step": 24155
    },
    {
      "epoch": 423.86725663716817,
      "grad_norm": 3.595997213778901e-06,
      "learning_rate": 1.6978269402737866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13722576,
      "step": 24160
    },
    {
      "epoch": 423.95575221238937,
      "grad_norm": 1.3696160294784931e-06,
      "learning_rate": 1.696897165819743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13725440,
      "step": 24165
    },
    {
      "epoch": 424.0353982300885,
      "grad_norm": 3.823710812866921e-06,
      "learning_rate": 1.6959675152142487e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13728208,
      "step": 24170
    },
    {
      "epoch": 424.12389380530976,
      "grad_norm": 2.4431112706224667e-06,
      "learning_rate": 1.6950379886006667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13731456,
      "step": 24175
    },
    {
      "epoch": 424.21238938053096,
      "grad_norm": 2.578751718829153e-06,
      "learning_rate": 1.6941085861223438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13734144,
      "step": 24180
    },
    {
      "epoch": 424.3008849557522,
      "grad_norm": 1.226781819241296e-06,
      "learning_rate": 1.6931793079226034e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13736816,
      "step": 24185
    },
    {
      "epoch": 424.3893805309734,
      "grad_norm": 1.1118441989310668e-06,
      "learning_rate": 1.692250154144754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13739264,
      "step": 24190
    },
    {
      "epoch": 424.4778761061947,
      "grad_norm": 9.555992619425524e-07,
      "learning_rate": 1.6913211249320807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13742176,
      "step": 24195
    },
    {
      "epoch": 424.56637168141594,
      "grad_norm": 1.0115005579791614e-06,
      "learning_rate": 1.6903922204278522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13745200,
      "step": 24200
    },
    {
      "epoch": 424.56637168141594,
      "eval_loss": 0.49358347058296204,
      "eval_runtime": 1.0796,
      "eval_samples_per_second": 23.156,
      "eval_steps_per_second": 12.041,
      "num_input_tokens_seen": 13745200,
      "step": 24200
    },
    {
      "epoch": 424.65486725663715,
      "grad_norm": 1.7055592707038159e-06,
      "learning_rate": 1.6894634407753186e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13748144,
      "step": 24205
    },
    {
      "epoch": 424.7433628318584,
      "grad_norm": 1.0927784614978009e-06,
      "learning_rate": 1.6885347861177077e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13751504,
      "step": 24210
    },
    {
      "epoch": 424.83185840707966,
      "grad_norm": 3.881550583173521e-06,
      "learning_rate": 1.6876062565982298e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13754336,
      "step": 24215
    },
    {
      "epoch": 424.92035398230087,
      "grad_norm": 4.395184532768326e-06,
      "learning_rate": 1.6866778523600774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13757184,
      "step": 24220
    },
    {
      "epoch": 425.0,
      "grad_norm": 7.491780706914142e-06,
      "learning_rate": 1.6857495735464195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13759352,
      "step": 24225
    },
    {
      "epoch": 425.08849557522126,
      "grad_norm": 2.962990492960671e-06,
      "learning_rate": 1.6848214203004115e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13762296,
      "step": 24230
    },
    {
      "epoch": 425.17699115044246,
      "grad_norm": 6.962067345739342e-06,
      "learning_rate": 1.6838933927651835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13765320,
      "step": 24235
    },
    {
      "epoch": 425.2654867256637,
      "grad_norm": 1.1171513278895873e-06,
      "learning_rate": 1.6829654910838506e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13768184,
      "step": 24240
    },
    {
      "epoch": 425.353982300885,
      "grad_norm": 1.275399654332432e-06,
      "learning_rate": 1.6820377153995065e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13770888,
      "step": 24245
    },
    {
      "epoch": 425.4424778761062,
      "grad_norm": 6.105977945480845e-07,
      "learning_rate": 1.681110065855226e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13773432,
      "step": 24250
    },
    {
      "epoch": 425.53097345132744,
      "grad_norm": 1.0341749430153868e-06,
      "learning_rate": 1.6801825425940642e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13776440,
      "step": 24255
    },
    {
      "epoch": 425.6194690265487,
      "grad_norm": 8.394205224249163e-07,
      "learning_rate": 1.679255145759056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13779320,
      "step": 24260
    },
    {
      "epoch": 425.7079646017699,
      "grad_norm": 5.675396550941514e-06,
      "learning_rate": 1.6783278754932187e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13782168,
      "step": 24265
    },
    {
      "epoch": 425.79646017699116,
      "grad_norm": 3.061396455450449e-06,
      "learning_rate": 1.6774007319395496e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13785288,
      "step": 24270
    },
    {
      "epoch": 425.88495575221236,
      "grad_norm": 4.246197022439446e-06,
      "learning_rate": 1.6764737152410243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13788056,
      "step": 24275
    },
    {
      "epoch": 425.9734513274336,
      "grad_norm": 1.5542626670139725e-06,
      "learning_rate": 1.6755468255406016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13790920,
      "step": 24280
    },
    {
      "epoch": 426.05309734513276,
      "grad_norm": 7.014898528723279e-06,
      "learning_rate": 1.674620062981219e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13793464,
      "step": 24285
    },
    {
      "epoch": 426.14159292035396,
      "grad_norm": 5.510081336979056e-06,
      "learning_rate": 1.6736934277057947e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13796424,
      "step": 24290
    },
    {
      "epoch": 426.2300884955752,
      "grad_norm": 1.1415747849241598e-06,
      "learning_rate": 1.6727669198572286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13799064,
      "step": 24295
    },
    {
      "epoch": 426.3185840707965,
      "grad_norm": 3.0577143661503214e-06,
      "learning_rate": 1.6718405395783984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13802408,
      "step": 24300
    },
    {
      "epoch": 426.4070796460177,
      "grad_norm": 3.787782134168083e-06,
      "learning_rate": 1.6709142870121643e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13804920,
      "step": 24305
    },
    {
      "epoch": 426.49557522123894,
      "grad_norm": 2.7576520551519934e-06,
      "learning_rate": 1.669988162301367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13808008,
      "step": 24310
    },
    {
      "epoch": 426.5840707964602,
      "grad_norm": 1.577579382683325e-06,
      "learning_rate": 1.6690621655888243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13811160,
      "step": 24315
    },
    {
      "epoch": 426.6725663716814,
      "grad_norm": 1.2895628742626286e-06,
      "learning_rate": 1.6681362970173386e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13814472,
      "step": 24320
    },
    {
      "epoch": 426.76106194690266,
      "grad_norm": 2.889762754421099e-06,
      "learning_rate": 1.6672105567296904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13817528,
      "step": 24325
    },
    {
      "epoch": 426.8495575221239,
      "grad_norm": 1.1387251106498297e-06,
      "learning_rate": 1.666284944868639e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13819960,
      "step": 24330
    },
    {
      "epoch": 426.9380530973451,
      "grad_norm": 2.134988108082325e-06,
      "learning_rate": 1.665359461576927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13822712,
      "step": 24335
    },
    {
      "epoch": 427.01769911504425,
      "grad_norm": 1.5303902500818367e-06,
      "learning_rate": 1.6644341069972736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13825240,
      "step": 24340
    },
    {
      "epoch": 427.1061946902655,
      "grad_norm": 8.348664550794638e-07,
      "learning_rate": 1.6635088812723813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13827688,
      "step": 24345
    },
    {
      "epoch": 427.1946902654867,
      "grad_norm": 2.849028760465444e-06,
      "learning_rate": 1.6625837845449328e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13830840,
      "step": 24350
    },
    {
      "epoch": 427.283185840708,
      "grad_norm": 8.777485049904499e-07,
      "learning_rate": 1.6616588169575874e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13834264,
      "step": 24355
    },
    {
      "epoch": 427.37168141592923,
      "grad_norm": 1.643871996748203e-06,
      "learning_rate": 1.6607339786529878e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13837336,
      "step": 24360
    },
    {
      "epoch": 427.46017699115043,
      "grad_norm": 7.717748076174757e-07,
      "learning_rate": 1.659809269773756e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13840168,
      "step": 24365
    },
    {
      "epoch": 427.5486725663717,
      "grad_norm": 7.212130412881379e-07,
      "learning_rate": 1.658884690462493e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13842760,
      "step": 24370
    },
    {
      "epoch": 427.6371681415929,
      "grad_norm": 6.750630291207926e-06,
      "learning_rate": 1.6579602408617813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13845592,
      "step": 24375
    },
    {
      "epoch": 427.72566371681415,
      "grad_norm": 8.98831160611735e-07,
      "learning_rate": 1.657035921114181e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13848568,
      "step": 24380
    },
    {
      "epoch": 427.8141592920354,
      "grad_norm": 9.8607988547883e-06,
      "learning_rate": 1.656111731362236e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13851464,
      "step": 24385
    },
    {
      "epoch": 427.9026548672566,
      "grad_norm": 9.624151289244764e-07,
      "learning_rate": 1.6551876717484666e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13854744,
      "step": 24390
    },
    {
      "epoch": 427.9911504424779,
      "grad_norm": 1.0259614100505132e-06,
      "learning_rate": 1.6542637424153752e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13857256,
      "step": 24395
    },
    {
      "epoch": 428.070796460177,
      "grad_norm": 1.2096062391719897e-06,
      "learning_rate": 1.6533399435054418e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13859752,
      "step": 24400
    },
    {
      "epoch": 428.070796460177,
      "eval_loss": 0.5260993242263794,
      "eval_runtime": 1.062,
      "eval_samples_per_second": 23.54,
      "eval_steps_per_second": 12.241,
      "num_input_tokens_seen": 13859752,
      "step": 24400
    },
    {
      "epoch": 428.1592920353982,
      "grad_norm": 9.896382380247815e-07,
      "learning_rate": 1.6524162751611304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13863672,
      "step": 24405
    },
    {
      "epoch": 428.24778761061947,
      "grad_norm": 4.519103640632238e-06,
      "learning_rate": 1.6514927375248796e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13866632,
      "step": 24410
    },
    {
      "epoch": 428.3362831858407,
      "grad_norm": 1.4516801911668153e-06,
      "learning_rate": 1.6505693307391127e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13869592,
      "step": 24415
    },
    {
      "epoch": 428.42477876106193,
      "grad_norm": 3.9156052480393555e-06,
      "learning_rate": 1.6496460549462288e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13872472,
      "step": 24420
    },
    {
      "epoch": 428.5132743362832,
      "grad_norm": 1.5261305179592455e-06,
      "learning_rate": 1.6487229102886097e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13875032,
      "step": 24425
    },
    {
      "epoch": 428.60176991150445,
      "grad_norm": 8.407857308156963e-07,
      "learning_rate": 1.6477998969086155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13877624,
      "step": 24430
    },
    {
      "epoch": 428.69026548672565,
      "grad_norm": 1.4998205415395205e-06,
      "learning_rate": 1.646877014948587e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13880056,
      "step": 24435
    },
    {
      "epoch": 428.7787610619469,
      "grad_norm": 1.2382743079797365e-06,
      "learning_rate": 1.6459542645508433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13883192,
      "step": 24440
    },
    {
      "epoch": 428.86725663716817,
      "grad_norm": 1.3852245501766447e-06,
      "learning_rate": 1.6450316458576852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13885720,
      "step": 24445
    },
    {
      "epoch": 428.95575221238937,
      "grad_norm": 1.4536259413944208e-06,
      "learning_rate": 1.6441091590113912e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13888776,
      "step": 24450
    },
    {
      "epoch": 429.0353982300885,
      "grad_norm": 2.307497879883158e-06,
      "learning_rate": 1.6431868041542213e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13891224,
      "step": 24455
    },
    {
      "epoch": 429.12389380530976,
      "grad_norm": 1.2047797781633562e-06,
      "learning_rate": 1.6422645814284123e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13894136,
      "step": 24460
    },
    {
      "epoch": 429.21238938053096,
      "grad_norm": 3.0683190743729938e-06,
      "learning_rate": 1.6413424909761846e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13896744,
      "step": 24465
    },
    {
      "epoch": 429.3008849557522,
      "grad_norm": 1.2386337402858771e-05,
      "learning_rate": 1.640420532939736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13899512,
      "step": 24470
    },
    {
      "epoch": 429.3893805309734,
      "grad_norm": 9.790277317733853e-07,
      "learning_rate": 1.639498707461242e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13902616,
      "step": 24475
    },
    {
      "epoch": 429.4778761061947,
      "grad_norm": 7.366340355474676e-07,
      "learning_rate": 1.6385770146828614e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13905256,
      "step": 24480
    },
    {
      "epoch": 429.56637168141594,
      "grad_norm": 8.094949066617119e-07,
      "learning_rate": 1.637655454746731e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13908536,
      "step": 24485
    },
    {
      "epoch": 429.65486725663715,
      "grad_norm": 1.3162690493118134e-06,
      "learning_rate": 1.6367340277949658e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13911384,
      "step": 24490
    },
    {
      "epoch": 429.7433628318584,
      "grad_norm": 1.0857164625122095e-06,
      "learning_rate": 1.635812733969663e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13914520,
      "step": 24495
    },
    {
      "epoch": 429.83185840707966,
      "grad_norm": 3.004145128215896e-06,
      "learning_rate": 1.634891573412896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13917448,
      "step": 24500
    },
    {
      "epoch": 429.92035398230087,
      "grad_norm": 3.1228667012328515e-06,
      "learning_rate": 1.6339705462667196e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13920136,
      "step": 24505
    },
    {
      "epoch": 430.0,
      "grad_norm": 8.603004744145437e-07,
      "learning_rate": 1.633049652673169e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13922320,
      "step": 24510
    },
    {
      "epoch": 430.08849557522126,
      "grad_norm": 1.0017251952376682e-06,
      "learning_rate": 1.632128892774256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13925120,
      "step": 24515
    },
    {
      "epoch": 430.17699115044246,
      "grad_norm": 2.7200985641684383e-06,
      "learning_rate": 1.6312082667119737e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13928400,
      "step": 24520
    },
    {
      "epoch": 430.2654867256637,
      "grad_norm": 2.845150447683409e-06,
      "learning_rate": 1.630287774628296e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13930816,
      "step": 24525
    },
    {
      "epoch": 430.353982300885,
      "grad_norm": 9.855160669758334e-07,
      "learning_rate": 1.6293674166651718e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13933712,
      "step": 24530
    },
    {
      "epoch": 430.4424778761062,
      "grad_norm": 9.299616067437455e-06,
      "learning_rate": 1.6284471929645338e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13936576,
      "step": 24535
    },
    {
      "epoch": 430.53097345132744,
      "grad_norm": 9.966901188818156e-07,
      "learning_rate": 1.627527103668291e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13939824,
      "step": 24540
    },
    {
      "epoch": 430.6194690265487,
      "grad_norm": 3.0459439130936516e-06,
      "learning_rate": 1.6266071489183327e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13942752,
      "step": 24545
    },
    {
      "epoch": 430.7079646017699,
      "grad_norm": 2.8266883873584447e-06,
      "learning_rate": 1.6256873288565283e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13945504,
      "step": 24550
    },
    {
      "epoch": 430.79646017699116,
      "grad_norm": 2.9179359444242436e-06,
      "learning_rate": 1.6247676436247245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13948240,
      "step": 24555
    },
    {
      "epoch": 430.88495575221236,
      "grad_norm": 1.1391593943699263e-06,
      "learning_rate": 1.6238480933647486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13951040,
      "step": 24560
    },
    {
      "epoch": 430.9734513274336,
      "grad_norm": 7.478247425751761e-07,
      "learning_rate": 1.6229286782184083e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13953808,
      "step": 24565
    },
    {
      "epoch": 431.05309734513276,
      "grad_norm": 2.639255171743571e-06,
      "learning_rate": 1.622009398327487e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13956216,
      "step": 24570
    },
    {
      "epoch": 431.14159292035396,
      "grad_norm": 2.5969711714424193e-06,
      "learning_rate": 1.6210902538337502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13958872,
      "step": 24575
    },
    {
      "epoch": 431.2300884955752,
      "grad_norm": 1.3859730643162038e-06,
      "learning_rate": 1.6201712448789413e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13961848,
      "step": 24580
    },
    {
      "epoch": 431.3185840707965,
      "grad_norm": 1.953033006429905e-06,
      "learning_rate": 1.6192523716047827e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13964632,
      "step": 24585
    },
    {
      "epoch": 431.4070796460177,
      "grad_norm": 2.622510237415554e-06,
      "learning_rate": 1.6183336341529776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13967048,
      "step": 24590
    },
    {
      "epoch": 431.49557522123894,
      "grad_norm": 2.828768629115075e-06,
      "learning_rate": 1.6174150326652047e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13969480,
      "step": 24595
    },
    {
      "epoch": 431.5840707964602,
      "grad_norm": 1.218737111230439e-06,
      "learning_rate": 1.6164965672831256e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13972648,
      "step": 24600
    },
    {
      "epoch": 431.5840707964602,
      "eval_loss": 0.5022317171096802,
      "eval_runtime": 1.0637,
      "eval_samples_per_second": 23.503,
      "eval_steps_per_second": 12.222,
      "num_input_tokens_seen": 13972648,
      "step": 24600
    },
    {
      "epoch": 431.6725663716814,
      "grad_norm": 9.26009761315072e-07,
      "learning_rate": 1.6155782381483784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13975832,
      "step": 24605
    },
    {
      "epoch": 431.76106194690266,
      "grad_norm": 1.4301097053248668e-06,
      "learning_rate": 1.6146600454025813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13979112,
      "step": 24610
    },
    {
      "epoch": 431.8495575221239,
      "grad_norm": 1.350045636172581e-06,
      "learning_rate": 1.6137419891873317e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13981960,
      "step": 24615
    },
    {
      "epoch": 431.9380530973451,
      "grad_norm": 1.565350430610124e-06,
      "learning_rate": 1.6128240696442038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13984536,
      "step": 24620
    },
    {
      "epoch": 432.01769911504425,
      "grad_norm": 2.9592592909466475e-06,
      "learning_rate": 1.611906286914753e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13986872,
      "step": 24625
    },
    {
      "epoch": 432.1061946902655,
      "grad_norm": 9.763632533577038e-07,
      "learning_rate": 1.6109886411405144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13989352,
      "step": 24630
    },
    {
      "epoch": 432.1946902654867,
      "grad_norm": 3.2117598038894357e-06,
      "learning_rate": 1.6100711324629985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13992280,
      "step": 24635
    },
    {
      "epoch": 432.283185840708,
      "grad_norm": 3.889714207616635e-06,
      "learning_rate": 1.609153761023698e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13995256,
      "step": 24640
    },
    {
      "epoch": 432.37168141592923,
      "grad_norm": 9.704993999548606e-07,
      "learning_rate": 1.608236526964083e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 13997992,
      "step": 24645
    },
    {
      "epoch": 432.46017699115043,
      "grad_norm": 4.5096416556589247e-07,
      "learning_rate": 1.607319430425601e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14000792,
      "step": 24650
    },
    {
      "epoch": 432.5486725663717,
      "grad_norm": 6.726513674948364e-07,
      "learning_rate": 1.606402471549682e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14003352,
      "step": 24655
    },
    {
      "epoch": 432.6371681415929,
      "grad_norm": 8.280323413600854e-07,
      "learning_rate": 1.6054856504777312e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14007016,
      "step": 24660
    },
    {
      "epoch": 432.72566371681415,
      "grad_norm": 6.97346763445239e-07,
      "learning_rate": 1.6045689673511334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14009928,
      "step": 24665
    },
    {
      "epoch": 432.8141592920354,
      "grad_norm": 1.7508187966086552e-06,
      "learning_rate": 1.6036524223112548e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14012792,
      "step": 24670
    },
    {
      "epoch": 432.9026548672566,
      "grad_norm": 9.628289490137831e-07,
      "learning_rate": 1.602736015499436e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14015400,
      "step": 24675
    },
    {
      "epoch": 432.9911504424779,
      "grad_norm": 5.275448074826272e-06,
      "learning_rate": 1.601819747057e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14018920,
      "step": 24680
    },
    {
      "epoch": 433.070796460177,
      "grad_norm": 2.656995548022678e-06,
      "learning_rate": 1.6009036171252465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14021720,
      "step": 24685
    },
    {
      "epoch": 433.1592920353982,
      "grad_norm": 8.982435701909708e-07,
      "learning_rate": 1.599987625845453e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14025016,
      "step": 24690
    },
    {
      "epoch": 433.24778761061947,
      "grad_norm": 1.360706050945737e-06,
      "learning_rate": 1.599071773358879e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14028056,
      "step": 24695
    },
    {
      "epoch": 433.3362831858407,
      "grad_norm": 2.067044079012703e-06,
      "learning_rate": 1.598156059806758e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14030888,
      "step": 24700
    },
    {
      "epoch": 433.42477876106193,
      "grad_norm": 4.4574113644557656e-07,
      "learning_rate": 1.5972404853303062e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14033480,
      "step": 24705
    },
    {
      "epoch": 433.5132743362832,
      "grad_norm": 1.1345262009854196e-06,
      "learning_rate": 1.5963250500707172e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14036648,
      "step": 24710
    },
    {
      "epoch": 433.60176991150445,
      "grad_norm": 1.0866601769521367e-06,
      "learning_rate": 1.5954097541691612e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14039096,
      "step": 24715
    },
    {
      "epoch": 433.69026548672565,
      "grad_norm": 8.474839887639973e-07,
      "learning_rate": 1.5944945977667884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14042024,
      "step": 24720
    },
    {
      "epoch": 433.7787610619469,
      "grad_norm": 8.875317121237458e-07,
      "learning_rate": 1.593579581004729e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14045032,
      "step": 24725
    },
    {
      "epoch": 433.86725663716817,
      "grad_norm": 1.152954268945905e-06,
      "learning_rate": 1.592664704024088e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14047384,
      "step": 24730
    },
    {
      "epoch": 433.95575221238937,
      "grad_norm": 1.2516961760411505e-06,
      "learning_rate": 1.591749966965953e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14050136,
      "step": 24735
    },
    {
      "epoch": 434.0353982300885,
      "grad_norm": 3.135061660941574e-06,
      "learning_rate": 1.5908353699713856e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14052824,
      "step": 24740
    },
    {
      "epoch": 434.12389380530976,
      "grad_norm": 1.5204416285996558e-06,
      "learning_rate": 1.5899209131814298e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14055816,
      "step": 24745
    },
    {
      "epoch": 434.21238938053096,
      "grad_norm": 1.1998463378404267e-06,
      "learning_rate": 1.5890065967371067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14058712,
      "step": 24750
    },
    {
      "epoch": 434.3008849557522,
      "grad_norm": 4.422429356054636e-06,
      "learning_rate": 1.5880924207794144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14061176,
      "step": 24755
    },
    {
      "epoch": 434.3893805309734,
      "grad_norm": 4.487095338845393e-06,
      "learning_rate": 1.5871783854493298e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14064056,
      "step": 24760
    },
    {
      "epoch": 434.4778761061947,
      "grad_norm": 9.703201158117736e-07,
      "learning_rate": 1.5862644908878106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14067096,
      "step": 24765
    },
    {
      "epoch": 434.56637168141594,
      "grad_norm": 8.263232302851975e-07,
      "learning_rate": 1.5853507372357885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14070264,
      "step": 24770
    },
    {
      "epoch": 434.65486725663715,
      "grad_norm": 9.150141409008938e-07,
      "learning_rate": 1.5844371246341776e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14072728,
      "step": 24775
    },
    {
      "epoch": 434.7433628318584,
      "grad_norm": 1.0863644774872228e-06,
      "learning_rate": 1.5835236532238674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14075464,
      "step": 24780
    },
    {
      "epoch": 434.83185840707966,
      "grad_norm": 4.763543529406888e-06,
      "learning_rate": 1.582610323145727e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14078776,
      "step": 24785
    },
    {
      "epoch": 434.92035398230087,
      "grad_norm": 3.078285544688697e-06,
      "learning_rate": 1.5816971345406035e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14081576,
      "step": 24790
    },
    {
      "epoch": 435.0,
      "grad_norm": 2.5839030058705248e-05,
      "learning_rate": 1.5807840875493225e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14083896,
      "step": 24795
    },
    {
      "epoch": 435.08849557522126,
      "grad_norm": 1.8245970068164752e-06,
      "learning_rate": 1.5798711823126854e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14086360,
      "step": 24800
    },
    {
      "epoch": 435.08849557522126,
      "eval_loss": 0.5776745080947876,
      "eval_runtime": 1.0708,
      "eval_samples_per_second": 23.347,
      "eval_steps_per_second": 12.141,
      "num_input_tokens_seen": 14086360,
      "step": 24800
    },
    {
      "epoch": 435.17699115044246,
      "grad_norm": 3.5659445529745426e-06,
      "learning_rate": 1.578958418971477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14089400,
      "step": 24805
    },
    {
      "epoch": 435.2654867256637,
      "grad_norm": 1.5398729829030344e-06,
      "learning_rate": 1.578045797666453e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14092600,
      "step": 24810
    },
    {
      "epoch": 435.353982300885,
      "grad_norm": 4.000652552349493e-06,
      "learning_rate": 1.5771333185383548e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14095240,
      "step": 24815
    },
    {
      "epoch": 435.4424778761062,
      "grad_norm": 2.857400659195264e-06,
      "learning_rate": 1.576220981727895e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14098200,
      "step": 24820
    },
    {
      "epoch": 435.53097345132744,
      "grad_norm": 7.123363161554153e-07,
      "learning_rate": 1.575308787375769e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14101480,
      "step": 24825
    },
    {
      "epoch": 435.6194690265487,
      "grad_norm": 6.126202265477332e-07,
      "learning_rate": 1.5743967356226492e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14103960,
      "step": 24830
    },
    {
      "epoch": 435.7079646017699,
      "grad_norm": 6.781047545700858e-07,
      "learning_rate": 1.5734848266091835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14107048,
      "step": 24835
    },
    {
      "epoch": 435.79646017699116,
      "grad_norm": 6.761304689462122e-07,
      "learning_rate": 1.572573060476001e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14110008,
      "step": 24840
    },
    {
      "epoch": 435.88495575221236,
      "grad_norm": 7.999594799912302e-07,
      "learning_rate": 1.5716614373637085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14113000,
      "step": 24845
    },
    {
      "epoch": 435.9734513274336,
      "grad_norm": 8.139320470945677e-07,
      "learning_rate": 1.570749957412887e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14115752,
      "step": 24850
    },
    {
      "epoch": 436.05309734513276,
      "grad_norm": 4.382645784062333e-06,
      "learning_rate": 1.5698386207641013e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14118128,
      "step": 24855
    },
    {
      "epoch": 436.14159292035396,
      "grad_norm": 9.585673979017884e-07,
      "learning_rate": 1.5689274275578884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14121072,
      "step": 24860
    },
    {
      "epoch": 436.2300884955752,
      "grad_norm": 1.242105554410955e-06,
      "learning_rate": 1.5680163779347667e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14124304,
      "step": 24865
    },
    {
      "epoch": 436.3185840707965,
      "grad_norm": 2.221689555881312e-06,
      "learning_rate": 1.5671054720352327e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14127056,
      "step": 24870
    },
    {
      "epoch": 436.4070796460177,
      "grad_norm": 2.9867894681956386e-06,
      "learning_rate": 1.566194709999757e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14129776,
      "step": 24875
    },
    {
      "epoch": 436.49557522123894,
      "grad_norm": 1.5244395399349742e-06,
      "learning_rate": 1.5652840919687933e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14132400,
      "step": 24880
    },
    {
      "epoch": 436.5840707964602,
      "grad_norm": 3.865499365929281e-06,
      "learning_rate": 1.5643736180827676e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14135280,
      "step": 24885
    },
    {
      "epoch": 436.6725663716814,
      "grad_norm": 1.4133092918200418e-06,
      "learning_rate": 1.5634632884820878e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14137776,
      "step": 24890
    },
    {
      "epoch": 436.76106194690266,
      "grad_norm": 5.505098215508042e-06,
      "learning_rate": 1.5625531033071395e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14140768,
      "step": 24895
    },
    {
      "epoch": 436.8495575221239,
      "grad_norm": 2.7031678655475844e-06,
      "learning_rate": 1.5616430626982828e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14143744,
      "step": 24900
    },
    {
      "epoch": 436.9380530973451,
      "grad_norm": 5.086569672130281e-06,
      "learning_rate": 1.5607331667958575e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14146448,
      "step": 24905
    },
    {
      "epoch": 437.01769911504425,
      "grad_norm": 1.0030587418441428e-06,
      "learning_rate": 1.5598234157401824e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14149152,
      "step": 24910
    },
    {
      "epoch": 437.1061946902655,
      "grad_norm": 1.0860056818273733e-06,
      "learning_rate": 1.5589138096715503e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14151952,
      "step": 24915
    },
    {
      "epoch": 437.1946902654867,
      "grad_norm": 1.1913389244000427e-06,
      "learning_rate": 1.5580043487302365e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14154784,
      "step": 24920
    },
    {
      "epoch": 437.283185840708,
      "grad_norm": 2.7048977244703565e-06,
      "learning_rate": 1.5570950330564888e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14157872,
      "step": 24925
    },
    {
      "epoch": 437.37168141592923,
      "grad_norm": 1.1503333325890708e-06,
      "learning_rate": 1.5561858627905367e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14161040,
      "step": 24930
    },
    {
      "epoch": 437.46017699115043,
      "grad_norm": 1.2381564147290192e-06,
      "learning_rate": 1.5552768380725857e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14163872,
      "step": 24935
    },
    {
      "epoch": 437.5486725663717,
      "grad_norm": 1.4264151104725897e-06,
      "learning_rate": 1.5543679590428183e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14167152,
      "step": 24940
    },
    {
      "epoch": 437.6371681415929,
      "grad_norm": 1.1584693311306182e-06,
      "learning_rate": 1.5534592258413943e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14169888,
      "step": 24945
    },
    {
      "epoch": 437.72566371681415,
      "grad_norm": 2.7356140890333336e-06,
      "learning_rate": 1.5525506386084538e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14172992,
      "step": 24950
    },
    {
      "epoch": 437.8141592920354,
      "grad_norm": 2.449052317388123e-06,
      "learning_rate": 1.55164219748411e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14175568,
      "step": 24955
    },
    {
      "epoch": 437.9026548672566,
      "grad_norm": 1.1575449434531038e-06,
      "learning_rate": 1.550733902608459e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14178144,
      "step": 24960
    },
    {
      "epoch": 437.9911504424779,
      "grad_norm": 2.681321575437323e-06,
      "learning_rate": 1.549825754121568e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14181104,
      "step": 24965
    },
    {
      "epoch": 438.070796460177,
      "grad_norm": 1.0313340681022964e-05,
      "learning_rate": 1.5489177521634864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14183192,
      "step": 24970
    },
    {
      "epoch": 438.1592920353982,
      "grad_norm": 4.052221356687369e-06,
      "learning_rate": 1.5480098968742402e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14185976,
      "step": 24975
    },
    {
      "epoch": 438.24778761061947,
      "grad_norm": 6.212222274371015e-07,
      "learning_rate": 1.5471021883938304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14189144,
      "step": 24980
    },
    {
      "epoch": 438.3362831858407,
      "grad_norm": 1.9584642814152176e-06,
      "learning_rate": 1.546194626862238e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14192472,
      "step": 24985
    },
    {
      "epoch": 438.42477876106193,
      "grad_norm": 9.275059937863261e-07,
      "learning_rate": 1.5452872124194216e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14195560,
      "step": 24990
    },
    {
      "epoch": 438.5132743362832,
      "grad_norm": 2.433367626508698e-06,
      "learning_rate": 1.5443799452053136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14198760,
      "step": 24995
    },
    {
      "epoch": 438.60176991150445,
      "grad_norm": 5.8525770327833015e-06,
      "learning_rate": 1.543472825359828e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14201656,
      "step": 25000
    },
    {
      "epoch": 438.60176991150445,
      "eval_loss": 0.5152356624603271,
      "eval_runtime": 1.0596,
      "eval_samples_per_second": 23.593,
      "eval_steps_per_second": 12.268,
      "num_input_tokens_seen": 14201656,
      "step": 25000
    },
    {
      "epoch": 438.69026548672565,
      "grad_norm": 1.7878400058179977e-06,
      "learning_rate": 1.5425658530228522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14204424,
      "step": 25005
    },
    {
      "epoch": 438.7787610619469,
      "grad_norm": 2.72212832896912e-06,
      "learning_rate": 1.5416590283342546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14207144,
      "step": 25010
    },
    {
      "epoch": 438.86725663716817,
      "grad_norm": 1.5224951539494214e-06,
      "learning_rate": 1.5407523514338783e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14209656,
      "step": 25015
    },
    {
      "epoch": 438.95575221238937,
      "grad_norm": 6.2072881519270595e-06,
      "learning_rate": 1.539845822461543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14212424,
      "step": 25020
    },
    {
      "epoch": 439.0353982300885,
      "grad_norm": 2.5412384729861515e-06,
      "learning_rate": 1.538939441557048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14214888,
      "step": 25025
    },
    {
      "epoch": 439.12389380530976,
      "grad_norm": 3.00792271445971e-06,
      "learning_rate": 1.5380332088601696e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14217768,
      "step": 25030
    },
    {
      "epoch": 439.21238938053096,
      "grad_norm": 1.0592067383186077e-06,
      "learning_rate": 1.537127124510658e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14220328,
      "step": 25035
    },
    {
      "epoch": 439.3008849557522,
      "grad_norm": 9.99482494989934e-07,
      "learning_rate": 1.5362211886482457e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14222840,
      "step": 25040
    },
    {
      "epoch": 439.3893805309734,
      "grad_norm": 4.362138497526757e-06,
      "learning_rate": 1.5353154014126363e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14225352,
      "step": 25045
    },
    {
      "epoch": 439.4778761061947,
      "grad_norm": 9.391906132805161e-07,
      "learning_rate": 1.534409762943515e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14228072,
      "step": 25050
    },
    {
      "epoch": 439.56637168141594,
      "grad_norm": 1.2570376384246629e-06,
      "learning_rate": 1.5335042733805438e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14231080,
      "step": 25055
    },
    {
      "epoch": 439.65486725663715,
      "grad_norm": 2.526627213228494e-06,
      "learning_rate": 1.532598932863358e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14234200,
      "step": 25060
    },
    {
      "epoch": 439.7433628318584,
      "grad_norm": 9.839634458330693e-07,
      "learning_rate": 1.531693741531574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14237048,
      "step": 25065
    },
    {
      "epoch": 439.83185840707966,
      "grad_norm": 4.976307650395029e-07,
      "learning_rate": 1.5307886995247844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14239992,
      "step": 25070
    },
    {
      "epoch": 439.92035398230087,
      "grad_norm": 9.97986148831842e-07,
      "learning_rate": 1.529883806982557e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14243480,
      "step": 25075
    },
    {
      "epoch": 440.0,
      "grad_norm": 2.1216412733338075e-06,
      "learning_rate": 1.5289790640444376e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14246096,
      "step": 25080
    },
    {
      "epoch": 440.08849557522126,
      "grad_norm": 8.654089356241457e-07,
      "learning_rate": 1.5280744708499494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14249216,
      "step": 25085
    },
    {
      "epoch": 440.17699115044246,
      "grad_norm": 1.8442282225805684e-06,
      "learning_rate": 1.527170027538591e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14252000,
      "step": 25090
    },
    {
      "epoch": 440.2654867256637,
      "grad_norm": 4.291327513783472e-06,
      "learning_rate": 1.5262657342498407e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14255216,
      "step": 25095
    },
    {
      "epoch": 440.353982300885,
      "grad_norm": 2.966867668874329e-06,
      "learning_rate": 1.52536159112315e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14257904,
      "step": 25100
    },
    {
      "epoch": 440.4424778761062,
      "grad_norm": 3.30776288137713e-06,
      "learning_rate": 1.5244575982979497e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14260496,
      "step": 25105
    },
    {
      "epoch": 440.53097345132744,
      "grad_norm": 1.1131866131108836e-06,
      "learning_rate": 1.5235537559136487e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14263680,
      "step": 25110
    },
    {
      "epoch": 440.6194690265487,
      "grad_norm": 9.804483624975546e-07,
      "learning_rate": 1.5226500641096286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14266272,
      "step": 25115
    },
    {
      "epoch": 440.7079646017699,
      "grad_norm": 8.906262678465282e-07,
      "learning_rate": 1.5217465230252509e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14268784,
      "step": 25120
    },
    {
      "epoch": 440.79646017699116,
      "grad_norm": 4.676020125771174e-06,
      "learning_rate": 1.5208431327998523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14271952,
      "step": 25125
    },
    {
      "epoch": 440.88495575221236,
      "grad_norm": 1.6904987205634825e-06,
      "learning_rate": 1.5199398935727477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14274608,
      "step": 25130
    },
    {
      "epoch": 440.9734513274336,
      "grad_norm": 9.202929049934028e-07,
      "learning_rate": 1.5190368054832282e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14277520,
      "step": 25135
    },
    {
      "epoch": 441.05309734513276,
      "grad_norm": 1.8561615888756933e-06,
      "learning_rate": 1.5181338686705601e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14280120,
      "step": 25140
    },
    {
      "epoch": 441.14159292035396,
      "grad_norm": 3.494184511509957e-06,
      "learning_rate": 1.5172310832739889e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14283176,
      "step": 25145
    },
    {
      "epoch": 441.2300884955752,
      "grad_norm": 3.3499213714094367e-06,
      "learning_rate": 1.5163284494327346e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14285800,
      "step": 25150
    },
    {
      "epoch": 441.3185840707965,
      "grad_norm": 2.1093796931381803e-06,
      "learning_rate": 1.5154259672859952e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14288680,
      "step": 25155
    },
    {
      "epoch": 441.4070796460177,
      "grad_norm": 1.0535361525398912e-06,
      "learning_rate": 1.5145236369729452e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14291416,
      "step": 25160
    },
    {
      "epoch": 441.49557522123894,
      "grad_norm": 2.9132984309399035e-06,
      "learning_rate": 1.5136214586327335e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14294424,
      "step": 25165
    },
    {
      "epoch": 441.5840707964602,
      "grad_norm": 1.0904529972322052e-06,
      "learning_rate": 1.5127194324044885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14297048,
      "step": 25170
    },
    {
      "epoch": 441.6725663716814,
      "grad_norm": 4.721188361145323e-06,
      "learning_rate": 1.5118175584273148e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14299672,
      "step": 25175
    },
    {
      "epoch": 441.76106194690266,
      "grad_norm": 2.3322897959587863e-06,
      "learning_rate": 1.5109158368402909e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14302568,
      "step": 25180
    },
    {
      "epoch": 441.8495575221239,
      "grad_norm": 7.946275673020864e-07,
      "learning_rate": 1.5100142677824753e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14305448,
      "step": 25185
    },
    {
      "epoch": 441.9380530973451,
      "grad_norm": 1.3023058045291691e-06,
      "learning_rate": 1.509112851392901e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14308680,
      "step": 25190
    },
    {
      "epoch": 442.01769911504425,
      "grad_norm": 7.119199381122598e-07,
      "learning_rate": 1.5082115878105763e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14311808,
      "step": 25195
    },
    {
      "epoch": 442.1061946902655,
      "grad_norm": 8.910229212233389e-07,
      "learning_rate": 1.5073104771744892e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14314736,
      "step": 25200
    },
    {
      "epoch": 442.1061946902655,
      "eval_loss": 0.5149388313293457,
      "eval_runtime": 1.0744,
      "eval_samples_per_second": 23.268,
      "eval_steps_per_second": 12.099,
      "num_input_tokens_seen": 14314736,
      "step": 25200
    },
    {
      "epoch": 442.1946902654867,
      "grad_norm": 9.671600309957284e-06,
      "learning_rate": 1.5064095196236006e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14317232,
      "step": 25205
    },
    {
      "epoch": 442.283185840708,
      "grad_norm": 3.344347305755946e-06,
      "learning_rate": 1.50550871529685e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14319824,
      "step": 25210
    },
    {
      "epoch": 442.37168141592923,
      "grad_norm": 2.560136863394291e-06,
      "learning_rate": 1.5046080643331546e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14323552,
      "step": 25215
    },
    {
      "epoch": 442.46017699115043,
      "grad_norm": 7.996986823854968e-07,
      "learning_rate": 1.5037075668714028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14326240,
      "step": 25220
    },
    {
      "epoch": 442.5486725663717,
      "grad_norm": 7.930835295155703e-07,
      "learning_rate": 1.5028072230504656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14329824,
      "step": 25225
    },
    {
      "epoch": 442.6371681415929,
      "grad_norm": 2.939465275630937e-06,
      "learning_rate": 1.5019070330091861e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14332720,
      "step": 25230
    },
    {
      "epoch": 442.72566371681415,
      "grad_norm": 1.2289967799006263e-06,
      "learning_rate": 1.5010069968863843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14335168,
      "step": 25235
    },
    {
      "epoch": 442.8141592920354,
      "grad_norm": 8.495571250932699e-07,
      "learning_rate": 1.5001071148208584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14337936,
      "step": 25240
    },
    {
      "epoch": 442.9026548672566,
      "grad_norm": 8.701329534233082e-06,
      "learning_rate": 1.49920738695138e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14340288,
      "step": 25245
    },
    {
      "epoch": 442.9911504424779,
      "grad_norm": 2.8849558475485537e-06,
      "learning_rate": 1.4983078134166995e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14343200,
      "step": 25250
    },
    {
      "epoch": 443.070796460177,
      "grad_norm": 7.707317308813799e-07,
      "learning_rate": 1.4974083943555428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14345424,
      "step": 25255
    },
    {
      "epoch": 443.1592920353982,
      "grad_norm": 7.569513513772108e-07,
      "learning_rate": 1.496509129906611e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14348080,
      "step": 25260
    },
    {
      "epoch": 443.24778761061947,
      "grad_norm": 1.201000713990652e-06,
      "learning_rate": 1.4956100202085809e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14351088,
      "step": 25265
    },
    {
      "epoch": 443.3362831858407,
      "grad_norm": 9.145104513663682e-07,
      "learning_rate": 1.4947110654001093e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14354304,
      "step": 25270
    },
    {
      "epoch": 443.42477876106193,
      "grad_norm": 6.246339125937084e-07,
      "learning_rate": 1.4938122656198234e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14356976,
      "step": 25275
    },
    {
      "epoch": 443.5132743362832,
      "grad_norm": 9.042598776431987e-07,
      "learning_rate": 1.4929136210063316e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14359952,
      "step": 25280
    },
    {
      "epoch": 443.60176991150445,
      "grad_norm": 1.5503251233894844e-06,
      "learning_rate": 1.4920151316982146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14362992,
      "step": 25285
    },
    {
      "epoch": 443.69026548672565,
      "grad_norm": 3.923008534911787e-06,
      "learning_rate": 1.4911167978340312e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14365696,
      "step": 25290
    },
    {
      "epoch": 443.7787610619469,
      "grad_norm": 1.5739863101771334e-06,
      "learning_rate": 1.4902186195523166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14368800,
      "step": 25295
    },
    {
      "epoch": 443.86725663716817,
      "grad_norm": 8.369635793314956e-07,
      "learning_rate": 1.4893205969915805e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14371440,
      "step": 25300
    },
    {
      "epoch": 443.95575221238937,
      "grad_norm": 2.638872047100449e-06,
      "learning_rate": 1.4884227302903086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14374960,
      "step": 25305
    },
    {
      "epoch": 444.0353982300885,
      "grad_norm": 4.17698129240307e-06,
      "learning_rate": 1.4875250195869653e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14377168,
      "step": 25310
    },
    {
      "epoch": 444.12389380530976,
      "grad_norm": 2.2881265522300964e-06,
      "learning_rate": 1.4866274650199862e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14380288,
      "step": 25315
    },
    {
      "epoch": 444.21238938053096,
      "grad_norm": 2.8576091608556453e-06,
      "learning_rate": 1.485730066727788e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14382640,
      "step": 25320
    },
    {
      "epoch": 444.3008849557522,
      "grad_norm": 2.0649758880608715e-06,
      "learning_rate": 1.4848328248487586e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14385440,
      "step": 25325
    },
    {
      "epoch": 444.3893805309734,
      "grad_norm": 1.370899667563208e-06,
      "learning_rate": 1.4839357395212656e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14388304,
      "step": 25330
    },
    {
      "epoch": 444.4778761061947,
      "grad_norm": 2.78682909993222e-06,
      "learning_rate": 1.4830388108836502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14391376,
      "step": 25335
    },
    {
      "epoch": 444.56637168141594,
      "grad_norm": 9.360588819617988e-07,
      "learning_rate": 1.4821420390742299e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14394336,
      "step": 25340
    },
    {
      "epoch": 444.65486725663715,
      "grad_norm": 1.5427970083692344e-06,
      "learning_rate": 1.4812454242312979e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14396880,
      "step": 25345
    },
    {
      "epoch": 444.7433628318584,
      "grad_norm": 2.7143871648149798e-06,
      "learning_rate": 1.4803489664931253e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14399520,
      "step": 25350
    },
    {
      "epoch": 444.83185840707966,
      "grad_norm": 9.928561439664918e-07,
      "learning_rate": 1.4794526659979544e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14402304,
      "step": 25355
    },
    {
      "epoch": 444.92035398230087,
      "grad_norm": 1.1242234450037358e-06,
      "learning_rate": 1.4785565228840086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14405232,
      "step": 25360
    },
    {
      "epoch": 445.0,
      "grad_norm": 1.3469438044921844e-06,
      "learning_rate": 1.4776605372894819e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14407480,
      "step": 25365
    },
    {
      "epoch": 445.08849557522126,
      "grad_norm": 1.1611413128775894e-06,
      "learning_rate": 1.4767647093525488e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14410616,
      "step": 25370
    },
    {
      "epoch": 445.17699115044246,
      "grad_norm": 1.1144172731292201e-06,
      "learning_rate": 1.4758690392113566e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14413176,
      "step": 25375
    },
    {
      "epoch": 445.2654867256637,
      "grad_norm": 7.011756792962842e-07,
      "learning_rate": 1.4749735270040276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14416200,
      "step": 25380
    },
    {
      "epoch": 445.353982300885,
      "grad_norm": 4.3245522647339385e-06,
      "learning_rate": 1.4740781728686623e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14419464,
      "step": 25385
    },
    {
      "epoch": 445.4424778761062,
      "grad_norm": 1.097613676392939e-06,
      "learning_rate": 1.4731829769433358e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14421992,
      "step": 25390
    },
    {
      "epoch": 445.53097345132744,
      "grad_norm": 1.6138714045155211e-06,
      "learning_rate": 1.4722879393660976e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14425256,
      "step": 25395
    },
    {
      "epoch": 445.6194690265487,
      "grad_norm": 4.740087661048165e-06,
      "learning_rate": 1.4713930602749748e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14428104,
      "step": 25400
    },
    {
      "epoch": 445.6194690265487,
      "eval_loss": 0.5317846536636353,
      "eval_runtime": 1.0667,
      "eval_samples_per_second": 23.436,
      "eval_steps_per_second": 12.187,
      "num_input_tokens_seen": 14428104,
      "step": 25400
    },
    {
      "epoch": 445.7079646017699,
      "grad_norm": 1.159801968242391e-06,
      "learning_rate": 1.470498339807968e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14430552,
      "step": 25405
    },
    {
      "epoch": 445.79646017699116,
      "grad_norm": 1.0583811445030733e-06,
      "learning_rate": 1.4696037781030542e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14433528,
      "step": 25410
    },
    {
      "epoch": 445.88495575221236,
      "grad_norm": 4.998825261282036e-06,
      "learning_rate": 1.4687093752981876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14436056,
      "step": 25415
    },
    {
      "epoch": 445.9734513274336,
      "grad_norm": 1.6062383565440541e-06,
      "learning_rate": 1.4678151315312943e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14439160,
      "step": 25420
    },
    {
      "epoch": 446.05309734513276,
      "grad_norm": 2.768265630948008e-06,
      "learning_rate": 1.4669210469402789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14441688,
      "step": 25425
    },
    {
      "epoch": 446.14159292035396,
      "grad_norm": 2.483783418938401e-06,
      "learning_rate": 1.4660271216630218e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14444552,
      "step": 25430
    },
    {
      "epoch": 446.2300884955752,
      "grad_norm": 3.4326969853282208e-06,
      "learning_rate": 1.4651333558373748e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14447160,
      "step": 25435
    },
    {
      "epoch": 446.3185840707965,
      "grad_norm": 6.973370432206139e-07,
      "learning_rate": 1.4642397496011707e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14450104,
      "step": 25440
    },
    {
      "epoch": 446.4070796460177,
      "grad_norm": 1.1933951782339136e-06,
      "learning_rate": 1.4633463030922129e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14453128,
      "step": 25445
    },
    {
      "epoch": 446.49557522123894,
      "grad_norm": 5.097728603686846e-07,
      "learning_rate": 1.462453016448282e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14456008,
      "step": 25450
    },
    {
      "epoch": 446.5840707964602,
      "grad_norm": 8.69094435529405e-07,
      "learning_rate": 1.4615598898071354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14459208,
      "step": 25455
    },
    {
      "epoch": 446.6725663716814,
      "grad_norm": 8.096037618088303e-07,
      "learning_rate": 1.4606669233065026e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14461800,
      "step": 25460
    },
    {
      "epoch": 446.76106194690266,
      "grad_norm": 6.187494818732375e-06,
      "learning_rate": 1.4597741170840914e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14464744,
      "step": 25465
    },
    {
      "epoch": 446.8495575221239,
      "grad_norm": 1.2207318604851025e-06,
      "learning_rate": 1.4588814712775853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14467672,
      "step": 25470
    },
    {
      "epoch": 446.9380530973451,
      "grad_norm": 1.107581852011208e-06,
      "learning_rate": 1.4579889860246382e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14470456,
      "step": 25475
    },
    {
      "epoch": 447.01769911504425,
      "grad_norm": 1.5363607417384628e-06,
      "learning_rate": 1.457096661462885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14472848,
      "step": 25480
    },
    {
      "epoch": 447.1061946902655,
      "grad_norm": 7.286477057277807e-07,
      "learning_rate": 1.4562044977299322e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14475520,
      "step": 25485
    },
    {
      "epoch": 447.1946902654867,
      "grad_norm": 1.0684457265597302e-06,
      "learning_rate": 1.4553124949633623e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14478144,
      "step": 25490
    },
    {
      "epoch": 447.283185840708,
      "grad_norm": 1.0175950819757418e-06,
      "learning_rate": 1.4544206533007354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14481280,
      "step": 25495
    },
    {
      "epoch": 447.37168141592923,
      "grad_norm": 3.651089855338796e-06,
      "learning_rate": 1.4535289728795821e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14484608,
      "step": 25500
    },
    {
      "epoch": 447.46017699115043,
      "grad_norm": 6.09898097536643e-06,
      "learning_rate": 1.4526374538374132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14487248,
      "step": 25505
    },
    {
      "epoch": 447.5486725663717,
      "grad_norm": 6.415823463612469e-07,
      "learning_rate": 1.4517460963117097e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14489776,
      "step": 25510
    },
    {
      "epoch": 447.6371681415929,
      "grad_norm": 1.7672474541541305e-06,
      "learning_rate": 1.4508549004399314e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14492768,
      "step": 25515
    },
    {
      "epoch": 447.72566371681415,
      "grad_norm": 6.163325565466948e-07,
      "learning_rate": 1.449963866359513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14495664,
      "step": 25520
    },
    {
      "epoch": 447.8141592920354,
      "grad_norm": 9.709563073556637e-07,
      "learning_rate": 1.4490729942078607e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14498544,
      "step": 25525
    },
    {
      "epoch": 447.9026548672566,
      "grad_norm": 4.089973572263261e-06,
      "learning_rate": 1.4481822841223608e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14501520,
      "step": 25530
    },
    {
      "epoch": 447.9911504424779,
      "grad_norm": 1.4647968100689468e-06,
      "learning_rate": 1.4472917362403704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14504656,
      "step": 25535
    },
    {
      "epoch": 448.070796460177,
      "grad_norm": 4.88032924295112e-07,
      "learning_rate": 1.4464013506992224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14506960,
      "step": 25540
    },
    {
      "epoch": 448.1592920353982,
      "grad_norm": 8.684900763000769e-07,
      "learning_rate": 1.4455111276362277e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14510720,
      "step": 25545
    },
    {
      "epoch": 448.24778761061947,
      "grad_norm": 4.964241497873445e-07,
      "learning_rate": 1.4446210671886676e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14513552,
      "step": 25550
    },
    {
      "epoch": 448.3362831858407,
      "grad_norm": 3.0704632081324235e-06,
      "learning_rate": 1.4437311694938015e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14516592,
      "step": 25555
    },
    {
      "epoch": 448.42477876106193,
      "grad_norm": 2.760935558399069e-06,
      "learning_rate": 1.442841434688864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14519216,
      "step": 25560
    },
    {
      "epoch": 448.5132743362832,
      "grad_norm": 3.2267635106109083e-06,
      "learning_rate": 1.4419518629110615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14522608,
      "step": 25565
    },
    {
      "epoch": 448.60176991150445,
      "grad_norm": 6.206592388480203e-06,
      "learning_rate": 1.4410624542975778e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14524928,
      "step": 25570
    },
    {
      "epoch": 448.69026548672565,
      "grad_norm": 1.2754561566907796e-06,
      "learning_rate": 1.4401732089855724e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14527616,
      "step": 25575
    },
    {
      "epoch": 448.7787610619469,
      "grad_norm": 1.099312498809013e-06,
      "learning_rate": 1.4392841271121754e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14530688,
      "step": 25580
    },
    {
      "epoch": 448.86725663716817,
      "grad_norm": 8.505414825776825e-07,
      "learning_rate": 1.438395208814497e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14533152,
      "step": 25585
    },
    {
      "epoch": 448.95575221238937,
      "grad_norm": 3.916306923201773e-06,
      "learning_rate": 1.4375064542296174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14535984,
      "step": 25590
    },
    {
      "epoch": 449.0353982300885,
      "grad_norm": 1.3262437050798326e-06,
      "learning_rate": 1.4366178634945946e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14537984,
      "step": 25595
    },
    {
      "epoch": 449.12389380530976,
      "grad_norm": 8.152295549734845e-07,
      "learning_rate": 1.4357294367464616e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14541136,
      "step": 25600
    },
    {
      "epoch": 449.12389380530976,
      "eval_loss": 0.48938480019569397,
      "eval_runtime": 1.0677,
      "eval_samples_per_second": 23.415,
      "eval_steps_per_second": 12.176,
      "num_input_tokens_seen": 14541136,
      "step": 25600
    },
    {
      "epoch": 449.21238938053096,
      "grad_norm": 9.144769705926592e-07,
      "learning_rate": 1.434841174122224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14543632,
      "step": 25605
    },
    {
      "epoch": 449.3008849557522,
      "grad_norm": 3.057629555769381e-06,
      "learning_rate": 1.4339530757588615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14546160,
      "step": 25610
    },
    {
      "epoch": 449.3893805309734,
      "grad_norm": 1.1067252216889756e-06,
      "learning_rate": 1.433065141793333e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14548832,
      "step": 25615
    },
    {
      "epoch": 449.4778761061947,
      "grad_norm": 4.236069344187854e-06,
      "learning_rate": 1.4321773723625665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14551792,
      "step": 25620
    },
    {
      "epoch": 449.56637168141594,
      "grad_norm": 3.987283434980782e-06,
      "learning_rate": 1.4312897676034693e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14554608,
      "step": 25625
    },
    {
      "epoch": 449.65486725663715,
      "grad_norm": 8.523198289367429e-07,
      "learning_rate": 1.4304023276529188e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14558080,
      "step": 25630
    },
    {
      "epoch": 449.7433628318584,
      "grad_norm": 2.3297588995774277e-06,
      "learning_rate": 1.4295150526477712e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14561024,
      "step": 25635
    },
    {
      "epoch": 449.83185840707966,
      "grad_norm": 1.2627323258129763e-06,
      "learning_rate": 1.4286279427248562e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14564240,
      "step": 25640
    },
    {
      "epoch": 449.92035398230087,
      "grad_norm": 2.924627551692538e-06,
      "learning_rate": 1.4277409980209747e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14567040,
      "step": 25645
    },
    {
      "epoch": 450.0,
      "grad_norm": 1.7751859786585555e-06,
      "learning_rate": 1.4268542186729061e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14569608,
      "step": 25650
    },
    {
      "epoch": 450.08849557522126,
      "grad_norm": 1.7908793097376474e-06,
      "learning_rate": 1.4259676048174043e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14572744,
      "step": 25655
    },
    {
      "epoch": 450.17699115044246,
      "grad_norm": 1.5637082469766028e-05,
      "learning_rate": 1.4250811565911937e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14575960,
      "step": 25660
    },
    {
      "epoch": 450.2654867256637,
      "grad_norm": 4.786474164575338e-06,
      "learning_rate": 1.4241948741309782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14578824,
      "step": 25665
    },
    {
      "epoch": 450.353982300885,
      "grad_norm": 1.324374807154527e-06,
      "learning_rate": 1.4233087575734317e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14581752,
      "step": 25670
    },
    {
      "epoch": 450.4424778761062,
      "grad_norm": 6.45399040877237e-07,
      "learning_rate": 1.422422807055206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14584520,
      "step": 25675
    },
    {
      "epoch": 450.53097345132744,
      "grad_norm": 4.472012733458541e-06,
      "learning_rate": 1.4215370227129243e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14587320,
      "step": 25680
    },
    {
      "epoch": 450.6194690265487,
      "grad_norm": 1.208902517646493e-06,
      "learning_rate": 1.4206514046831876e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14590216,
      "step": 25685
    },
    {
      "epoch": 450.7079646017699,
      "grad_norm": 1.4716916894030874e-06,
      "learning_rate": 1.419765953102567e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14592824,
      "step": 25690
    },
    {
      "epoch": 450.79646017699116,
      "grad_norm": 4.455112502910197e-06,
      "learning_rate": 1.4188806681076125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14595464,
      "step": 25695
    },
    {
      "epoch": 450.88495575221236,
      "grad_norm": 1.0971600659104297e-06,
      "learning_rate": 1.4179955498348443e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14598680,
      "step": 25700
    },
    {
      "epoch": 450.9734513274336,
      "grad_norm": 1.4157467376207933e-06,
      "learning_rate": 1.4171105984207605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14601352,
      "step": 25705
    },
    {
      "epoch": 451.05309734513276,
      "grad_norm": 1.1279043974354863e-06,
      "learning_rate": 1.4162258140018304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14603616,
      "step": 25710
    },
    {
      "epoch": 451.14159292035396,
      "grad_norm": 7.828085699657095e-07,
      "learning_rate": 1.4153411967144986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14606192,
      "step": 25715
    },
    {
      "epoch": 451.2300884955752,
      "grad_norm": 2.0526765638351208e-06,
      "learning_rate": 1.4144567466951864e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14608576,
      "step": 25720
    },
    {
      "epoch": 451.3185840707965,
      "grad_norm": 7.869191449572099e-07,
      "learning_rate": 1.4135724640802844e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14611856,
      "step": 25725
    },
    {
      "epoch": 451.4070796460177,
      "grad_norm": 9.278037964577379e-07,
      "learning_rate": 1.4126883490061615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14614592,
      "step": 25730
    },
    {
      "epoch": 451.49557522123894,
      "grad_norm": 3.4504068935348187e-06,
      "learning_rate": 1.4118044016091603e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14618064,
      "step": 25735
    },
    {
      "epoch": 451.5840707964602,
      "grad_norm": 2.385482048339327e-06,
      "learning_rate": 1.410920622025594e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14620832,
      "step": 25740
    },
    {
      "epoch": 451.6725663716814,
      "grad_norm": 2.2484518922283314e-06,
      "learning_rate": 1.4100370103917554e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14623856,
      "step": 25745
    },
    {
      "epoch": 451.76106194690266,
      "grad_norm": 1.3917818932895898e-06,
      "learning_rate": 1.409153566843907e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14626592,
      "step": 25750
    },
    {
      "epoch": 451.8495575221239,
      "grad_norm": 3.839066721411655e-06,
      "learning_rate": 1.408270291518286e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14629616,
      "step": 25755
    },
    {
      "epoch": 451.9380530973451,
      "grad_norm": 1.2770095736414078e-06,
      "learning_rate": 1.407387184551107e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14632288,
      "step": 25760
    },
    {
      "epoch": 452.01769911504425,
      "grad_norm": 2.8247384307178436e-06,
      "learning_rate": 1.4065042460785532e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14635040,
      "step": 25765
    },
    {
      "epoch": 452.1061946902655,
      "grad_norm": 5.6988974392879754e-06,
      "learning_rate": 1.405621476236787e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14638080,
      "step": 25770
    },
    {
      "epoch": 452.1946902654867,
      "grad_norm": 1.1651684417302022e-06,
      "learning_rate": 1.4047388751619423e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14640768,
      "step": 25775
    },
    {
      "epoch": 452.283185840708,
      "grad_norm": 2.5648032533354126e-06,
      "learning_rate": 1.4038564429901264e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14643808,
      "step": 25780
    },
    {
      "epoch": 452.37168141592923,
      "grad_norm": 3.989111974078696e-06,
      "learning_rate": 1.4029741798574227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14646704,
      "step": 25785
    },
    {
      "epoch": 452.46017699115043,
      "grad_norm": 2.432710516586667e-06,
      "learning_rate": 1.402092085899886e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14649808,
      "step": 25790
    },
    {
      "epoch": 452.5486725663717,
      "grad_norm": 2.5419083158340072e-06,
      "learning_rate": 1.4012101612535464e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14652400,
      "step": 25795
    },
    {
      "epoch": 452.6371681415929,
      "grad_norm": 3.0217972835089313e-06,
      "learning_rate": 1.4003284060544092e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14655696,
      "step": 25800
    },
    {
      "epoch": 452.6371681415929,
      "eval_loss": 0.516354501247406,
      "eval_runtime": 1.0633,
      "eval_samples_per_second": 23.512,
      "eval_steps_per_second": 12.226,
      "num_input_tokens_seen": 14655696,
      "step": 25800
    },
    {
      "epoch": 452.72566371681415,
      "grad_norm": 3.2528721476410283e-06,
      "learning_rate": 1.3994468204384504e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14658528,
      "step": 25805
    },
    {
      "epoch": 452.8141592920354,
      "grad_norm": 2.538337867008522e-06,
      "learning_rate": 1.398565404541622e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14661792,
      "step": 25810
    },
    {
      "epoch": 452.9026548672566,
      "grad_norm": 1.4152708445180906e-06,
      "learning_rate": 1.3976841584998513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14664272,
      "step": 25815
    },
    {
      "epoch": 452.9911504424779,
      "grad_norm": 3.0938344934838824e-06,
      "learning_rate": 1.3968030824490352e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14666800,
      "step": 25820
    },
    {
      "epoch": 453.070796460177,
      "grad_norm": 2.9292873477970716e-06,
      "learning_rate": 1.3959221765250469e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14669096,
      "step": 25825
    },
    {
      "epoch": 453.1592920353982,
      "grad_norm": 1.2686640502579394e-06,
      "learning_rate": 1.3950414408637343e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14672008,
      "step": 25830
    },
    {
      "epoch": 453.24778761061947,
      "grad_norm": 1.0823898719536373e-06,
      "learning_rate": 1.3941608756009166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14674920,
      "step": 25835
    },
    {
      "epoch": 453.3362831858407,
      "grad_norm": 1.131339445237245e-06,
      "learning_rate": 1.3932804808723898e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14678232,
      "step": 25840
    },
    {
      "epoch": 453.42477876106193,
      "grad_norm": 1.5071759662532713e-06,
      "learning_rate": 1.3924002568139194e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14680744,
      "step": 25845
    },
    {
      "epoch": 453.5132743362832,
      "grad_norm": 5.024903202865971e-06,
      "learning_rate": 1.3915202035612485e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14683560,
      "step": 25850
    },
    {
      "epoch": 453.60176991150445,
      "grad_norm": 6.0302572819637135e-06,
      "learning_rate": 1.3906403212500935e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14686808,
      "step": 25855
    },
    {
      "epoch": 453.69026548672565,
      "grad_norm": 1.178371462629002e-06,
      "learning_rate": 1.3897606100161409e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14689640,
      "step": 25860
    },
    {
      "epoch": 453.7787610619469,
      "grad_norm": 3.521748567436589e-06,
      "learning_rate": 1.388881069995055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14692248,
      "step": 25865
    },
    {
      "epoch": 453.86725663716817,
      "grad_norm": 3.767877160498756e-06,
      "learning_rate": 1.3880017013224708e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14695032,
      "step": 25870
    },
    {
      "epoch": 453.95575221238937,
      "grad_norm": 6.160689167700184e-07,
      "learning_rate": 1.3871225041339984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14697960,
      "step": 25875
    },
    {
      "epoch": 454.0353982300885,
      "grad_norm": 1.4384436326508876e-06,
      "learning_rate": 1.386243478565222e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14700632,
      "step": 25880
    },
    {
      "epoch": 454.12389380530976,
      "grad_norm": 2.8930292046425166e-06,
      "learning_rate": 1.3853646247516966e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14703112,
      "step": 25885
    },
    {
      "epoch": 454.21238938053096,
      "grad_norm": 1.0330741133657284e-06,
      "learning_rate": 1.3844859428289545e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14705736,
      "step": 25890
    },
    {
      "epoch": 454.3008849557522,
      "grad_norm": 1.5480965203096275e-06,
      "learning_rate": 1.3836074329324984e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14708888,
      "step": 25895
    },
    {
      "epoch": 454.3893805309734,
      "grad_norm": 1.3166140888642985e-06,
      "learning_rate": 1.3827290951978044e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14711496,
      "step": 25900
    },
    {
      "epoch": 454.4778761061947,
      "grad_norm": 3.85509520128835e-06,
      "learning_rate": 1.381850929760326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14713992,
      "step": 25905
    },
    {
      "epoch": 454.56637168141594,
      "grad_norm": 3.1351969482784625e-06,
      "learning_rate": 1.3809729367554842e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14716824,
      "step": 25910
    },
    {
      "epoch": 454.65486725663715,
      "grad_norm": 2.8798144739994314e-06,
      "learning_rate": 1.3800951163186784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14719784,
      "step": 25915
    },
    {
      "epoch": 454.7433628318584,
      "grad_norm": 8.3408923501338e-07,
      "learning_rate": 1.3792174685852801e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14722456,
      "step": 25920
    },
    {
      "epoch": 454.83185840707966,
      "grad_norm": 5.0795106290024705e-06,
      "learning_rate": 1.378339993690632e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14725944,
      "step": 25925
    },
    {
      "epoch": 454.92035398230087,
      "grad_norm": 9.272289389627986e-06,
      "learning_rate": 1.3774626917700523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14728984,
      "step": 25930
    },
    {
      "epoch": 455.0,
      "grad_norm": 1.1123199783469317e-06,
      "learning_rate": 1.3765855629588334e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14731520,
      "step": 25935
    },
    {
      "epoch": 455.08849557522126,
      "grad_norm": 5.390130581872654e-07,
      "learning_rate": 1.3757086073922374e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14733952,
      "step": 25940
    },
    {
      "epoch": 455.17699115044246,
      "grad_norm": 2.2905805963091552e-06,
      "learning_rate": 1.3748318252055038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14737200,
      "step": 25945
    },
    {
      "epoch": 455.2654867256637,
      "grad_norm": 8.178972166206222e-07,
      "learning_rate": 1.3739552165338416e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14740208,
      "step": 25950
    },
    {
      "epoch": 455.353982300885,
      "grad_norm": 9.362160540149489e-07,
      "learning_rate": 1.3730787815124354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14742944,
      "step": 25955
    },
    {
      "epoch": 455.4424778761062,
      "grad_norm": 1.5403669522129348e-06,
      "learning_rate": 1.3722025202764443e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14745824,
      "step": 25960
    },
    {
      "epoch": 455.53097345132744,
      "grad_norm": 2.1626317447953625e-06,
      "learning_rate": 1.371326432960997e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14748576,
      "step": 25965
    },
    {
      "epoch": 455.6194690265487,
      "grad_norm": 2.9820914733136306e-06,
      "learning_rate": 1.3704505197011969e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14750992,
      "step": 25970
    },
    {
      "epoch": 455.7079646017699,
      "grad_norm": 1.5479467947443482e-06,
      "learning_rate": 1.3695747806321224e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14753936,
      "step": 25975
    },
    {
      "epoch": 455.79646017699116,
      "grad_norm": 2.5391473172931e-06,
      "learning_rate": 1.3686992158888212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14756976,
      "step": 25980
    },
    {
      "epoch": 455.88495575221236,
      "grad_norm": 2.9246084523038007e-06,
      "learning_rate": 1.367823825606319e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14760288,
      "step": 25985
    },
    {
      "epoch": 455.9734513274336,
      "grad_norm": 1.5139313518375275e-06,
      "learning_rate": 1.36694860991961e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14763120,
      "step": 25990
    },
    {
      "epoch": 456.05309734513276,
      "grad_norm": 3.1303359264711617e-06,
      "learning_rate": 1.3660735689636636e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14765480,
      "step": 25995
    },
    {
      "epoch": 456.14159292035396,
      "grad_norm": 2.5613339857954998e-06,
      "learning_rate": 1.365198702873424e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14768168,
      "step": 26000
    },
    {
      "epoch": 456.14159292035396,
      "eval_loss": 0.5152615904808044,
      "eval_runtime": 1.0826,
      "eval_samples_per_second": 23.092,
      "eval_steps_per_second": 12.008,
      "num_input_tokens_seen": 14768168,
      "step": 26000
    },
    {
      "epoch": 456.2300884955752,
      "grad_norm": 4.766399797517806e-06,
      "learning_rate": 1.364324011783804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14770904,
      "step": 26005
    },
    {
      "epoch": 456.3185840707965,
      "grad_norm": 1.1509774822116015e-06,
      "learning_rate": 1.3634494958296934e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14774024,
      "step": 26010
    },
    {
      "epoch": 456.4070796460177,
      "grad_norm": 8.249481311395357e-07,
      "learning_rate": 1.3625751551459542e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14777208,
      "step": 26015
    },
    {
      "epoch": 456.49557522123894,
      "grad_norm": 5.775498266302748e-07,
      "learning_rate": 1.3617009898674188e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14779928,
      "step": 26020
    },
    {
      "epoch": 456.5840707964602,
      "grad_norm": 3.130288860120345e-06,
      "learning_rate": 1.3608270001288967e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14783064,
      "step": 26025
    },
    {
      "epoch": 456.6725663716814,
      "grad_norm": 2.737471731961705e-06,
      "learning_rate": 1.359953186065166e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14786072,
      "step": 26030
    },
    {
      "epoch": 456.76106194690266,
      "grad_norm": 2.01050079340348e-06,
      "learning_rate": 1.3590795478109814e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14788584,
      "step": 26035
    },
    {
      "epoch": 456.8495575221239,
      "grad_norm": 1.287816076001036e-06,
      "learning_rate": 1.3582060855010675e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14791480,
      "step": 26040
    },
    {
      "epoch": 456.9380530973451,
      "grad_norm": 1.4686180520584458e-06,
      "learning_rate": 1.3573327992701245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14794696,
      "step": 26045
    },
    {
      "epoch": 457.01769911504425,
      "grad_norm": 3.7464581055246526e-06,
      "learning_rate": 1.356459689252823e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14796992,
      "step": 26050
    },
    {
      "epoch": 457.1061946902655,
      "grad_norm": 1.1389452083676588e-06,
      "learning_rate": 1.3555867555838087e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14800080,
      "step": 26055
    },
    {
      "epoch": 457.1946902654867,
      "grad_norm": 1.1528987897690968e-06,
      "learning_rate": 1.3547139983976975e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14802624,
      "step": 26060
    },
    {
      "epoch": 457.283185840708,
      "grad_norm": 3.337176394779817e-06,
      "learning_rate": 1.3538414178290815e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14805408,
      "step": 26065
    },
    {
      "epoch": 457.37168141592923,
      "grad_norm": 4.262339189153863e-06,
      "learning_rate": 1.3529690140125209e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14808416,
      "step": 26070
    },
    {
      "epoch": 457.46017699115043,
      "grad_norm": 1.4561693433279288e-06,
      "learning_rate": 1.352096787082553e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14810992,
      "step": 26075
    },
    {
      "epoch": 457.5486725663717,
      "grad_norm": 9.506280775894993e-07,
      "learning_rate": 1.3512247371736871e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14813776,
      "step": 26080
    },
    {
      "epoch": 457.6371681415929,
      "grad_norm": 1.0550575098022819e-06,
      "learning_rate": 1.3503528644204022e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14817152,
      "step": 26085
    },
    {
      "epoch": 457.72566371681415,
      "grad_norm": 1.4561491070708144e-06,
      "learning_rate": 1.349481168957153e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14819984,
      "step": 26090
    },
    {
      "epoch": 457.8141592920354,
      "grad_norm": 9.995181926569785e-07,
      "learning_rate": 1.3486096509183665e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14823344,
      "step": 26095
    },
    {
      "epoch": 457.9026548672566,
      "grad_norm": 1.2885766409453936e-06,
      "learning_rate": 1.3477383104384406e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14826016,
      "step": 26100
    },
    {
      "epoch": 457.9911504424779,
      "grad_norm": 3.514806849125307e-06,
      "learning_rate": 1.3468671476517481e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14828992,
      "step": 26105
    },
    {
      "epoch": 458.070796460177,
      "grad_norm": 2.9870002435927745e-06,
      "learning_rate": 1.3459961626926326e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14831296,
      "step": 26110
    },
    {
      "epoch": 458.1592920353982,
      "grad_norm": 7.56469626139733e-06,
      "learning_rate": 1.3451253556954101e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14834464,
      "step": 26115
    },
    {
      "epoch": 458.24778761061947,
      "grad_norm": 1.8807733113135328e-06,
      "learning_rate": 1.3442547267943717e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14837632,
      "step": 26120
    },
    {
      "epoch": 458.3362831858407,
      "grad_norm": 1.2203818187117577e-06,
      "learning_rate": 1.3433842761237774e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14840336,
      "step": 26125
    },
    {
      "epoch": 458.42477876106193,
      "grad_norm": 2.6660491130314767e-06,
      "learning_rate": 1.3425140038178639e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14842496,
      "step": 26130
    },
    {
      "epoch": 458.5132743362832,
      "grad_norm": 9.936665037457715e-07,
      "learning_rate": 1.3416439100108358e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14845360,
      "step": 26135
    },
    {
      "epoch": 458.60176991150445,
      "grad_norm": 1.1663286159091513e-06,
      "learning_rate": 1.3407739948368734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14848080,
      "step": 26140
    },
    {
      "epoch": 458.69026548672565,
      "grad_norm": 2.4634327928652056e-06,
      "learning_rate": 1.3399042584301298e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14850864,
      "step": 26145
    },
    {
      "epoch": 458.7787610619469,
      "grad_norm": 1.1130739494547015e-06,
      "learning_rate": 1.3390347009247272e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14854176,
      "step": 26150
    },
    {
      "epoch": 458.86725663716817,
      "grad_norm": 1.364682475468726e-06,
      "learning_rate": 1.3381653224547635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14857696,
      "step": 26155
    },
    {
      "epoch": 458.95575221238937,
      "grad_norm": 9.813114729695371e-07,
      "learning_rate": 1.3372961231543086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14860256,
      "step": 26160
    },
    {
      "epoch": 459.0353982300885,
      "grad_norm": 1.2293793361095595e-06,
      "learning_rate": 1.3364271031574016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14862496,
      "step": 26165
    },
    {
      "epoch": 459.12389380530976,
      "grad_norm": 1.9366477772564394e-06,
      "learning_rate": 1.335558262598059e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14865568,
      "step": 26170
    },
    {
      "epoch": 459.21238938053096,
      "grad_norm": 1.058142402143858e-06,
      "learning_rate": 1.3346896016102645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14868448,
      "step": 26175
    },
    {
      "epoch": 459.3008849557522,
      "grad_norm": 3.5574807952798437e-06,
      "learning_rate": 1.3338211203279788e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14871296,
      "step": 26180
    },
    {
      "epoch": 459.3893805309734,
      "grad_norm": 5.615014288196107e-06,
      "learning_rate": 1.3329528188851303e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14873776,
      "step": 26185
    },
    {
      "epoch": 459.4778761061947,
      "grad_norm": 1.260948238268611e-06,
      "learning_rate": 1.3320846974156242e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14876544,
      "step": 26190
    },
    {
      "epoch": 459.56637168141594,
      "grad_norm": 1.0353450079492177e-06,
      "learning_rate": 1.3312167560533337e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14879248,
      "step": 26195
    },
    {
      "epoch": 459.65486725663715,
      "grad_norm": 1.593697447788145e-06,
      "learning_rate": 1.3303489949321082e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14882048,
      "step": 26200
    },
    {
      "epoch": 459.65486725663715,
      "eval_loss": 0.5005371570587158,
      "eval_runtime": 1.065,
      "eval_samples_per_second": 23.474,
      "eval_steps_per_second": 12.206,
      "num_input_tokens_seen": 14882048,
      "step": 26200
    },
    {
      "epoch": 459.7433628318584,
      "grad_norm": 1.0975885516018025e-06,
      "learning_rate": 1.3294814141857653e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14885712,
      "step": 26205
    },
    {
      "epoch": 459.83185840707966,
      "grad_norm": 1.1323868420731742e-06,
      "learning_rate": 1.3286140139480992e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14888896,
      "step": 26210
    },
    {
      "epoch": 459.92035398230087,
      "grad_norm": 1.4115217936705449e-06,
      "learning_rate": 1.3277467943528719e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14891792,
      "step": 26215
    },
    {
      "epoch": 460.0,
      "grad_norm": 2.1462662971316604e-06,
      "learning_rate": 1.3268797555338203e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14893968,
      "step": 26220
    },
    {
      "epoch": 460.08849557522126,
      "grad_norm": 2.9441389415296726e-06,
      "learning_rate": 1.3260128976246533e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14897008,
      "step": 26225
    },
    {
      "epoch": 460.17699115044246,
      "grad_norm": 1.8390510376775637e-06,
      "learning_rate": 1.32514622075905e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14899680,
      "step": 26230
    },
    {
      "epoch": 460.2654867256637,
      "grad_norm": 1.1482233048809576e-06,
      "learning_rate": 1.3242797250706638e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14902800,
      "step": 26235
    },
    {
      "epoch": 460.353982300885,
      "grad_norm": 6.947760766706779e-07,
      "learning_rate": 1.3234134106931195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14905440,
      "step": 26240
    },
    {
      "epoch": 460.4424778761062,
      "grad_norm": 1.5427042399096536e-06,
      "learning_rate": 1.322547277760013e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14908336,
      "step": 26245
    },
    {
      "epoch": 460.53097345132744,
      "grad_norm": 6.053742822587083e-07,
      "learning_rate": 1.3216813264049132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14910976,
      "step": 26250
    },
    {
      "epoch": 460.6194690265487,
      "grad_norm": 8.48220952320844e-06,
      "learning_rate": 1.32081555676136e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14913952,
      "step": 26255
    },
    {
      "epoch": 460.7079646017699,
      "grad_norm": 9.675048886492732e-07,
      "learning_rate": 1.3199499689628674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14917056,
      "step": 26260
    },
    {
      "epoch": 460.79646017699116,
      "grad_norm": 1.1136733064631699e-06,
      "learning_rate": 1.3190845631429192e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14919856,
      "step": 26265
    },
    {
      "epoch": 460.88495575221236,
      "grad_norm": 2.3072946078173118e-06,
      "learning_rate": 1.3182193394349704e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14922096,
      "step": 26270
    },
    {
      "epoch": 460.9734513274336,
      "grad_norm": 1.2908827784485766e-06,
      "learning_rate": 1.3173542979724507e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14925632,
      "step": 26275
    },
    {
      "epoch": 461.05309734513276,
      "grad_norm": 2.669148216227768e-06,
      "learning_rate": 1.3164894388887617e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14927848,
      "step": 26280
    },
    {
      "epoch": 461.14159292035396,
      "grad_norm": 8.608521397945879e-07,
      "learning_rate": 1.3156247623172727e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14930584,
      "step": 26285
    },
    {
      "epoch": 461.2300884955752,
      "grad_norm": 1.3481890164257493e-06,
      "learning_rate": 1.3147602683913302e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14933784,
      "step": 26290
    },
    {
      "epoch": 461.3185840707965,
      "grad_norm": 1.169881215901114e-06,
      "learning_rate": 1.3138959572442481e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14936216,
      "step": 26295
    },
    {
      "epoch": 461.4070796460177,
      "grad_norm": 6.57657437841408e-07,
      "learning_rate": 1.3130318290093146e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14939304,
      "step": 26300
    },
    {
      "epoch": 461.49557522123894,
      "grad_norm": 1.2450194617485977e-06,
      "learning_rate": 1.3121678838197909e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14942392,
      "step": 26305
    },
    {
      "epoch": 461.5840707964602,
      "grad_norm": 1.879177261798759e-06,
      "learning_rate": 1.3113041218089056e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14945288,
      "step": 26310
    },
    {
      "epoch": 461.6725663716814,
      "grad_norm": 1.0368863740950474e-06,
      "learning_rate": 1.3104405431098626e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14948376,
      "step": 26315
    },
    {
      "epoch": 461.76106194690266,
      "grad_norm": 1.2272288358872174e-06,
      "learning_rate": 1.3095771478558377e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14951320,
      "step": 26320
    },
    {
      "epoch": 461.8495575221239,
      "grad_norm": 1.7271113392780535e-06,
      "learning_rate": 1.3087139361799766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14954360,
      "step": 26325
    },
    {
      "epoch": 461.9380530973451,
      "grad_norm": 1.4869846154397237e-06,
      "learning_rate": 1.3078509082153964e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14956952,
      "step": 26330
    },
    {
      "epoch": 462.01769911504425,
      "grad_norm": 1.3368039617489558e-06,
      "learning_rate": 1.3069880640951885e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14959064,
      "step": 26335
    },
    {
      "epoch": 462.1061946902655,
      "grad_norm": 3.9413112062902655e-06,
      "learning_rate": 1.3061254039524123e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14961480,
      "step": 26340
    },
    {
      "epoch": 462.1946902654867,
      "grad_norm": 3.7753670767415315e-06,
      "learning_rate": 1.3052629279201028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14964344,
      "step": 26345
    },
    {
      "epoch": 462.283185840708,
      "grad_norm": 1.0607103604343138e-06,
      "learning_rate": 1.3044006361312633e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14966744,
      "step": 26350
    },
    {
      "epoch": 462.37168141592923,
      "grad_norm": 1.7558249965077266e-06,
      "learning_rate": 1.30353852871887e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14970328,
      "step": 26355
    },
    {
      "epoch": 462.46017699115043,
      "grad_norm": 1.0932059240076342e-06,
      "learning_rate": 1.302676605815873e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14973160,
      "step": 26360
    },
    {
      "epoch": 462.5486725663717,
      "grad_norm": 1.8763838625091012e-06,
      "learning_rate": 1.3018148675551884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14976136,
      "step": 26365
    },
    {
      "epoch": 462.6371681415929,
      "grad_norm": 1.1319870054649073e-06,
      "learning_rate": 1.3009533140697094e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14978712,
      "step": 26370
    },
    {
      "epoch": 462.72566371681415,
      "grad_norm": 2.547127905927482e-06,
      "learning_rate": 1.3000919454922966e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14981528,
      "step": 26375
    },
    {
      "epoch": 462.8141592920354,
      "grad_norm": 3.185660034432658e-06,
      "learning_rate": 1.299230761955785e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14984488,
      "step": 26380
    },
    {
      "epoch": 462.9026548672566,
      "grad_norm": 2.801275968522532e-06,
      "learning_rate": 1.2983697635929807e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14987208,
      "step": 26385
    },
    {
      "epoch": 462.9911504424779,
      "grad_norm": 1.2130682307542884e-06,
      "learning_rate": 1.2975089505366584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14990504,
      "step": 26390
    },
    {
      "epoch": 463.070796460177,
      "grad_norm": 1.1525456784511334e-06,
      "learning_rate": 1.2966483229195683e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14992888,
      "step": 26395
    },
    {
      "epoch": 463.1592920353982,
      "grad_norm": 2.9159346013329923e-06,
      "learning_rate": 1.2957878808744283e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14996008,
      "step": 26400
    },
    {
      "epoch": 463.1592920353982,
      "eval_loss": 0.5167677402496338,
      "eval_runtime": 1.0736,
      "eval_samples_per_second": 23.286,
      "eval_steps_per_second": 12.109,
      "num_input_tokens_seen": 14996008,
      "step": 26400
    },
    {
      "epoch": 463.24778761061947,
      "grad_norm": 2.699227934499504e-06,
      "learning_rate": 1.294927624533931e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 14998904,
      "step": 26405
    },
    {
      "epoch": 463.3362831858407,
      "grad_norm": 2.596830881884671e-06,
      "learning_rate": 1.2940675540307378e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15001768,
      "step": 26410
    },
    {
      "epoch": 463.42477876106193,
      "grad_norm": 8.613019417680334e-07,
      "learning_rate": 1.2932076694974814e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15004696,
      "step": 26415
    },
    {
      "epoch": 463.5132743362832,
      "grad_norm": 1.861246460066468e-06,
      "learning_rate": 1.2923479710667682e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15007400,
      "step": 26420
    },
    {
      "epoch": 463.60176991150445,
      "grad_norm": 7.018377914391749e-07,
      "learning_rate": 1.2914884588711751e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 15010040,
      "step": 26425
    },
    {
      "epoch": 463.69026548672565,
      "grad_norm": 0.8978846669197083,
      "learning_rate": 1.2906291330432475e-05,
      "loss": 0.2589,
      "num_input_tokens_seen": 15012792,
      "step": 26430
    },
    {
      "epoch": 463.7787610619469,
      "grad_norm": 0.004068547859787941,
      "learning_rate": 1.2897699937155055e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15015944,
      "step": 26435
    },
    {
      "epoch": 463.86725663716817,
      "grad_norm": 0.00032665274920873344,
      "learning_rate": 1.2889110410204403e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 15018952,
      "step": 26440
    },
    {
      "epoch": 463.95575221238937,
      "grad_norm": 2.1734662368544377e-05,
      "learning_rate": 1.2880522750905111e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15021608,
      "step": 26445
    },
    {
      "epoch": 464.0353982300885,
      "grad_norm": 6.360437510011252e-06,
      "learning_rate": 1.2871936960581523e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15024248,
      "step": 26450
    },
    {
      "epoch": 464.12389380530976,
      "grad_norm": 1.0574379302852321e-05,
      "learning_rate": 1.2863353040557658e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15027448,
      "step": 26455
    },
    {
      "epoch": 464.21238938053096,
      "grad_norm": 5.732876616093563e-06,
      "learning_rate": 1.2854770992157273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15030568,
      "step": 26460
    },
    {
      "epoch": 464.3008849557522,
      "grad_norm": 2.6342850105720572e-05,
      "learning_rate": 1.2846190816703835e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15032920,
      "step": 26465
    },
    {
      "epoch": 464.3893805309734,
      "grad_norm": 1.4654239748779219e-05,
      "learning_rate": 1.2837612515520498e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15036120,
      "step": 26470
    },
    {
      "epoch": 464.4778761061947,
      "grad_norm": 0.0015227219555526972,
      "learning_rate": 1.2829036089930163e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15039080,
      "step": 26475
    },
    {
      "epoch": 464.56637168141594,
      "grad_norm": 0.035120971500873566,
      "learning_rate": 1.2820461541255412e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15041496,
      "step": 26480
    },
    {
      "epoch": 464.65486725663715,
      "grad_norm": 0.0010651787742972374,
      "learning_rate": 1.2811888870818543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15044024,
      "step": 26485
    },
    {
      "epoch": 464.7433628318584,
      "grad_norm": 1.3459843103191815e-05,
      "learning_rate": 1.2803318079941581e-05,
      "loss": 0.0904,
      "num_input_tokens_seen": 15046536,
      "step": 26490
    },
    {
      "epoch": 464.83185840707966,
      "grad_norm": 0.0053971935994923115,
      "learning_rate": 1.2794749169946235e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15050024,
      "step": 26495
    },
    {
      "epoch": 464.92035398230087,
      "grad_norm": 1.1763980864998302e-06,
      "learning_rate": 1.2786182142153952e-05,
      "loss": 0.1704,
      "num_input_tokens_seen": 15053384,
      "step": 26500
    },
    {
      "epoch": 465.0,
      "grad_norm": 2.413292349956464e-05,
      "learning_rate": 1.2777616997885878e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15055576,
      "step": 26505
    },
    {
      "epoch": 465.08849557522126,
      "grad_norm": 4.552322207018733e-05,
      "learning_rate": 1.2769053738462847e-05,
      "loss": 0.2264,
      "num_input_tokens_seen": 15058232,
      "step": 26510
    },
    {
      "epoch": 465.17699115044246,
      "grad_norm": 1.0915108532572049e-06,
      "learning_rate": 1.2760492365205434e-05,
      "loss": 0.1045,
      "num_input_tokens_seen": 15061512,
      "step": 26515
    },
    {
      "epoch": 465.2654867256637,
      "grad_norm": 8.404960567531816e-07,
      "learning_rate": 1.2751932879433919e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15064536,
      "step": 26520
    },
    {
      "epoch": 465.353982300885,
      "grad_norm": 5.160314685781486e-05,
      "learning_rate": 1.2743375282468267e-05,
      "loss": 0.2155,
      "num_input_tokens_seen": 15067416,
      "step": 26525
    },
    {
      "epoch": 465.4424778761062,
      "grad_norm": 6.08109303357196e-06,
      "learning_rate": 1.2734819575628182e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15070200,
      "step": 26530
    },
    {
      "epoch": 465.53097345132744,
      "grad_norm": 0.0006843023584224284,
      "learning_rate": 1.2726265760233039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15072808,
      "step": 26535
    },
    {
      "epoch": 465.6194690265487,
      "grad_norm": 1.762384817993734e-05,
      "learning_rate": 1.271771383760197e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15075800,
      "step": 26540
    },
    {
      "epoch": 465.7079646017699,
      "grad_norm": 6.921401109138969e-06,
      "learning_rate": 1.2709163809053764e-05,
      "loss": 0.0001,
      "num_input_tokens_seen": 15078232,
      "step": 26545
    },
    {
      "epoch": 465.79646017699116,
      "grad_norm": 7.967484634718858e-06,
      "learning_rate": 1.2700615675906963e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15081096,
      "step": 26550
    },
    {
      "epoch": 465.88495575221236,
      "grad_norm": 8.723499377083499e-06,
      "learning_rate": 1.269206943947978e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15084616,
      "step": 26555
    },
    {
      "epoch": 465.9734513274336,
      "grad_norm": 3.8560152461286634e-05,
      "learning_rate": 1.2683525101090177e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15087160,
      "step": 26560
    },
    {
      "epoch": 466.05309734513276,
      "grad_norm": 1.7556625380166224e-06,
      "learning_rate": 1.2674982662055765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15089240,
      "step": 26565
    },
    {
      "epoch": 466.14159292035396,
      "grad_norm": 1.635164335311856e-05,
      "learning_rate": 1.2666442123693922e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15092200,
      "step": 26570
    },
    {
      "epoch": 466.2300884955752,
      "grad_norm": 7.849899702705443e-06,
      "learning_rate": 1.265790348732169e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15094808,
      "step": 26575
    },
    {
      "epoch": 466.3185840707965,
      "grad_norm": 1.280233027500799e-05,
      "learning_rate": 1.264936675425584e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15097560,
      "step": 26580
    },
    {
      "epoch": 466.4070796460177,
      "grad_norm": 2.0700201275758445e-05,
      "learning_rate": 1.2640831925812852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15100824,
      "step": 26585
    },
    {
      "epoch": 466.49557522123894,
      "grad_norm": 2.516400854801759e-06,
      "learning_rate": 1.263229900330889e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15103368,
      "step": 26590
    },
    {
      "epoch": 466.5840707964602,
      "grad_norm": 5.211844381847186e-06,
      "learning_rate": 1.2623767988059843e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15106504,
      "step": 26595
    },
    {
      "epoch": 466.6725663716814,
      "grad_norm": 1.0983038009726442e-05,
      "learning_rate": 1.2615238881381309e-05,
      "loss": 0.139,
      "num_input_tokens_seen": 15109352,
      "step": 26600
    },
    {
      "epoch": 466.6725663716814,
      "eval_loss": 0.8271105885505676,
      "eval_runtime": 1.0711,
      "eval_samples_per_second": 23.341,
      "eval_steps_per_second": 12.138,
      "num_input_tokens_seen": 15109352,
      "step": 26600
    },
    {
      "epoch": 466.76106194690266,
      "grad_norm": 5.224718734098133e-06,
      "learning_rate": 1.2606711684588568e-05,
      "loss": 0.0005,
      "num_input_tokens_seen": 15112392,
      "step": 26605
    },
    {
      "epoch": 466.8495575221239,
      "grad_norm": 6.59678903502936e-07,
      "learning_rate": 1.2598186398996636e-05,
      "loss": 0.1392,
      "num_input_tokens_seen": 15115688,
      "step": 26610
    },
    {
      "epoch": 466.9380530973451,
      "grad_norm": 1.0046042916656006e-05,
      "learning_rate": 1.2589663025920207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15118104,
      "step": 26615
    },
    {
      "epoch": 467.01769911504425,
      "grad_norm": 1.5686231336076162e-06,
      "learning_rate": 1.2581141566673705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15120496,
      "step": 26620
    },
    {
      "epoch": 467.1061946902655,
      "grad_norm": 1.0685114375519333e-06,
      "learning_rate": 1.257262202257124e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15123328,
      "step": 26625
    },
    {
      "epoch": 467.1946902654867,
      "grad_norm": 2.0221320937707787e-06,
      "learning_rate": 1.2564104394926618e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15125856,
      "step": 26630
    },
    {
      "epoch": 467.283185840708,
      "grad_norm": 0.0010105854598805308,
      "learning_rate": 1.2555588685053383e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15128512,
      "step": 26635
    },
    {
      "epoch": 467.37168141592923,
      "grad_norm": 0.6982900500297546,
      "learning_rate": 1.2547074894264762e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 15131520,
      "step": 26640
    },
    {
      "epoch": 467.46017699115043,
      "grad_norm": 7.659015750505205e-07,
      "learning_rate": 1.2538563023873679e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15134128,
      "step": 26645
    },
    {
      "epoch": 467.5486725663717,
      "grad_norm": 7.223436568892794e-07,
      "learning_rate": 1.2530053075192789e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15136816,
      "step": 26650
    },
    {
      "epoch": 467.6371681415929,
      "grad_norm": 5.44935176094441e-07,
      "learning_rate": 1.252154504953441e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15140256,
      "step": 26655
    },
    {
      "epoch": 467.72566371681415,
      "grad_norm": 0.05883975699543953,
      "learning_rate": 1.25130389482106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15143248,
      "step": 26660
    },
    {
      "epoch": 467.8141592920354,
      "grad_norm": 4.759689886668639e-07,
      "learning_rate": 1.2504534772533116e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15146240,
      "step": 26665
    },
    {
      "epoch": 467.9026548672566,
      "grad_norm": 7.577697260785499e-07,
      "learning_rate": 1.2496032523813387e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15149520,
      "step": 26670
    },
    {
      "epoch": 467.9911504424779,
      "grad_norm": 1.3266790119814686e-06,
      "learning_rate": 1.2487532203362576e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15152272,
      "step": 26675
    },
    {
      "epoch": 468.070796460177,
      "grad_norm": 3.879926225636154e-06,
      "learning_rate": 1.247903381249155e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15155144,
      "step": 26680
    },
    {
      "epoch": 468.1592920353982,
      "grad_norm": 9.523677135803155e-07,
      "learning_rate": 1.2470537352510853e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15158280,
      "step": 26685
    },
    {
      "epoch": 468.24778761061947,
      "grad_norm": 8.757662044445169e-07,
      "learning_rate": 1.2462042824730758e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15161736,
      "step": 26690
    },
    {
      "epoch": 468.3362831858407,
      "grad_norm": 1.7692235587674077e-06,
      "learning_rate": 1.245355023046122e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15164072,
      "step": 26695
    },
    {
      "epoch": 468.42477876106193,
      "grad_norm": 2.9100371534696023e-07,
      "learning_rate": 1.2445059571011896e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15167192,
      "step": 26700
    },
    {
      "epoch": 468.5132743362832,
      "grad_norm": 9.235368452209514e-07,
      "learning_rate": 1.2436570847692173e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15170264,
      "step": 26705
    },
    {
      "epoch": 468.60176991150445,
      "grad_norm": 1.7959689557756064e-06,
      "learning_rate": 1.2428084061811096e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15173768,
      "step": 26710
    },
    {
      "epoch": 468.69026548672565,
      "grad_norm": 1.2862342373409774e-05,
      "learning_rate": 1.2419599214677447e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15176584,
      "step": 26715
    },
    {
      "epoch": 468.7787610619469,
      "grad_norm": 5.715594397770474e-06,
      "learning_rate": 1.2411116307599702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15179032,
      "step": 26720
    },
    {
      "epoch": 468.86725663716817,
      "grad_norm": 6.629103154409677e-06,
      "learning_rate": 1.2402635341886016e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15181624,
      "step": 26725
    },
    {
      "epoch": 468.95575221238937,
      "grad_norm": 3.8387372569559375e-07,
      "learning_rate": 1.2394156318844278e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15184088,
      "step": 26730
    },
    {
      "epoch": 469.0353982300885,
      "grad_norm": 3.018797087861458e-07,
      "learning_rate": 1.2385679239782039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15186016,
      "step": 26735
    },
    {
      "epoch": 469.12389380530976,
      "grad_norm": 6.8011650000698864e-06,
      "learning_rate": 1.2377204106006585e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15188768,
      "step": 26740
    },
    {
      "epoch": 469.21238938053096,
      "grad_norm": 1.1413852689656778e-06,
      "learning_rate": 1.2368730918824891e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15191872,
      "step": 26745
    },
    {
      "epoch": 469.3008849557522,
      "grad_norm": 2.1458743049151963e-06,
      "learning_rate": 1.236025967954362e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15194592,
      "step": 26750
    },
    {
      "epoch": 469.3893805309734,
      "grad_norm": 1.3124082443027874e-06,
      "learning_rate": 1.2351790389469153e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15197680,
      "step": 26755
    },
    {
      "epoch": 469.4778761061947,
      "grad_norm": 3.604701760195894e-06,
      "learning_rate": 1.234332304990755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15200560,
      "step": 26760
    },
    {
      "epoch": 469.56637168141594,
      "grad_norm": 8.71788233780535e-06,
      "learning_rate": 1.2334857662164593e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15203216,
      "step": 26765
    },
    {
      "epoch": 469.65486725663715,
      "grad_norm": 4.6006445586499467e-07,
      "learning_rate": 1.2326394227545743e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15205920,
      "step": 26770
    },
    {
      "epoch": 469.7433628318584,
      "grad_norm": 1.475809995099553e-06,
      "learning_rate": 1.2317932747356162e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15208880,
      "step": 26775
    },
    {
      "epoch": 469.83185840707966,
      "grad_norm": 4.92734613999346e-07,
      "learning_rate": 1.2309473222900726e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15211968,
      "step": 26780
    },
    {
      "epoch": 469.92035398230087,
      "grad_norm": 1.2195021099614678e-06,
      "learning_rate": 1.2301015655484006e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15215120,
      "step": 26785
    },
    {
      "epoch": 470.0,
      "grad_norm": 1.663189209466509e-06,
      "learning_rate": 1.2292560046410245e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15217640,
      "step": 26790
    },
    {
      "epoch": 470.08849557522126,
      "grad_norm": 2.6272815034644736e-07,
      "learning_rate": 1.228410639698343e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15220568,
      "step": 26795
    },
    {
      "epoch": 470.17699115044246,
      "grad_norm": 2.4470114112773445e-06,
      "learning_rate": 1.2275654708507195e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15223592,
      "step": 26800
    },
    {
      "epoch": 470.17699115044246,
      "eval_loss": 0.9104436039924622,
      "eval_runtime": 1.0642,
      "eval_samples_per_second": 23.492,
      "eval_steps_per_second": 12.216,
      "num_input_tokens_seen": 15223592,
      "step": 26800
    },
    {
      "epoch": 470.2654867256637,
      "grad_norm": 2.584563105756388e-07,
      "learning_rate": 1.2267204982284908e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15226888,
      "step": 26805
    },
    {
      "epoch": 470.353982300885,
      "grad_norm": 1.2032493259539478e-06,
      "learning_rate": 1.2258757219619635e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15229496,
      "step": 26810
    },
    {
      "epoch": 470.4424778761062,
      "grad_norm": 1.0254561857436784e-06,
      "learning_rate": 1.2250311421814104e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15232088,
      "step": 26815
    },
    {
      "epoch": 470.53097345132744,
      "grad_norm": 1.4846754083919222e-06,
      "learning_rate": 1.2241867590170772e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15235576,
      "step": 26820
    },
    {
      "epoch": 470.6194690265487,
      "grad_norm": 5.137064817972714e-07,
      "learning_rate": 1.2233425725991799e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15238120,
      "step": 26825
    },
    {
      "epoch": 470.7079646017699,
      "grad_norm": 7.736598490737379e-07,
      "learning_rate": 1.2224985830579003e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15241112,
      "step": 26830
    },
    {
      "epoch": 470.79646017699116,
      "grad_norm": 2.496550735031633e-07,
      "learning_rate": 1.2216547905233944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15243688,
      "step": 26835
    },
    {
      "epoch": 470.88495575221236,
      "grad_norm": 1.5954311720633996e-06,
      "learning_rate": 1.2208111951257842e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15246424,
      "step": 26840
    },
    {
      "epoch": 470.9734513274336,
      "grad_norm": 5.577377351073665e-07,
      "learning_rate": 1.2199677969951622e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15249320,
      "step": 26845
    },
    {
      "epoch": 471.05309734513276,
      "grad_norm": 4.765087396663148e-06,
      "learning_rate": 1.2191245962615927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15251360,
      "step": 26850
    },
    {
      "epoch": 471.14159292035396,
      "grad_norm": 4.1790170257627324e-07,
      "learning_rate": 1.218281593055106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15254256,
      "step": 26855
    },
    {
      "epoch": 471.2300884955752,
      "grad_norm": 4.105494326722692e-07,
      "learning_rate": 1.217438787505705e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15257424,
      "step": 26860
    },
    {
      "epoch": 471.3185840707965,
      "grad_norm": 6.455153993556451e-07,
      "learning_rate": 1.2165961797433615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15260464,
      "step": 26865
    },
    {
      "epoch": 471.4070796460177,
      "grad_norm": 1.8748534102996928e-06,
      "learning_rate": 1.215753769898014e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15263712,
      "step": 26870
    },
    {
      "epoch": 471.49557522123894,
      "grad_norm": 6.114307780080708e-06,
      "learning_rate": 1.2149115580995755e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15266256,
      "step": 26875
    },
    {
      "epoch": 471.5840707964602,
      "grad_norm": 3.266801229528937e-07,
      "learning_rate": 1.2140695444779227e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15269312,
      "step": 26880
    },
    {
      "epoch": 471.6725663716814,
      "grad_norm": 2.4624009142826253e-07,
      "learning_rate": 1.2132277291629066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15272032,
      "step": 26885
    },
    {
      "epoch": 471.76106194690266,
      "grad_norm": 1.5346489590228884e-06,
      "learning_rate": 1.2123861122843458e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15274912,
      "step": 26890
    },
    {
      "epoch": 471.8495575221239,
      "grad_norm": 5.905049420107389e-06,
      "learning_rate": 1.2115446939720271e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15277952,
      "step": 26895
    },
    {
      "epoch": 471.9380530973451,
      "grad_norm": 1.2243575611137203e-06,
      "learning_rate": 1.210703474355708e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15280512,
      "step": 26900
    },
    {
      "epoch": 472.01769911504425,
      "grad_norm": 1.6419686289737e-05,
      "learning_rate": 1.2098624535651164e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15282872,
      "step": 26905
    },
    {
      "epoch": 472.1061946902655,
      "grad_norm": 1.3584017324319575e-06,
      "learning_rate": 1.2090216317299477e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15286456,
      "step": 26910
    },
    {
      "epoch": 472.1946902654867,
      "grad_norm": 1.071879864866787e-06,
      "learning_rate": 1.2081810089798668e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15289368,
      "step": 26915
    },
    {
      "epoch": 472.283185840708,
      "grad_norm": 4.456307749478583e-07,
      "learning_rate": 1.2073405854445072e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15292488,
      "step": 26920
    },
    {
      "epoch": 472.37168141592923,
      "grad_norm": 5.906461524318729e-07,
      "learning_rate": 1.206500361253474e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15294952,
      "step": 26925
    },
    {
      "epoch": 472.46017699115043,
      "grad_norm": 7.087942321959417e-06,
      "learning_rate": 1.2056603365363409e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15297928,
      "step": 26930
    },
    {
      "epoch": 472.5486725663717,
      "grad_norm": 4.79069797165721e-07,
      "learning_rate": 1.2048205114226487e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15300760,
      "step": 26935
    },
    {
      "epoch": 472.6371681415929,
      "grad_norm": 4.974995135853533e-07,
      "learning_rate": 1.2039808860419102e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15303640,
      "step": 26940
    },
    {
      "epoch": 472.72566371681415,
      "grad_norm": 3.783744091379049e-07,
      "learning_rate": 1.2031414605236066e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15306424,
      "step": 26945
    },
    {
      "epoch": 472.8141592920354,
      "grad_norm": 3.6705029060613015e-07,
      "learning_rate": 1.2023022349971862e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15309560,
      "step": 26950
    },
    {
      "epoch": 472.9026548672566,
      "grad_norm": 5.530334874492837e-06,
      "learning_rate": 1.20146320959207e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15312440,
      "step": 26955
    },
    {
      "epoch": 472.9911504424779,
      "grad_norm": 1.6433889413747238e-06,
      "learning_rate": 1.2006243844376445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15314872,
      "step": 26960
    },
    {
      "epoch": 473.070796460177,
      "grad_norm": 3.952817223762395e-06,
      "learning_rate": 1.1997857596632678e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15317736,
      "step": 26965
    },
    {
      "epoch": 473.1592920353982,
      "grad_norm": 8.604496883890533e-07,
      "learning_rate": 1.1989473353982672e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15320888,
      "step": 26970
    },
    {
      "epoch": 473.24778761061947,
      "grad_norm": 2.2288218133326154e-06,
      "learning_rate": 1.198109111771937e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15323944,
      "step": 26975
    },
    {
      "epoch": 473.3362831858407,
      "grad_norm": 9.381049039802747e-07,
      "learning_rate": 1.197271088913543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15327048,
      "step": 26980
    },
    {
      "epoch": 473.42477876106193,
      "grad_norm": 2.9349106966947147e-07,
      "learning_rate": 1.1964332669523182e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15329816,
      "step": 26985
    },
    {
      "epoch": 473.5132743362832,
      "grad_norm": 8.99658516573254e-07,
      "learning_rate": 1.1955956460174645e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15332728,
      "step": 26990
    },
    {
      "epoch": 473.60176991150445,
      "grad_norm": 2.303684823345975e-06,
      "learning_rate": 1.1947582262381552e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15335192,
      "step": 26995
    },
    {
      "epoch": 473.69026548672565,
      "grad_norm": 1.8651051050255774e-06,
      "learning_rate": 1.1939210077435293e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15338072,
      "step": 27000
    },
    {
      "epoch": 473.69026548672565,
      "eval_loss": 0.9008995890617371,
      "eval_runtime": 1.064,
      "eval_samples_per_second": 23.496,
      "eval_steps_per_second": 12.218,
      "num_input_tokens_seen": 15338072,
      "step": 27000
    },
    {
      "epoch": 473.7787610619469,
      "grad_norm": 4.7488120458183403e-07,
      "learning_rate": 1.193083990662697e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15341128,
      "step": 27005
    },
    {
      "epoch": 473.86725663716817,
      "grad_norm": 4.6711704726476455e-07,
      "learning_rate": 1.192247175124738e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15343592,
      "step": 27010
    },
    {
      "epoch": 473.95575221238937,
      "grad_norm": 8.492571623719414e-07,
      "learning_rate": 1.191410561258698e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15346248,
      "step": 27015
    },
    {
      "epoch": 474.0353982300885,
      "grad_norm": 2.454014065733645e-06,
      "learning_rate": 1.1905741491935944e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15348456,
      "step": 27020
    },
    {
      "epoch": 474.12389380530976,
      "grad_norm": 1.3313945146364858e-06,
      "learning_rate": 1.1897379390584129e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15351736,
      "step": 27025
    },
    {
      "epoch": 474.21238938053096,
      "grad_norm": 2.8226213544257917e-07,
      "learning_rate": 1.1889019309821062e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15354824,
      "step": 27030
    },
    {
      "epoch": 474.3008849557522,
      "grad_norm": 5.794362891720084e-07,
      "learning_rate": 1.188066125093599e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15357480,
      "step": 27035
    },
    {
      "epoch": 474.3893805309734,
      "grad_norm": 1.8753279391603428e-06,
      "learning_rate": 1.1872305215217811e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15360664,
      "step": 27040
    },
    {
      "epoch": 474.4778761061947,
      "grad_norm": 5.001158456252597e-07,
      "learning_rate": 1.186395120395514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15363368,
      "step": 27045
    },
    {
      "epoch": 474.56637168141594,
      "grad_norm": 2.9157426979509182e-06,
      "learning_rate": 1.1855599218436283e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15366024,
      "step": 27050
    },
    {
      "epoch": 474.65486725663715,
      "grad_norm": 1.4787328836973757e-06,
      "learning_rate": 1.1847249259949209e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15368904,
      "step": 27055
    },
    {
      "epoch": 474.7433628318584,
      "grad_norm": 7.63383297908149e-07,
      "learning_rate": 1.1838901329781574e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15371720,
      "step": 27060
    },
    {
      "epoch": 474.83185840707966,
      "grad_norm": 3.9474059576605214e-07,
      "learning_rate": 1.1830555429220758e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15375016,
      "step": 27065
    },
    {
      "epoch": 474.92035398230087,
      "grad_norm": 3.860870037897257e-06,
      "learning_rate": 1.1822211559553784e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15377832,
      "step": 27070
    },
    {
      "epoch": 475.0,
      "grad_norm": 5.0859334521646815e-08,
      "learning_rate": 1.18138697220674e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15379912,
      "step": 27075
    },
    {
      "epoch": 475.08849557522126,
      "grad_norm": 7.012130254224758e-07,
      "learning_rate": 1.1805529918048e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15382568,
      "step": 27080
    },
    {
      "epoch": 475.17699115044246,
      "grad_norm": 6.994309842411894e-07,
      "learning_rate": 1.1797192148781702e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15385384,
      "step": 27085
    },
    {
      "epoch": 475.2654867256637,
      "grad_norm": 3.0257612593231897e-07,
      "learning_rate": 1.1788856415554297e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15388488,
      "step": 27090
    },
    {
      "epoch": 475.353982300885,
      "grad_norm": 1.658807718740718e-06,
      "learning_rate": 1.1780522719651249e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15391576,
      "step": 27095
    },
    {
      "epoch": 475.4424778761062,
      "grad_norm": 8.576000141147233e-07,
      "learning_rate": 1.1772191062357721e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15394008,
      "step": 27100
    },
    {
      "epoch": 475.53097345132744,
      "grad_norm": 1.2159382549725706e-06,
      "learning_rate": 1.1763861444958573e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15396680,
      "step": 27105
    },
    {
      "epoch": 475.6194690265487,
      "grad_norm": 6.774230314476881e-07,
      "learning_rate": 1.1755533868738317e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15399592,
      "step": 27110
    },
    {
      "epoch": 475.7079646017699,
      "grad_norm": 1.0339199434383772e-06,
      "learning_rate": 1.1747208334981185e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15402920,
      "step": 27115
    },
    {
      "epoch": 475.79646017699116,
      "grad_norm": 1.6682927252986701e-06,
      "learning_rate": 1.1738884844971067e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15405896,
      "step": 27120
    },
    {
      "epoch": 475.88495575221236,
      "grad_norm": 7.290584562724689e-07,
      "learning_rate": 1.1730563399991563e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15408936,
      "step": 27125
    },
    {
      "epoch": 475.9734513274336,
      "grad_norm": 6.612307288378361e-07,
      "learning_rate": 1.1722244001325938e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15411768,
      "step": 27130
    },
    {
      "epoch": 476.05309734513276,
      "grad_norm": 1.327762674918631e-06,
      "learning_rate": 1.1713926650257137e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15414264,
      "step": 27135
    },
    {
      "epoch": 476.14159292035396,
      "grad_norm": 8.031285005927202e-07,
      "learning_rate": 1.170561134806781e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15417240,
      "step": 27140
    },
    {
      "epoch": 476.2300884955752,
      "grad_norm": 2.4858047709130915e-07,
      "learning_rate": 1.1697298096040287e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15420360,
      "step": 27145
    },
    {
      "epoch": 476.3185840707965,
      "grad_norm": 1.1138273293909151e-05,
      "learning_rate": 1.1688986895456567e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15423784,
      "step": 27150
    },
    {
      "epoch": 476.4070796460177,
      "grad_norm": 5.118829449202167e-07,
      "learning_rate": 1.1680677747598349e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15426936,
      "step": 27155
    },
    {
      "epoch": 476.49557522123894,
      "grad_norm": 9.713901363284094e-07,
      "learning_rate": 1.1672370653746995e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15429864,
      "step": 27160
    },
    {
      "epoch": 476.5840707964602,
      "grad_norm": 1.50254300024244e-06,
      "learning_rate": 1.166406561518357e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15432792,
      "step": 27165
    },
    {
      "epoch": 476.6725663716814,
      "grad_norm": 6.326511083898367e-07,
      "learning_rate": 1.1655762633188826e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15436024,
      "step": 27170
    },
    {
      "epoch": 476.76106194690266,
      "grad_norm": 5.426257416729641e-07,
      "learning_rate": 1.1647461709043172e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15438648,
      "step": 27175
    },
    {
      "epoch": 476.8495575221239,
      "grad_norm": 2.1358403046178864e-07,
      "learning_rate": 1.1639162844026722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15441224,
      "step": 27180
    },
    {
      "epoch": 476.9380530973451,
      "grad_norm": 2.2177057417138712e-06,
      "learning_rate": 1.163086603941927e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15443704,
      "step": 27185
    },
    {
      "epoch": 477.01769911504425,
      "grad_norm": 9.438728625354997e-07,
      "learning_rate": 1.1622571296500273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15445760,
      "step": 27190
    },
    {
      "epoch": 477.1061946902655,
      "grad_norm": 2.0860996130522835e-07,
      "learning_rate": 1.1614278616548904e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15448704,
      "step": 27195
    },
    {
      "epoch": 477.1946902654867,
      "grad_norm": 2.2038062752471888e-07,
      "learning_rate": 1.1605988000843986e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15451312,
      "step": 27200
    },
    {
      "epoch": 477.1946902654867,
      "eval_loss": 0.9213019013404846,
      "eval_runtime": 1.0752,
      "eval_samples_per_second": 23.251,
      "eval_steps_per_second": 12.091,
      "num_input_tokens_seen": 15451312,
      "step": 27200
    },
    {
      "epoch": 477.283185840708,
      "grad_norm": 1.4439578990277369e-06,
      "learning_rate": 1.1597699450664028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15454432,
      "step": 27205
    },
    {
      "epoch": 477.37168141592923,
      "grad_norm": 1.1492545581859304e-06,
      "learning_rate": 1.1589412967287252e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15456992,
      "step": 27210
    },
    {
      "epoch": 477.46017699115043,
      "grad_norm": 1.659172568224676e-07,
      "learning_rate": 1.1581128551991514e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15460000,
      "step": 27215
    },
    {
      "epoch": 477.5486725663717,
      "grad_norm": 3.0049354791117366e-06,
      "learning_rate": 1.1572846206054383e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15462736,
      "step": 27220
    },
    {
      "epoch": 477.6371681415929,
      "grad_norm": 1.7494820667707245e-06,
      "learning_rate": 1.1564565930753113e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15465216,
      "step": 27225
    },
    {
      "epoch": 477.72566371681415,
      "grad_norm": 1.1242508435316267e-06,
      "learning_rate": 1.1556287727364606e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15468384,
      "step": 27230
    },
    {
      "epoch": 477.8141592920354,
      "grad_norm": 1.5456547544090427e-06,
      "learning_rate": 1.1548011597165489e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15471680,
      "step": 27235
    },
    {
      "epoch": 477.9026548672566,
      "grad_norm": 1.853462663348182e-07,
      "learning_rate": 1.1539737541432019e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15474256,
      "step": 27240
    },
    {
      "epoch": 477.9911504424779,
      "grad_norm": 4.390669801068725e-07,
      "learning_rate": 1.1531465561440174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15477216,
      "step": 27245
    },
    {
      "epoch": 478.070796460177,
      "grad_norm": 8.418881520810828e-07,
      "learning_rate": 1.1523195658465605e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15479632,
      "step": 27250
    },
    {
      "epoch": 478.1592920353982,
      "grad_norm": 7.14428551873425e-07,
      "learning_rate": 1.1514927833783618e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15482416,
      "step": 27255
    },
    {
      "epoch": 478.24778761061947,
      "grad_norm": 5.638436505250866e-06,
      "learning_rate": 1.150666208866922e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15485280,
      "step": 27260
    },
    {
      "epoch": 478.3362831858407,
      "grad_norm": 2.982967259868019e-07,
      "learning_rate": 1.1498398424397106e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15487776,
      "step": 27265
    },
    {
      "epoch": 478.42477876106193,
      "grad_norm": 8.300235094793607e-07,
      "learning_rate": 1.1490136842241628e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15490944,
      "step": 27270
    },
    {
      "epoch": 478.5132743362832,
      "grad_norm": 1.1064087175327586e-06,
      "learning_rate": 1.1481877343476813e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15493584,
      "step": 27275
    },
    {
      "epoch": 478.60176991150445,
      "grad_norm": 5.96902907545882e-07,
      "learning_rate": 1.14736199293764e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15496672,
      "step": 27280
    },
    {
      "epoch": 478.69026548672565,
      "grad_norm": 5.479967057908652e-07,
      "learning_rate": 1.1465364601213771e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15499664,
      "step": 27285
    },
    {
      "epoch": 478.7787610619469,
      "grad_norm": 4.448575509741204e-06,
      "learning_rate": 1.1457111360262012e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15502736,
      "step": 27290
    },
    {
      "epoch": 478.86725663716817,
      "grad_norm": 8.28266934149724e-07,
      "learning_rate": 1.1448860207793869e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15505280,
      "step": 27295
    },
    {
      "epoch": 478.95575221238937,
      "grad_norm": 2.52499972930309e-07,
      "learning_rate": 1.144061114508177e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15508256,
      "step": 27300
    },
    {
      "epoch": 479.0353982300885,
      "grad_norm": 2.052348122560943e-07,
      "learning_rate": 1.1432364173397842e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15510952,
      "step": 27305
    },
    {
      "epoch": 479.12389380530976,
      "grad_norm": 3.075747372349724e-05,
      "learning_rate": 1.1424119294013852e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15513672,
      "step": 27310
    },
    {
      "epoch": 479.21238938053096,
      "grad_norm": 7.091379075063742e-07,
      "learning_rate": 1.1415876508201279e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15516376,
      "step": 27315
    },
    {
      "epoch": 479.3008849557522,
      "grad_norm": 1.3738005009145127e-06,
      "learning_rate": 1.140763581723125e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15519176,
      "step": 27320
    },
    {
      "epoch": 479.3893805309734,
      "grad_norm": 1.120688011724269e-05,
      "learning_rate": 1.1399397222374588e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15522648,
      "step": 27325
    },
    {
      "epoch": 479.4778761061947,
      "grad_norm": 5.758777774644841e-07,
      "learning_rate": 1.1391160724901804e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15525672,
      "step": 27330
    },
    {
      "epoch": 479.56637168141594,
      "grad_norm": 1.2959922059962992e-06,
      "learning_rate": 1.138292632608304e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15528648,
      "step": 27335
    },
    {
      "epoch": 479.65486725663715,
      "grad_norm": 1.063925878952432e-06,
      "learning_rate": 1.1374694027188174e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15531416,
      "step": 27340
    },
    {
      "epoch": 479.7433628318584,
      "grad_norm": 1.1710875696735457e-05,
      "learning_rate": 1.1366463829486711e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15533992,
      "step": 27345
    },
    {
      "epoch": 479.83185840707966,
      "grad_norm": 3.1465421557186346e-07,
      "learning_rate": 1.1358235734247849e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15536600,
      "step": 27350
    },
    {
      "epoch": 479.92035398230087,
      "grad_norm": 2.6456754653736425e-07,
      "learning_rate": 1.1350009742740478e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15539992,
      "step": 27355
    },
    {
      "epoch": 480.0,
      "grad_norm": 4.6318143631651765e-07,
      "learning_rate": 1.134178585623313e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15542312,
      "step": 27360
    },
    {
      "epoch": 480.08849557522126,
      "grad_norm": 7.548263738499372e-07,
      "learning_rate": 1.1333564075994047e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15545032,
      "step": 27365
    },
    {
      "epoch": 480.17699115044246,
      "grad_norm": 1.6540324168090592e-06,
      "learning_rate": 1.1325344403291133e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15548616,
      "step": 27370
    },
    {
      "epoch": 480.2654867256637,
      "grad_norm": 1.0401656709291274e-06,
      "learning_rate": 1.1317126839391951e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15551624,
      "step": 27375
    },
    {
      "epoch": 480.353982300885,
      "grad_norm": 1.6231280142164906e-06,
      "learning_rate": 1.1308911385563766e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15555016,
      "step": 27380
    },
    {
      "epoch": 480.4424778761062,
      "grad_norm": 4.989935405319557e-06,
      "learning_rate": 1.1300698043073494e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15557608,
      "step": 27385
    },
    {
      "epoch": 480.53097345132744,
      "grad_norm": 1.1088410474258126e-06,
      "learning_rate": 1.1292486813187736e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15560360,
      "step": 27390
    },
    {
      "epoch": 480.6194690265487,
      "grad_norm": 2.3985035113582853e-07,
      "learning_rate": 1.1284277697172782e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15562952,
      "step": 27395
    },
    {
      "epoch": 480.7079646017699,
      "grad_norm": 6.886896812829946e-07,
      "learning_rate": 1.127607069629456e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15565784,
      "step": 27400
    },
    {
      "epoch": 480.7079646017699,
      "eval_loss": 0.9219540953636169,
      "eval_runtime": 1.0673,
      "eval_samples_per_second": 23.423,
      "eval_steps_per_second": 12.18,
      "num_input_tokens_seen": 15565784,
      "step": 27400
    },
    {
      "epoch": 480.79646017699116,
      "grad_norm": 2.9859040751034627e-06,
      "learning_rate": 1.1267865811818701e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15568328,
      "step": 27405
    },
    {
      "epoch": 480.88495575221236,
      "grad_norm": 1.3236250424597529e-06,
      "learning_rate": 1.1259663045010513e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15571400,
      "step": 27410
    },
    {
      "epoch": 480.9734513274336,
      "grad_norm": 2.7867956760019297e-06,
      "learning_rate": 1.1251462397134957e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15574088,
      "step": 27415
    },
    {
      "epoch": 481.05309734513276,
      "grad_norm": 1.4088724356042803e-06,
      "learning_rate": 1.1243263869456664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15576984,
      "step": 27420
    },
    {
      "epoch": 481.14159292035396,
      "grad_norm": 5.897219921280339e-07,
      "learning_rate": 1.1235067463239967e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15579496,
      "step": 27425
    },
    {
      "epoch": 481.2300884955752,
      "grad_norm": 6.540265644616738e-07,
      "learning_rate": 1.122687317974884e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15582376,
      "step": 27430
    },
    {
      "epoch": 481.3185840707965,
      "grad_norm": 5.27126871929795e-07,
      "learning_rate": 1.1218681020246963e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15585416,
      "step": 27435
    },
    {
      "epoch": 481.4070796460177,
      "grad_norm": 1.3073571608401835e-05,
      "learning_rate": 1.1210490985997652e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15588184,
      "step": 27440
    },
    {
      "epoch": 481.49557522123894,
      "grad_norm": 2.4517123165423982e-05,
      "learning_rate": 1.1202303078263917e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15590984,
      "step": 27445
    },
    {
      "epoch": 481.5840707964602,
      "grad_norm": 3.5778427331933926e-07,
      "learning_rate": 1.1194117298308451e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15593608,
      "step": 27450
    },
    {
      "epoch": 481.6725663716814,
      "grad_norm": 1.3961450804345077e-06,
      "learning_rate": 1.1185933647393585e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15596728,
      "step": 27455
    },
    {
      "epoch": 481.76106194690266,
      "grad_norm": 1.0065963351735263e-06,
      "learning_rate": 1.1177752126781354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15599736,
      "step": 27460
    },
    {
      "epoch": 481.8495575221239,
      "grad_norm": 1.7010838746500667e-06,
      "learning_rate": 1.1169572737733441e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15602520,
      "step": 27465
    },
    {
      "epoch": 481.9380530973451,
      "grad_norm": 7.164929911596118e-07,
      "learning_rate": 1.1161395481511216e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15605320,
      "step": 27470
    },
    {
      "epoch": 482.01769911504425,
      "grad_norm": 2.355150030552977e-07,
      "learning_rate": 1.1153220359375722e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15607816,
      "step": 27475
    },
    {
      "epoch": 482.1061946902655,
      "grad_norm": 4.3127676008225535e-07,
      "learning_rate": 1.114504737258765e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15610568,
      "step": 27480
    },
    {
      "epoch": 482.1946902654867,
      "grad_norm": 2.5965207441913662e-06,
      "learning_rate": 1.1136876522407393e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15613480,
      "step": 27485
    },
    {
      "epoch": 482.283185840708,
      "grad_norm": 7.73050885527482e-07,
      "learning_rate": 1.1128707810094985e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15616344,
      "step": 27490
    },
    {
      "epoch": 482.37168141592923,
      "grad_norm": 6.931067559889925e-07,
      "learning_rate": 1.1120541236910157e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15618984,
      "step": 27495
    },
    {
      "epoch": 482.46017699115043,
      "grad_norm": 2.8773747544619255e-06,
      "learning_rate": 1.111237680411229e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15621560,
      "step": 27500
    },
    {
      "epoch": 482.5486725663717,
      "grad_norm": 8.619782647656393e-07,
      "learning_rate": 1.1104214512960433e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15624664,
      "step": 27505
    },
    {
      "epoch": 482.6371681415929,
      "grad_norm": 4.3246171799182775e-07,
      "learning_rate": 1.1096054364713327e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15627640,
      "step": 27510
    },
    {
      "epoch": 482.72566371681415,
      "grad_norm": 6.559396751981694e-06,
      "learning_rate": 1.1087896360629371e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15631112,
      "step": 27515
    },
    {
      "epoch": 482.8141592920354,
      "grad_norm": 5.350914875634771e-07,
      "learning_rate": 1.107974050196662e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15633832,
      "step": 27520
    },
    {
      "epoch": 482.9026548672566,
      "grad_norm": 1.925468950503273e-06,
      "learning_rate": 1.1071586789982816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15636792,
      "step": 27525
    },
    {
      "epoch": 482.9911504424779,
      "grad_norm": 7.95619257587532e-07,
      "learning_rate": 1.1063435225935373e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15639560,
      "step": 27530
    },
    {
      "epoch": 483.070796460177,
      "grad_norm": 2.6332705260756484e-07,
      "learning_rate": 1.1055285811081348e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15642000,
      "step": 27535
    },
    {
      "epoch": 483.1592920353982,
      "grad_norm": 1.6412390095865703e-06,
      "learning_rate": 1.1047138546677499e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15645056,
      "step": 27540
    },
    {
      "epoch": 483.24778761061947,
      "grad_norm": 2.4338757498298946e-07,
      "learning_rate": 1.1038993433980219e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15648016,
      "step": 27545
    },
    {
      "epoch": 483.3362831858407,
      "grad_norm": 1.0427564802739653e-06,
      "learning_rate": 1.1030850474245597e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15650528,
      "step": 27550
    },
    {
      "epoch": 483.42477876106193,
      "grad_norm": 2.978030693157052e-07,
      "learning_rate": 1.102270966872939e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15653824,
      "step": 27555
    },
    {
      "epoch": 483.5132743362832,
      "grad_norm": 1.0257108442601748e-06,
      "learning_rate": 1.1014571018687e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15656496,
      "step": 27560
    },
    {
      "epoch": 483.60176991150445,
      "grad_norm": 2.5899130378093105e-06,
      "learning_rate": 1.1006434525373502e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15659120,
      "step": 27565
    },
    {
      "epoch": 483.69026548672565,
      "grad_norm": 2.723955049077631e-06,
      "learning_rate": 1.0998300190043664e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15662032,
      "step": 27570
    },
    {
      "epoch": 483.7787610619469,
      "grad_norm": 1.5784685274411459e-06,
      "learning_rate": 1.0990168013951882e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15664544,
      "step": 27575
    },
    {
      "epoch": 483.86725663716817,
      "grad_norm": 1.200024712488812e-06,
      "learning_rate": 1.0982037998352263e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15667200,
      "step": 27580
    },
    {
      "epoch": 483.95575221238937,
      "grad_norm": 1.0292114893672988e-06,
      "learning_rate": 1.0973910144498534e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15670000,
      "step": 27585
    },
    {
      "epoch": 484.0353982300885,
      "grad_norm": 4.7038989237080386e-07,
      "learning_rate": 1.0965784453644123e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15672776,
      "step": 27590
    },
    {
      "epoch": 484.12389380530976,
      "grad_norm": 6.23340326910693e-07,
      "learning_rate": 1.0957660927042127e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15676328,
      "step": 27595
    },
    {
      "epoch": 484.21238938053096,
      "grad_norm": 8.839696761242521e-07,
      "learning_rate": 1.094953956594527e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15679720,
      "step": 27600
    },
    {
      "epoch": 484.21238938053096,
      "eval_loss": 0.9056763052940369,
      "eval_runtime": 1.0635,
      "eval_samples_per_second": 23.507,
      "eval_steps_per_second": 12.224,
      "num_input_tokens_seen": 15679720,
      "step": 27600
    },
    {
      "epoch": 484.3008849557522,
      "grad_norm": 3.6815160342484887e-07,
      "learning_rate": 1.0941420371605981e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15682680,
      "step": 27605
    },
    {
      "epoch": 484.3893805309734,
      "grad_norm": 1.5952646208461374e-06,
      "learning_rate": 1.0933303345276354e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15685336,
      "step": 27610
    },
    {
      "epoch": 484.4778761061947,
      "grad_norm": 6.599125299544539e-07,
      "learning_rate": 1.0925188488208112e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15687944,
      "step": 27615
    },
    {
      "epoch": 484.56637168141594,
      "grad_norm": 9.505480988991621e-07,
      "learning_rate": 1.0917075801652694e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15690776,
      "step": 27620
    },
    {
      "epoch": 484.65486725663715,
      "grad_norm": 1.5254419167831657e-06,
      "learning_rate": 1.0908965286861151e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15693704,
      "step": 27625
    },
    {
      "epoch": 484.7433628318584,
      "grad_norm": 1.0508344530535396e-05,
      "learning_rate": 1.090085694508425e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15696456,
      "step": 27630
    },
    {
      "epoch": 484.83185840707966,
      "grad_norm": 9.775766329767066e-07,
      "learning_rate": 1.089275077757238e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15699304,
      "step": 27635
    },
    {
      "epoch": 484.92035398230087,
      "grad_norm": 5.233441697782837e-07,
      "learning_rate": 1.0884646785575633e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15701880,
      "step": 27640
    },
    {
      "epoch": 485.0,
      "grad_norm": 2.0291638236358267e-07,
      "learning_rate": 1.0876544970343728e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15704328,
      "step": 27645
    },
    {
      "epoch": 485.08849557522126,
      "grad_norm": 2.2257427190197632e-05,
      "learning_rate": 1.0868445333126082e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15707336,
      "step": 27650
    },
    {
      "epoch": 485.17699115044246,
      "grad_norm": 7.122674787751748e-07,
      "learning_rate": 1.0860347875171745e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15709944,
      "step": 27655
    },
    {
      "epoch": 485.2654867256637,
      "grad_norm": 8.888649176697072e-07,
      "learning_rate": 1.0852252597729465e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15712472,
      "step": 27660
    },
    {
      "epoch": 485.353982300885,
      "grad_norm": 2.2054280179872876e-06,
      "learning_rate": 1.0844159502047615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15715112,
      "step": 27665
    },
    {
      "epoch": 485.4424778761062,
      "grad_norm": 4.546822310658172e-06,
      "learning_rate": 1.0836068589374265e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15717704,
      "step": 27670
    },
    {
      "epoch": 485.53097345132744,
      "grad_norm": 3.488051731892483e-07,
      "learning_rate": 1.0827979860957144e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15720360,
      "step": 27675
    },
    {
      "epoch": 485.6194690265487,
      "grad_norm": 1.4903633882568101e-06,
      "learning_rate": 1.0819893318043615e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15723352,
      "step": 27680
    },
    {
      "epoch": 485.7079646017699,
      "grad_norm": 5.211467851040652e-06,
      "learning_rate": 1.0811808961880734e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15726600,
      "step": 27685
    },
    {
      "epoch": 485.79646017699116,
      "grad_norm": 2.248723149023135e-06,
      "learning_rate": 1.080372679371522e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15729752,
      "step": 27690
    },
    {
      "epoch": 485.88495575221236,
      "grad_norm": 1.7010738702083472e-06,
      "learning_rate": 1.0795646814793428e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15732568,
      "step": 27695
    },
    {
      "epoch": 485.9734513274336,
      "grad_norm": 1.543937599990386e-07,
      "learning_rate": 1.078756902636141e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15735464,
      "step": 27700
    },
    {
      "epoch": 486.05309734513276,
      "grad_norm": 2.2563087895832723e-06,
      "learning_rate": 1.077949342966485e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15737976,
      "step": 27705
    },
    {
      "epoch": 486.14159292035396,
      "grad_norm": 2.578488249582733e-07,
      "learning_rate": 1.0771420025949103e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15740776,
      "step": 27710
    },
    {
      "epoch": 486.2300884955752,
      "grad_norm": 1.504801730334293e-06,
      "learning_rate": 1.0763348816459204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15744008,
      "step": 27715
    },
    {
      "epoch": 486.3185840707965,
      "grad_norm": 3.158402535063942e-07,
      "learning_rate": 1.0755279802439816e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15746824,
      "step": 27720
    },
    {
      "epoch": 486.4070796460177,
      "grad_norm": 2.2816941509518074e-06,
      "learning_rate": 1.0747212985135293e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15749496,
      "step": 27725
    },
    {
      "epoch": 486.49557522123894,
      "grad_norm": 1.852932598467305e-07,
      "learning_rate": 1.073914836578965e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15751848,
      "step": 27730
    },
    {
      "epoch": 486.5840707964602,
      "grad_norm": 1.4240189329939312e-06,
      "learning_rate": 1.0731085945646529e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15755016,
      "step": 27735
    },
    {
      "epoch": 486.6725663716814,
      "grad_norm": 1.424331458110828e-06,
      "learning_rate": 1.0723025725949285e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15758056,
      "step": 27740
    },
    {
      "epoch": 486.76106194690266,
      "grad_norm": 2.6872266971622594e-06,
      "learning_rate": 1.0714967707940875e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15760712,
      "step": 27745
    },
    {
      "epoch": 486.8495575221239,
      "grad_norm": 1.363938395115838e-06,
      "learning_rate": 1.0706911892863963e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15763544,
      "step": 27750
    },
    {
      "epoch": 486.9380530973451,
      "grad_norm": 7.671509365536622e-07,
      "learning_rate": 1.0698858281960866e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15766648,
      "step": 27755
    },
    {
      "epoch": 487.01769911504425,
      "grad_norm": 2.974332460325968e-07,
      "learning_rate": 1.069080687647353e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15769352,
      "step": 27760
    },
    {
      "epoch": 487.1061946902655,
      "grad_norm": 4.482244548853487e-06,
      "learning_rate": 1.0682757677643596e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15771960,
      "step": 27765
    },
    {
      "epoch": 487.1946902654867,
      "grad_norm": 1.4149067055768683e-06,
      "learning_rate": 1.0674710686712359e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15774872,
      "step": 27770
    },
    {
      "epoch": 487.283185840708,
      "grad_norm": 8.478380095766624e-07,
      "learning_rate": 1.0666665904920756e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15777672,
      "step": 27775
    },
    {
      "epoch": 487.37168141592923,
      "grad_norm": 4.838669838136411e-07,
      "learning_rate": 1.0658623333509385e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15780872,
      "step": 27780
    },
    {
      "epoch": 487.46017699115043,
      "grad_norm": 1.2348601785561186e-06,
      "learning_rate": 1.0650582973718532e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15784040,
      "step": 27785
    },
    {
      "epoch": 487.5486725663717,
      "grad_norm": 2.282265995745547e-06,
      "learning_rate": 1.0642544826788098e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15786696,
      "step": 27790
    },
    {
      "epoch": 487.6371681415929,
      "grad_norm": 8.686673027114011e-06,
      "learning_rate": 1.063450889395769e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15789272,
      "step": 27795
    },
    {
      "epoch": 487.72566371681415,
      "grad_norm": 4.1747074419618e-07,
      "learning_rate": 1.062647517646653e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15792680,
      "step": 27800
    },
    {
      "epoch": 487.72566371681415,
      "eval_loss": 0.9154664278030396,
      "eval_runtime": 1.0614,
      "eval_samples_per_second": 23.555,
      "eval_steps_per_second": 12.248,
      "num_input_tokens_seen": 15792680,
      "step": 27800
    },
    {
      "epoch": 487.8141592920354,
      "grad_norm": 1.357956762149115e-06,
      "learning_rate": 1.0618443675553527e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15795880,
      "step": 27805
    },
    {
      "epoch": 487.9026548672566,
      "grad_norm": 1.574624775457778e-06,
      "learning_rate": 1.0610414392457247e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15798424,
      "step": 27810
    },
    {
      "epoch": 487.9911504424779,
      "grad_norm": 9.663150422056788e-07,
      "learning_rate": 1.0602387328415888e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15801624,
      "step": 27815
    },
    {
      "epoch": 488.070796460177,
      "grad_norm": 3.879972894083039e-07,
      "learning_rate": 1.0594362484667347e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15803736,
      "step": 27820
    },
    {
      "epoch": 488.1592920353982,
      "grad_norm": 3.0803653316979762e-06,
      "learning_rate": 1.0586339862449132e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15806408,
      "step": 27825
    },
    {
      "epoch": 488.24778761061947,
      "grad_norm": 5.829708698001923e-07,
      "learning_rate": 1.0578319462998445e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15809416,
      "step": 27830
    },
    {
      "epoch": 488.3362831858407,
      "grad_norm": 1.9258864085713867e-06,
      "learning_rate": 1.057030128755214e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15812392,
      "step": 27835
    },
    {
      "epoch": 488.42477876106193,
      "grad_norm": 8.103942263915087e-07,
      "learning_rate": 1.0562285337346703e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15815896,
      "step": 27840
    },
    {
      "epoch": 488.5132743362832,
      "grad_norm": 1.4848974387859926e-06,
      "learning_rate": 1.0554271613618308e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15819336,
      "step": 27845
    },
    {
      "epoch": 488.60176991150445,
      "grad_norm": 4.142603131640499e-07,
      "learning_rate": 1.054626011760276e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15822296,
      "step": 27850
    },
    {
      "epoch": 488.69026548672565,
      "grad_norm": 4.462406479888159e-07,
      "learning_rate": 1.0538250850535549e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15825352,
      "step": 27855
    },
    {
      "epoch": 488.7787610619469,
      "grad_norm": 2.7367893835616997e-06,
      "learning_rate": 1.0530243813651794e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15827672,
      "step": 27860
    },
    {
      "epoch": 488.86725663716817,
      "grad_norm": 1.6179693602680345e-06,
      "learning_rate": 1.0522239008186271e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15830072,
      "step": 27865
    },
    {
      "epoch": 488.95575221238937,
      "grad_norm": 2.0618267626559827e-06,
      "learning_rate": 1.0514236435373434e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15832920,
      "step": 27870
    },
    {
      "epoch": 489.0353982300885,
      "grad_norm": 8.03117586656299e-07,
      "learning_rate": 1.0506236096447386e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15835384,
      "step": 27875
    },
    {
      "epoch": 489.12389380530976,
      "grad_norm": 0.00012537634756881744,
      "learning_rate": 1.049823799264186e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15838136,
      "step": 27880
    },
    {
      "epoch": 489.21238938053096,
      "grad_norm": 3.7441353128997434e-07,
      "learning_rate": 1.049024212519028e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15841240,
      "step": 27885
    },
    {
      "epoch": 489.3008849557522,
      "grad_norm": 5.945958037045784e-07,
      "learning_rate": 1.0482248495325713e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15843800,
      "step": 27890
    },
    {
      "epoch": 489.3893805309734,
      "grad_norm": 3.373728247879626e-07,
      "learning_rate": 1.047425710428086e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15846568,
      "step": 27895
    },
    {
      "epoch": 489.4778761061947,
      "grad_norm": 1.4252582332119346e-06,
      "learning_rate": 1.0466267953288114e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15849160,
      "step": 27900
    },
    {
      "epoch": 489.56637168141594,
      "grad_norm": 6.025668426445918e-07,
      "learning_rate": 1.0458281043579482e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15852296,
      "step": 27905
    },
    {
      "epoch": 489.65486725663715,
      "grad_norm": 5.150295237399405e-06,
      "learning_rate": 1.0450296376386657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15855480,
      "step": 27910
    },
    {
      "epoch": 489.7433628318584,
      "grad_norm": 8.337385679624276e-07,
      "learning_rate": 1.044231395294098e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15857880,
      "step": 27915
    },
    {
      "epoch": 489.83185840707966,
      "grad_norm": 2.976083521843975e-07,
      "learning_rate": 1.0434333774473435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15861416,
      "step": 27920
    },
    {
      "epoch": 489.92035398230087,
      "grad_norm": 2.361084398216917e-06,
      "learning_rate": 1.0426355842214657e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15863928,
      "step": 27925
    },
    {
      "epoch": 490.0,
      "grad_norm": 7.900802643234783e-07,
      "learning_rate": 1.0418380157394963e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15866288,
      "step": 27930
    },
    {
      "epoch": 490.08849557522126,
      "grad_norm": 1.2484911167121027e-06,
      "learning_rate": 1.0410406721244281e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15868624,
      "step": 27935
    },
    {
      "epoch": 490.17699115044246,
      "grad_norm": 6.43493535790185e-07,
      "learning_rate": 1.0402435534992238e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15871104,
      "step": 27940
    },
    {
      "epoch": 490.2654867256637,
      "grad_norm": 2.730721462285146e-07,
      "learning_rate": 1.0394466599868071e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15874144,
      "step": 27945
    },
    {
      "epoch": 490.353982300885,
      "grad_norm": 8.681357712703175e-07,
      "learning_rate": 1.0386499917100697e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15877168,
      "step": 27950
    },
    {
      "epoch": 490.4424778761062,
      "grad_norm": 2.3981428967090324e-06,
      "learning_rate": 1.0378535487918692e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15879920,
      "step": 27955
    },
    {
      "epoch": 490.53097345132744,
      "grad_norm": 4.97062819704297e-06,
      "learning_rate": 1.037057331355025e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15882768,
      "step": 27960
    },
    {
      "epoch": 490.6194690265487,
      "grad_norm": 4.1802547912084265e-07,
      "learning_rate": 1.0362613395223247e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15885952,
      "step": 27965
    },
    {
      "epoch": 490.7079646017699,
      "grad_norm": 2.3114405394153437e-06,
      "learning_rate": 1.0354655734165212e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15889264,
      "step": 27970
    },
    {
      "epoch": 490.79646017699116,
      "grad_norm": 1.4086881492403336e-06,
      "learning_rate": 1.03467003316033e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15891904,
      "step": 27975
    },
    {
      "epoch": 490.88495575221236,
      "grad_norm": 4.4752920302926213e-07,
      "learning_rate": 1.033874718876435e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15895296,
      "step": 27980
    },
    {
      "epoch": 490.9734513274336,
      "grad_norm": 3.408739246424375e-07,
      "learning_rate": 1.0330796306874818e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15897952,
      "step": 27985
    },
    {
      "epoch": 491.05309734513276,
      "grad_norm": 2.773382766463328e-05,
      "learning_rate": 1.032284768716085e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15900272,
      "step": 27990
    },
    {
      "epoch": 491.14159292035396,
      "grad_norm": 1.1576458973650006e-06,
      "learning_rate": 1.0314901330848206e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15903680,
      "step": 27995
    },
    {
      "epoch": 491.2300884955752,
      "grad_norm": 7.523898375438876e-07,
      "learning_rate": 1.030695723916233e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15906624,
      "step": 28000
    },
    {
      "epoch": 491.2300884955752,
      "eval_loss": 0.925339937210083,
      "eval_runtime": 1.06,
      "eval_samples_per_second": 23.585,
      "eval_steps_per_second": 12.264,
      "num_input_tokens_seen": 15906624,
      "step": 28000
    },
    {
      "epoch": 491.3185840707965,
      "grad_norm": 5.027106908528367e-07,
      "learning_rate": 1.0299015413328289e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15909952,
      "step": 28005
    },
    {
      "epoch": 491.4070796460177,
      "grad_norm": 8.132918196679384e-07,
      "learning_rate": 1.0291075854570809e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15912832,
      "step": 28010
    },
    {
      "epoch": 491.49557522123894,
      "grad_norm": 7.29210739791597e-07,
      "learning_rate": 1.0283138564114275e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15915616,
      "step": 28015
    },
    {
      "epoch": 491.5840707964602,
      "grad_norm": 4.3614090827759355e-06,
      "learning_rate": 1.027520354318273e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15918304,
      "step": 28020
    },
    {
      "epoch": 491.6725663716814,
      "grad_norm": 2.0481461433519144e-06,
      "learning_rate": 1.0267270792999828e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15921024,
      "step": 28025
    },
    {
      "epoch": 491.76106194690266,
      "grad_norm": 1.5587792177029769e-06,
      "learning_rate": 1.0259340314788919e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15924064,
      "step": 28030
    },
    {
      "epoch": 491.8495575221239,
      "grad_norm": 2.548856912198971e-07,
      "learning_rate": 1.0251412109772979e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15926880,
      "step": 28035
    },
    {
      "epoch": 491.9380530973451,
      "grad_norm": 1.93035111806239e-06,
      "learning_rate": 1.0243486179174627e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15929184,
      "step": 28040
    },
    {
      "epoch": 492.01769911504425,
      "grad_norm": 1.8156943326630426e-07,
      "learning_rate": 1.0235562524216158e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15931552,
      "step": 28045
    },
    {
      "epoch": 492.1061946902655,
      "grad_norm": 2.631910831496498e-07,
      "learning_rate": 1.022764114611948e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15934192,
      "step": 28050
    },
    {
      "epoch": 492.1946902654867,
      "grad_norm": 7.367377747868886e-07,
      "learning_rate": 1.0219722046106178e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15937104,
      "step": 28055
    },
    {
      "epoch": 492.283185840708,
      "grad_norm": 1.708687818791077e-06,
      "learning_rate": 1.0211805225397486e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15939728,
      "step": 28060
    },
    {
      "epoch": 492.37168141592923,
      "grad_norm": 6.432373993447982e-07,
      "learning_rate": 1.020389068521426e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15943264,
      "step": 28065
    },
    {
      "epoch": 492.46017699115043,
      "grad_norm": 0.00011726022057700902,
      "learning_rate": 1.0195978426777039e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15945904,
      "step": 28070
    },
    {
      "epoch": 492.5486725663717,
      "grad_norm": 5.36229720182746e-07,
      "learning_rate": 1.0188068451305982e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15949104,
      "step": 28075
    },
    {
      "epoch": 492.6371681415929,
      "grad_norm": 1.4840842368357698e-06,
      "learning_rate": 1.0180160760020902e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15952032,
      "step": 28080
    },
    {
      "epoch": 492.72566371681415,
      "grad_norm": 2.3540800953014696e-07,
      "learning_rate": 1.0172255354141278e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15954944,
      "step": 28085
    },
    {
      "epoch": 492.8141592920354,
      "grad_norm": 1.7856735894383746e-06,
      "learning_rate": 1.0164352234886205e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15957568,
      "step": 28090
    },
    {
      "epoch": 492.9026548672566,
      "grad_norm": 2.0211626861055265e-07,
      "learning_rate": 1.0156451403474454e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15960352,
      "step": 28095
    },
    {
      "epoch": 492.9911504424779,
      "grad_norm": 1.3597646102425642e-06,
      "learning_rate": 1.0148552861124443e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15963360,
      "step": 28100
    },
    {
      "epoch": 493.070796460177,
      "grad_norm": 2.1469749356128887e-07,
      "learning_rate": 1.0140656609054205e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15966096,
      "step": 28105
    },
    {
      "epoch": 493.1592920353982,
      "grad_norm": 8.973077001428464e-07,
      "learning_rate": 1.0132762648481455e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15968816,
      "step": 28110
    },
    {
      "epoch": 493.24778761061947,
      "grad_norm": 3.697712713801593e-07,
      "learning_rate": 1.0124870980623543e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15972160,
      "step": 28115
    },
    {
      "epoch": 493.3362831858407,
      "grad_norm": 8.301561820189818e-07,
      "learning_rate": 1.0116981606697453e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15974976,
      "step": 28120
    },
    {
      "epoch": 493.42477876106193,
      "grad_norm": 7.261751875375921e-07,
      "learning_rate": 1.0109094527919838e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15977984,
      "step": 28125
    },
    {
      "epoch": 493.5132743362832,
      "grad_norm": 4.028789248877729e-07,
      "learning_rate": 1.010120974550697e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15981376,
      "step": 28130
    },
    {
      "epoch": 493.60176991150445,
      "grad_norm": 1.878303692137706e-06,
      "learning_rate": 1.0093327260674795e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15983952,
      "step": 28135
    },
    {
      "epoch": 493.69026548672565,
      "grad_norm": 1.3784276120532013e-07,
      "learning_rate": 1.0085447074638878e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15987056,
      "step": 28140
    },
    {
      "epoch": 493.7787610619469,
      "grad_norm": 1.5837840692256577e-05,
      "learning_rate": 1.0077569188614461e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15989728,
      "step": 28145
    },
    {
      "epoch": 493.86725663716817,
      "grad_norm": 2.2443668967753183e-06,
      "learning_rate": 1.0069693603816393e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15992624,
      "step": 28150
    },
    {
      "epoch": 493.95575221238937,
      "grad_norm": 5.08786058617261e-07,
      "learning_rate": 1.0061820321459204e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15995184,
      "step": 28155
    },
    {
      "epoch": 494.0353982300885,
      "grad_norm": 5.110873644298408e-07,
      "learning_rate": 1.0053949342757038e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 15997584,
      "step": 28160
    },
    {
      "epoch": 494.12389380530976,
      "grad_norm": 9.139416192738281e-07,
      "learning_rate": 1.0046080668923717e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 16000352,
      "step": 28165
    },
    {
      "epoch": 494.21238938053096,
      "grad_norm": 1.296965820074547e-06,
      "learning_rate": 1.003821430117267e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 16003136,
      "step": 28170
    },
    {
      "epoch": 494.3008849557522,
      "grad_norm": 3.978399831794377e-07,
      "learning_rate": 1.0030350240716999e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 16005744,
      "step": 28175
    },
    {
      "epoch": 494.3893805309734,
      "grad_norm": 1.0867880462228641e-07,
      "learning_rate": 1.0022488488769449e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 16008176,
      "step": 28180
    },
    {
      "epoch": 494.4778761061947,
      "grad_norm": 5.528005431187921e-07,
      "learning_rate": 1.0014629046542387e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 16010960,
      "step": 28185
    },
    {
      "epoch": 494.56637168141594,
      "grad_norm": 3.1105846574064344e-06,
      "learning_rate": 1.0006771915247842e-05,
      "loss": 0.0,
      "num_input_tokens_seen": 16013568,
      "step": 28190
    },
    {
      "epoch": 494.65486725663715,
      "grad_norm": 9.642723171054968e-07,
      "learning_rate": 9.998917096097495e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16016880,
      "step": 28195
    },
    {
      "epoch": 494.7433628318584,
      "grad_norm": 0.0002820623049046844,
      "learning_rate": 9.991064590302638e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16019936,
      "step": 28200
    },
    {
      "epoch": 494.7433628318584,
      "eval_loss": 0.9103339910507202,
      "eval_runtime": 1.0652,
      "eval_samples_per_second": 23.469,
      "eval_steps_per_second": 12.204,
      "num_input_tokens_seen": 16019936,
      "step": 28200
    },
    {
      "epoch": 494.83185840707966,
      "grad_norm": 7.486558502023399e-07,
      "learning_rate": 9.983214399074241e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16022880,
      "step": 28205
    },
    {
      "epoch": 494.92035398230087,
      "grad_norm": 5.213549343352497e-07,
      "learning_rate": 9.975366523622893e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16025536,
      "step": 28210
    },
    {
      "epoch": 495.0,
      "grad_norm": 3.755674811145582e-07,
      "learning_rate": 9.967520965158841e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16028400,
      "step": 28215
    },
    {
      "epoch": 495.08849557522126,
      "grad_norm": 6.959613756407634e-07,
      "learning_rate": 9.95967772489197e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16031440,
      "step": 28220
    },
    {
      "epoch": 495.17699115044246,
      "grad_norm": 2.6357284355071897e-07,
      "learning_rate": 9.951836804031794e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16034208,
      "step": 28225
    },
    {
      "epoch": 495.2654867256637,
      "grad_norm": 8.2455981100793e-06,
      "learning_rate": 9.943998203787489e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16037008,
      "step": 28230
    },
    {
      "epoch": 495.353982300885,
      "grad_norm": 1.2096364798708237e-07,
      "learning_rate": 9.936161925367874e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16039616,
      "step": 28235
    },
    {
      "epoch": 495.4424778761062,
      "grad_norm": 2.3293428057513665e-06,
      "learning_rate": 9.928327969981386e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16042784,
      "step": 28240
    },
    {
      "epoch": 495.53097345132744,
      "grad_norm": 9.9839292033721e-07,
      "learning_rate": 9.920496338836135e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16045504,
      "step": 28245
    },
    {
      "epoch": 495.6194690265487,
      "grad_norm": 9.418839681529789e-07,
      "learning_rate": 9.912667033139844e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16048560,
      "step": 28250
    },
    {
      "epoch": 495.7079646017699,
      "grad_norm": 9.239109886038932e-07,
      "learning_rate": 9.904840054099893e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16051424,
      "step": 28255
    },
    {
      "epoch": 495.79646017699116,
      "grad_norm": 7.777087489557744e-07,
      "learning_rate": 9.897015402923312e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16054320,
      "step": 28260
    },
    {
      "epoch": 495.88495575221236,
      "grad_norm": 1.9917939653169014e-07,
      "learning_rate": 9.889193080816744e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16057616,
      "step": 28265
    },
    {
      "epoch": 495.9734513274336,
      "grad_norm": 1.507151978330512e-06,
      "learning_rate": 9.881373088986498e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16060192,
      "step": 28270
    },
    {
      "epoch": 496.05309734513276,
      "grad_norm": 1.4252347000365262e-06,
      "learning_rate": 9.873555428638523e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16062768,
      "step": 28275
    },
    {
      "epoch": 496.14159292035396,
      "grad_norm": 4.367442670627497e-06,
      "learning_rate": 9.865740100978383e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16065344,
      "step": 28280
    },
    {
      "epoch": 496.2300884955752,
      "grad_norm": 6.8976728471170645e-06,
      "learning_rate": 9.857927107211315e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16068416,
      "step": 28285
    },
    {
      "epoch": 496.3185840707965,
      "grad_norm": 3.3244401720367023e-07,
      "learning_rate": 9.850116448542177e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16071088,
      "step": 28290
    },
    {
      "epoch": 496.4070796460177,
      "grad_norm": 4.005789548955363e-07,
      "learning_rate": 9.842308126175457e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16073616,
      "step": 28295
    },
    {
      "epoch": 496.49557522123894,
      "grad_norm": 4.046918888889195e-07,
      "learning_rate": 9.834502141315315e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16076064,
      "step": 28300
    },
    {
      "epoch": 496.5840707964602,
      "grad_norm": 3.4350775877101114e-06,
      "learning_rate": 9.82669849516552e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16079136,
      "step": 28305
    },
    {
      "epoch": 496.6725663716814,
      "grad_norm": 7.523482281612814e-07,
      "learning_rate": 9.818897188929493e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16081920,
      "step": 28310
    },
    {
      "epoch": 496.76106194690266,
      "grad_norm": 1.4524061953125056e-06,
      "learning_rate": 9.811098223810309e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16085072,
      "step": 28315
    },
    {
      "epoch": 496.8495575221239,
      "grad_norm": 5.085695420348202e-07,
      "learning_rate": 9.803301601010641e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16088336,
      "step": 28320
    },
    {
      "epoch": 496.9380530973451,
      "grad_norm": 8.73264411893615e-08,
      "learning_rate": 9.795507321732853e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16091408,
      "step": 28325
    },
    {
      "epoch": 497.01769911504425,
      "grad_norm": 4.830088755625184e-07,
      "learning_rate": 9.787715387178898e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16094072,
      "step": 28330
    },
    {
      "epoch": 497.1061946902655,
      "grad_norm": 1.552429239382036e-06,
      "learning_rate": 9.779925798550399e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16096984,
      "step": 28335
    },
    {
      "epoch": 497.1946902654867,
      "grad_norm": 3.5708137602341594e-07,
      "learning_rate": 9.772138557048619e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16099432,
      "step": 28340
    },
    {
      "epoch": 497.283185840708,
      "grad_norm": 2.7964497917309927e-07,
      "learning_rate": 9.764353663874426e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16102472,
      "step": 28345
    },
    {
      "epoch": 497.37168141592923,
      "grad_norm": 7.697752153035253e-06,
      "learning_rate": 9.756571120228375e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16105208,
      "step": 28350
    },
    {
      "epoch": 497.46017699115043,
      "grad_norm": 7.252769864862785e-05,
      "learning_rate": 9.748790927310605e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16108248,
      "step": 28355
    },
    {
      "epoch": 497.5486725663717,
      "grad_norm": 1.8306069193840813e-07,
      "learning_rate": 9.741013086320946e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16111864,
      "step": 28360
    },
    {
      "epoch": 497.6371681415929,
      "grad_norm": 7.129259529392584e-07,
      "learning_rate": 9.733237598458821e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16114456,
      "step": 28365
    },
    {
      "epoch": 497.72566371681415,
      "grad_norm": 8.812317560114025e-07,
      "learning_rate": 9.725464464923308e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16117240,
      "step": 28370
    },
    {
      "epoch": 497.8141592920354,
      "grad_norm": 2.393099066466675e-06,
      "learning_rate": 9.717693686913123e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16119992,
      "step": 28375
    },
    {
      "epoch": 497.9026548672566,
      "grad_norm": 1.1010693015123252e-06,
      "learning_rate": 9.709925265626632e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16122744,
      "step": 28380
    },
    {
      "epoch": 497.9911504424779,
      "grad_norm": 1.6764735164542799e-06,
      "learning_rate": 9.702159202261801e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16125528,
      "step": 28385
    },
    {
      "epoch": 498.070796460177,
      "grad_norm": 3.055911520277732e-06,
      "learning_rate": 9.694395498016268e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16127848,
      "step": 28390
    },
    {
      "epoch": 498.1592920353982,
      "grad_norm": 1.567874278407544e-05,
      "learning_rate": 9.686634154087298e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16130488,
      "step": 28395
    },
    {
      "epoch": 498.24778761061947,
      "grad_norm": 1.791764475456148e-06,
      "learning_rate": 9.678875171671776e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16133784,
      "step": 28400
    },
    {
      "epoch": 498.24778761061947,
      "eval_loss": 0.9244890809059143,
      "eval_runtime": 1.0777,
      "eval_samples_per_second": 23.197,
      "eval_steps_per_second": 12.062,
      "num_input_tokens_seen": 16133784,
      "step": 28400
    },
    {
      "epoch": 498.3362831858407,
      "grad_norm": 3.195310398496076e-07,
      "learning_rate": 9.671118551966246e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16137032,
      "step": 28405
    },
    {
      "epoch": 498.42477876106193,
      "grad_norm": 8.960253694567655e-07,
      "learning_rate": 9.66336429616686e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16139944,
      "step": 28410
    },
    {
      "epoch": 498.5132743362832,
      "grad_norm": 3.8999547768980847e-07,
      "learning_rate": 9.655612405469436e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16142856,
      "step": 28415
    },
    {
      "epoch": 498.60176991150445,
      "grad_norm": 2.8561671570059843e-06,
      "learning_rate": 9.647862881069413e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16145848,
      "step": 28420
    },
    {
      "epoch": 498.69026548672565,
      "grad_norm": 7.822560291970149e-05,
      "learning_rate": 9.640115724161855e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16148488,
      "step": 28425
    },
    {
      "epoch": 498.7787610619469,
      "grad_norm": 4.1308186382593703e-07,
      "learning_rate": 9.632370935941483e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16151384,
      "step": 28430
    },
    {
      "epoch": 498.86725663716817,
      "grad_norm": 5.553350774789578e-07,
      "learning_rate": 9.624628517602634e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16154248,
      "step": 28435
    },
    {
      "epoch": 498.95575221238937,
      "grad_norm": 1.360376018055831e-06,
      "learning_rate": 9.61688847033928e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16157208,
      "step": 28440
    },
    {
      "epoch": 499.0353982300885,
      "grad_norm": 2.1353599777285126e-07,
      "learning_rate": 9.609150795345051e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16159192,
      "step": 28445
    },
    {
      "epoch": 499.12389380530976,
      "grad_norm": 4.661321327148471e-06,
      "learning_rate": 9.601415493813171e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16161672,
      "step": 28450
    },
    {
      "epoch": 499.21238938053096,
      "grad_norm": 3.599516276153736e-07,
      "learning_rate": 9.593682566936533e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16164632,
      "step": 28455
    },
    {
      "epoch": 499.3008849557522,
      "grad_norm": 2.724482328630984e-07,
      "learning_rate": 9.58595201590766e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16167432,
      "step": 28460
    },
    {
      "epoch": 499.3893805309734,
      "grad_norm": 3.4800277148860914e-07,
      "learning_rate": 9.578223841918681e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16170008,
      "step": 28465
    },
    {
      "epoch": 499.4778761061947,
      "grad_norm": 1.941491291290731e-06,
      "learning_rate": 9.570498046161389e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16172872,
      "step": 28470
    },
    {
      "epoch": 499.56637168141594,
      "grad_norm": 3.0260998755693436e-06,
      "learning_rate": 9.562774629827206e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16176408,
      "step": 28475
    },
    {
      "epoch": 499.65486725663715,
      "grad_norm": 4.6989759994175984e-07,
      "learning_rate": 9.555053594107163e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16179272,
      "step": 28480
    },
    {
      "epoch": 499.7433628318584,
      "grad_norm": 1.61751302130142e-06,
      "learning_rate": 9.547334940191957e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16182472,
      "step": 28485
    },
    {
      "epoch": 499.83185840707966,
      "grad_norm": 2.9051414003333775e-06,
      "learning_rate": 9.539618669271886e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16185496,
      "step": 28490
    },
    {
      "epoch": 499.92035398230087,
      "grad_norm": 4.954014229952008e-07,
      "learning_rate": 9.531904782536904e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16188728,
      "step": 28495
    },
    {
      "epoch": 500.0,
      "grad_norm": 1.1520114640006796e-06,
      "learning_rate": 9.524193281176597e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16190856,
      "step": 28500
    },
    {
      "epoch": 500.08849557522126,
      "grad_norm": 2.0175568238300912e-07,
      "learning_rate": 9.516484166380165e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16194104,
      "step": 28505
    },
    {
      "epoch": 500.17699115044246,
      "grad_norm": 1.4144617352940259e-06,
      "learning_rate": 9.508777439336447e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16196776,
      "step": 28510
    },
    {
      "epoch": 500.2654867256637,
      "grad_norm": 2.2639603969309974e-07,
      "learning_rate": 9.50107310123393e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16199848,
      "step": 28515
    },
    {
      "epoch": 500.353982300885,
      "grad_norm": 3.7758570670121117e-07,
      "learning_rate": 9.493371153260702e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16202504,
      "step": 28520
    },
    {
      "epoch": 500.4424778761062,
      "grad_norm": 3.054851731576491e-07,
      "learning_rate": 9.485671596604523e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16205448,
      "step": 28525
    },
    {
      "epoch": 500.53097345132744,
      "grad_norm": 3.7642064398824004e-06,
      "learning_rate": 9.477974432452738e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16208456,
      "step": 28530
    },
    {
      "epoch": 500.6194690265487,
      "grad_norm": 3.080903638874588e-07,
      "learning_rate": 9.470279661992356e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16211032,
      "step": 28535
    },
    {
      "epoch": 500.7079646017699,
      "grad_norm": 6.731983376084827e-06,
      "learning_rate": 9.462587286410021e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16214136,
      "step": 28540
    },
    {
      "epoch": 500.79646017699116,
      "grad_norm": 1.5030180122721504e-07,
      "learning_rate": 9.454897306891972e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16216984,
      "step": 28545
    },
    {
      "epoch": 500.88495575221236,
      "grad_norm": 5.718902116313984e-07,
      "learning_rate": 9.44720972462411e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16220072,
      "step": 28550
    },
    {
      "epoch": 500.9734513274336,
      "grad_norm": 1.325162998000451e-06,
      "learning_rate": 9.439524540791964e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16222648,
      "step": 28555
    },
    {
      "epoch": 501.05309734513276,
      "grad_norm": 8.698632427694974e-07,
      "learning_rate": 9.431841756580673e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16225112,
      "step": 28560
    },
    {
      "epoch": 501.14159292035396,
      "grad_norm": 4.348400750586734e-07,
      "learning_rate": 9.42416137317503e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16228200,
      "step": 28565
    },
    {
      "epoch": 501.2300884955752,
      "grad_norm": 1.174177782559127e-06,
      "learning_rate": 9.416483391759437e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16231048,
      "step": 28570
    },
    {
      "epoch": 501.3185840707965,
      "grad_norm": 2.942682294815313e-05,
      "learning_rate": 9.408807813517945e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16233848,
      "step": 28575
    },
    {
      "epoch": 501.4070796460177,
      "grad_norm": 5.931861437602493e-07,
      "learning_rate": 9.401134639634221e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16236648,
      "step": 28580
    },
    {
      "epoch": 501.49557522123894,
      "grad_norm": 4.349547850779345e-07,
      "learning_rate": 9.393463871291555e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16239512,
      "step": 28585
    },
    {
      "epoch": 501.5840707964602,
      "grad_norm": 1.4775466752325883e-06,
      "learning_rate": 9.385795509672881e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16242424,
      "step": 28590
    },
    {
      "epoch": 501.6725663716814,
      "grad_norm": 6.325751087388198e-07,
      "learning_rate": 9.378129555960771e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16245352,
      "step": 28595
    },
    {
      "epoch": 501.76106194690266,
      "grad_norm": 3.191273208358325e-05,
      "learning_rate": 9.370466011337392e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16248200,
      "step": 28600
    },
    {
      "epoch": 501.76106194690266,
      "eval_loss": 0.8963096141815186,
      "eval_runtime": 1.0658,
      "eval_samples_per_second": 23.456,
      "eval_steps_per_second": 12.197,
      "num_input_tokens_seen": 16248200,
      "step": 28600
    },
    {
      "epoch": 501.8495575221239,
      "grad_norm": 4.1992498722720484e-07,
      "learning_rate": 9.362804876984573e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16251432,
      "step": 28605
    },
    {
      "epoch": 501.9380530973451,
      "grad_norm": 1.569666210343712e-07,
      "learning_rate": 9.355146154083747e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16254008,
      "step": 28610
    },
    {
      "epoch": 502.01769911504425,
      "grad_norm": 2.6836050892597996e-07,
      "learning_rate": 9.347489843815987e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16256376,
      "step": 28615
    },
    {
      "epoch": 502.1061946902655,
      "grad_norm": 3.886034846800612e-06,
      "learning_rate": 9.339835947362002e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16258984,
      "step": 28620
    },
    {
      "epoch": 502.1946902654867,
      "grad_norm": 3.0194490818757913e-07,
      "learning_rate": 9.332184465902105e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16261784,
      "step": 28625
    },
    {
      "epoch": 502.283185840708,
      "grad_norm": 2.6907089250016725e-06,
      "learning_rate": 9.324535400616266e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16264840,
      "step": 28630
    },
    {
      "epoch": 502.37168141592923,
      "grad_norm": 3.799771377543948e-07,
      "learning_rate": 9.31688875268405e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16267768,
      "step": 28635
    },
    {
      "epoch": 502.46017699115043,
      "grad_norm": 9.413260499968601e-07,
      "learning_rate": 9.309244523284674e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16270488,
      "step": 28640
    },
    {
      "epoch": 502.5486725663717,
      "grad_norm": 7.17702164365619e-07,
      "learning_rate": 9.301602713596982e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16273240,
      "step": 28645
    },
    {
      "epoch": 502.6371681415929,
      "grad_norm": 1.7870820556709077e-07,
      "learning_rate": 9.293963324799432e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16276024,
      "step": 28650
    },
    {
      "epoch": 502.72566371681415,
      "grad_norm": 1.3835434629072552e-06,
      "learning_rate": 9.286326358070104e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16278952,
      "step": 28655
    },
    {
      "epoch": 502.8141592920354,
      "grad_norm": 5.470560040521377e-07,
      "learning_rate": 9.278691814586729e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16282248,
      "step": 28660
    },
    {
      "epoch": 502.9026548672566,
      "grad_norm": 1.556599613650178e-06,
      "learning_rate": 9.271059695526635e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16285032,
      "step": 28665
    },
    {
      "epoch": 502.9911504424779,
      "grad_norm": 2.665460442585754e-06,
      "learning_rate": 9.263430002066805e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16287864,
      "step": 28670
    },
    {
      "epoch": 503.070796460177,
      "grad_norm": 9.698583198769484e-07,
      "learning_rate": 9.25580273538382e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16290344,
      "step": 28675
    },
    {
      "epoch": 503.1592920353982,
      "grad_norm": 2.497222055808379e-07,
      "learning_rate": 9.248177896653907e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16292936,
      "step": 28680
    },
    {
      "epoch": 503.24778761061947,
      "grad_norm": 1.7178089137814823e-06,
      "learning_rate": 9.240555487052918e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16296264,
      "step": 28685
    },
    {
      "epoch": 503.3362831858407,
      "grad_norm": 4.6680639798069024e-07,
      "learning_rate": 9.232935507756313e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16298648,
      "step": 28690
    },
    {
      "epoch": 503.42477876106193,
      "grad_norm": 4.826387112188968e-07,
      "learning_rate": 9.225317959939193e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16301512,
      "step": 28695
    },
    {
      "epoch": 503.5132743362832,
      "grad_norm": 1.3481823089023237e-06,
      "learning_rate": 9.217702844776287e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16304440,
      "step": 28700
    },
    {
      "epoch": 503.60176991150445,
      "grad_norm": 1.1609006378421327e-06,
      "learning_rate": 9.210090163441929e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16307096,
      "step": 28705
    },
    {
      "epoch": 503.69026548672565,
      "grad_norm": 4.052407803101232e-06,
      "learning_rate": 9.202479917110105e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16310216,
      "step": 28710
    },
    {
      "epoch": 503.7787610619469,
      "grad_norm": 4.5042204988021695e-07,
      "learning_rate": 9.194872106954392e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16312904,
      "step": 28715
    },
    {
      "epoch": 503.86725663716817,
      "grad_norm": 1.6321110933859018e-06,
      "learning_rate": 9.187266734148029e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16316040,
      "step": 28720
    },
    {
      "epoch": 503.95575221238937,
      "grad_norm": 6.25294831024803e-07,
      "learning_rate": 9.179663799863849e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16319224,
      "step": 28725
    },
    {
      "epoch": 504.0353982300885,
      "grad_norm": 1.4184702195052523e-05,
      "learning_rate": 9.172063305274317e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16321928,
      "step": 28730
    },
    {
      "epoch": 504.12389380530976,
      "grad_norm": 4.793947709913482e-07,
      "learning_rate": 9.164465251551527e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16324424,
      "step": 28735
    },
    {
      "epoch": 504.21238938053096,
      "grad_norm": 8.643937690067105e-07,
      "learning_rate": 9.156869639867205e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16327096,
      "step": 28740
    },
    {
      "epoch": 504.3008849557522,
      "grad_norm": 2.4474388737871777e-07,
      "learning_rate": 9.149276471392677e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16329848,
      "step": 28745
    },
    {
      "epoch": 504.3893805309734,
      "grad_norm": 2.2507427388518408e-07,
      "learning_rate": 9.141685747298914e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16332984,
      "step": 28750
    },
    {
      "epoch": 504.4778761061947,
      "grad_norm": 4.875785748481576e-07,
      "learning_rate": 9.13409746875649e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16335496,
      "step": 28755
    },
    {
      "epoch": 504.56637168141594,
      "grad_norm": 2.4896291961340467e-07,
      "learning_rate": 9.12651163693562e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16338520,
      "step": 28760
    },
    {
      "epoch": 504.65486725663715,
      "grad_norm": 1.899794369819574e-06,
      "learning_rate": 9.11892825300614e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16341640,
      "step": 28765
    },
    {
      "epoch": 504.7433628318584,
      "grad_norm": 9.819417527978658e-07,
      "learning_rate": 9.111347318137491e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16344840,
      "step": 28770
    },
    {
      "epoch": 504.83185840707966,
      "grad_norm": 1.3051204632574809e-06,
      "learning_rate": 9.103768833498755e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16347896,
      "step": 28775
    },
    {
      "epoch": 504.92035398230087,
      "grad_norm": 5.036598622609745e-07,
      "learning_rate": 9.096192800258639e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16350840,
      "step": 28780
    },
    {
      "epoch": 505.0,
      "grad_norm": 4.343002046880429e-08,
      "learning_rate": 9.088619219585443e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16353448,
      "step": 28785
    },
    {
      "epoch": 505.08849557522126,
      "grad_norm": 4.648621541036846e-07,
      "learning_rate": 9.081048092647127e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16356024,
      "step": 28790
    },
    {
      "epoch": 505.17699115044246,
      "grad_norm": 1.9165068465554214e-07,
      "learning_rate": 9.073479420611245e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16359016,
      "step": 28795
    },
    {
      "epoch": 505.2654867256637,
      "grad_norm": 2.971759158754139e-07,
      "learning_rate": 9.065913204644974e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16361560,
      "step": 28800
    },
    {
      "epoch": 505.2654867256637,
      "eval_loss": 0.9023752808570862,
      "eval_runtime": 1.0645,
      "eval_samples_per_second": 23.485,
      "eval_steps_per_second": 12.212,
      "num_input_tokens_seen": 16361560,
      "step": 28800
    },
    {
      "epoch": 505.353982300885,
      "grad_norm": 7.4744202720467e-06,
      "learning_rate": 9.058349445915135e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16364184,
      "step": 28805
    },
    {
      "epoch": 505.4424778761062,
      "grad_norm": 1.2773523394571384e-06,
      "learning_rate": 9.050788145588138e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16367160,
      "step": 28810
    },
    {
      "epoch": 505.53097345132744,
      "grad_norm": 4.7582827278347395e-07,
      "learning_rate": 9.043229304830039e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16370264,
      "step": 28815
    },
    {
      "epoch": 505.6194690265487,
      "grad_norm": 4.958446425007423e-06,
      "learning_rate": 9.035672924806515e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16373224,
      "step": 28820
    },
    {
      "epoch": 505.7079646017699,
      "grad_norm": 8.773142781137722e-07,
      "learning_rate": 9.028119006682839e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16375992,
      "step": 28825
    },
    {
      "epoch": 505.79646017699116,
      "grad_norm": 1.422344666934805e-06,
      "learning_rate": 9.020567551623935e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16379000,
      "step": 28830
    },
    {
      "epoch": 505.88495575221236,
      "grad_norm": 1.5768064258736558e-06,
      "learning_rate": 9.013018560794318e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16381640,
      "step": 28835
    },
    {
      "epoch": 505.9734513274336,
      "grad_norm": 3.770529986013571e-07,
      "learning_rate": 9.005472035358139e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16384680,
      "step": 28840
    },
    {
      "epoch": 506.05309734513276,
      "grad_norm": 3.1400804800796323e-06,
      "learning_rate": 8.997927976479185e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16387568,
      "step": 28845
    },
    {
      "epoch": 506.14159292035396,
      "grad_norm": 3.978259883297142e-06,
      "learning_rate": 8.99038638532082e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16390688,
      "step": 28850
    },
    {
      "epoch": 506.2300884955752,
      "grad_norm": 3.389004632481374e-07,
      "learning_rate": 8.982847263046065e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16393184,
      "step": 28855
    },
    {
      "epoch": 506.3185840707965,
      "grad_norm": 1.2930047432746505e-06,
      "learning_rate": 8.975310610817555e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16396352,
      "step": 28860
    },
    {
      "epoch": 506.4070796460177,
      "grad_norm": 1.4074238379180315e-06,
      "learning_rate": 8.967776429797528e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16399376,
      "step": 28865
    },
    {
      "epoch": 506.49557522123894,
      "grad_norm": 2.6420507310831454e-06,
      "learning_rate": 8.960244721147842e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16402032,
      "step": 28870
    },
    {
      "epoch": 506.5840707964602,
      "grad_norm": 5.091241746413289e-06,
      "learning_rate": 8.952715486029995e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16405216,
      "step": 28875
    },
    {
      "epoch": 506.6725663716814,
      "grad_norm": 1.2489323353293003e-06,
      "learning_rate": 8.945188725605075e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16408240,
      "step": 28880
    },
    {
      "epoch": 506.76106194690266,
      "grad_norm": 9.834492402660544e-07,
      "learning_rate": 8.937664441033817e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16410608,
      "step": 28885
    },
    {
      "epoch": 506.8495575221239,
      "grad_norm": 1.5331143288221938e-07,
      "learning_rate": 8.930142633476549e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16413552,
      "step": 28890
    },
    {
      "epoch": 506.9380530973451,
      "grad_norm": 9.866886330200941e-07,
      "learning_rate": 8.92262330409323e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16416544,
      "step": 28895
    },
    {
      "epoch": 507.01769911504425,
      "grad_norm": 2.5295778982581396e-07,
      "learning_rate": 8.915106454043448e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16419008,
      "step": 28900
    },
    {
      "epoch": 507.1061946902655,
      "grad_norm": 1.3207604752096813e-06,
      "learning_rate": 8.90759208448638e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16422096,
      "step": 28905
    },
    {
      "epoch": 507.1946902654867,
      "grad_norm": 9.827573376242071e-05,
      "learning_rate": 8.900080196580848e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16424464,
      "step": 28910
    },
    {
      "epoch": 507.283185840708,
      "grad_norm": 1.1312048400213826e-06,
      "learning_rate": 8.892570791485267e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16427456,
      "step": 28915
    },
    {
      "epoch": 507.37168141592923,
      "grad_norm": 9.286337672165246e-07,
      "learning_rate": 8.885063870357688e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16429840,
      "step": 28920
    },
    {
      "epoch": 507.46017699115043,
      "grad_norm": 6.972400683480373e-07,
      "learning_rate": 8.87755943435578e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16432976,
      "step": 28925
    },
    {
      "epoch": 507.5486725663717,
      "grad_norm": 1.3220969208305178e-07,
      "learning_rate": 8.87005748463681e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16435584,
      "step": 28930
    },
    {
      "epoch": 507.6371681415929,
      "grad_norm": 8.58770135891973e-07,
      "learning_rate": 8.862558022357681e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16438512,
      "step": 28935
    },
    {
      "epoch": 507.72566371681415,
      "grad_norm": 4.879403604718391e-06,
      "learning_rate": 8.855061048674903e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16441184,
      "step": 28940
    },
    {
      "epoch": 507.8141592920354,
      "grad_norm": 2.7961195883108303e-06,
      "learning_rate": 8.847566564744595e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16443776,
      "step": 28945
    },
    {
      "epoch": 507.9026548672566,
      "grad_norm": 7.978422900123405e-07,
      "learning_rate": 8.840074571722512e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16446544,
      "step": 28950
    },
    {
      "epoch": 507.9911504424779,
      "grad_norm": 1.6984438389044954e-06,
      "learning_rate": 8.832585070764002e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16450080,
      "step": 28955
    },
    {
      "epoch": 508.070796460177,
      "grad_norm": 2.3649431568628643e-07,
      "learning_rate": 8.825098063024045e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16452536,
      "step": 28960
    },
    {
      "epoch": 508.1592920353982,
      "grad_norm": 1.6764089139087446e-07,
      "learning_rate": 8.817613549657244e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16455144,
      "step": 28965
    },
    {
      "epoch": 508.24778761061947,
      "grad_norm": 5.774523401669285e-07,
      "learning_rate": 8.810131531817783e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16458168,
      "step": 28970
    },
    {
      "epoch": 508.3362831858407,
      "grad_norm": 1.514033669991477e-06,
      "learning_rate": 8.802652010659496e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16461112,
      "step": 28975
    },
    {
      "epoch": 508.42477876106193,
      "grad_norm": 1.1798768184689834e-07,
      "learning_rate": 8.795174987335827e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16464376,
      "step": 28980
    },
    {
      "epoch": 508.5132743362832,
      "grad_norm": 6.195385253704444e-07,
      "learning_rate": 8.787700462999807e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16467480,
      "step": 28985
    },
    {
      "epoch": 508.60176991150445,
      "grad_norm": 2.370536094531417e-06,
      "learning_rate": 8.780228438804122e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16470344,
      "step": 28990
    },
    {
      "epoch": 508.69026548672565,
      "grad_norm": 2.0935206066496903e-06,
      "learning_rate": 8.772758915901032e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16472792,
      "step": 28995
    },
    {
      "epoch": 508.7787610619469,
      "grad_norm": 2.7074190711573465e-07,
      "learning_rate": 8.765291895442443e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16475624,
      "step": 29000
    },
    {
      "epoch": 508.7787610619469,
      "eval_loss": 0.9256235361099243,
      "eval_runtime": 1.0639,
      "eval_samples_per_second": 23.499,
      "eval_steps_per_second": 12.22,
      "num_input_tokens_seen": 16475624,
      "step": 29000
    },
    {
      "epoch": 508.86725663716817,
      "grad_norm": 2.704438429645961e-06,
      "learning_rate": 8.75782737857987e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16478728,
      "step": 29005
    },
    {
      "epoch": 508.95575221238937,
      "grad_norm": 4.106492724531563e-06,
      "learning_rate": 8.750365366464425e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16481352,
      "step": 29010
    },
    {
      "epoch": 509.0353982300885,
      "grad_norm": 3.162082862218085e-07,
      "learning_rate": 8.742905860246838e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16483560,
      "step": 29015
    },
    {
      "epoch": 509.12389380530976,
      "grad_norm": 3.279935185673821e-07,
      "learning_rate": 8.735448861077478e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16486200,
      "step": 29020
    },
    {
      "epoch": 509.21238938053096,
      "grad_norm": 7.178368264249002e-07,
      "learning_rate": 8.727994370106288e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16488856,
      "step": 29025
    },
    {
      "epoch": 509.3008849557522,
      "grad_norm": 2.7034704430661805e-07,
      "learning_rate": 8.720542388482861e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16491672,
      "step": 29030
    },
    {
      "epoch": 509.3893805309734,
      "grad_norm": 1.7730587842379464e-06,
      "learning_rate": 8.71309291735637e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16494904,
      "step": 29035
    },
    {
      "epoch": 509.4778761061947,
      "grad_norm": 1.1627813591985614e-06,
      "learning_rate": 8.705645957875621e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16497816,
      "step": 29040
    },
    {
      "epoch": 509.56637168141594,
      "grad_norm": 1.521054514341813e-06,
      "learning_rate": 8.698201511189048e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16500984,
      "step": 29045
    },
    {
      "epoch": 509.65486725663715,
      "grad_norm": 5.028557552577695e-07,
      "learning_rate": 8.690759578444649e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16503864,
      "step": 29050
    },
    {
      "epoch": 509.7433628318584,
      "grad_norm": 2.7877081265614834e-07,
      "learning_rate": 8.68332016079008e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16507080,
      "step": 29055
    },
    {
      "epoch": 509.83185840707966,
      "grad_norm": 5.76797901885584e-07,
      "learning_rate": 8.6758832593726e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16509768,
      "step": 29060
    },
    {
      "epoch": 509.92035398230087,
      "grad_norm": 2.740846980486822e-07,
      "learning_rate": 8.668448875339053e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16512296,
      "step": 29065
    },
    {
      "epoch": 510.0,
      "grad_norm": 5.683767199116119e-07,
      "learning_rate": 8.661017009835933e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16514688,
      "step": 29070
    },
    {
      "epoch": 510.08849557522126,
      "grad_norm": 7.417308438562031e-07,
      "learning_rate": 8.653587664009311e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16517760,
      "step": 29075
    },
    {
      "epoch": 510.17699115044246,
      "grad_norm": 6.732306587764469e-07,
      "learning_rate": 8.646160839004902e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16520672,
      "step": 29080
    },
    {
      "epoch": 510.2654867256637,
      "grad_norm": 1.0033847956947284e-06,
      "learning_rate": 8.638736535967998e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16523504,
      "step": 29085
    },
    {
      "epoch": 510.353982300885,
      "grad_norm": 7.536381758654898e-07,
      "learning_rate": 8.631314756043535e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16526496,
      "step": 29090
    },
    {
      "epoch": 510.4424778761062,
      "grad_norm": 1.1028201925000758e-06,
      "learning_rate": 8.62389550037603e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16529376,
      "step": 29095
    },
    {
      "epoch": 510.53097345132744,
      "grad_norm": 3.4369924151178566e-07,
      "learning_rate": 8.616478770109646e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16532480,
      "step": 29100
    },
    {
      "epoch": 510.6194690265487,
      "grad_norm": 4.4420963263291924e-07,
      "learning_rate": 8.609064566388111e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16535088,
      "step": 29105
    },
    {
      "epoch": 510.7079646017699,
      "grad_norm": 2.7435004312792444e-07,
      "learning_rate": 8.601652890354815e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16538384,
      "step": 29110
    },
    {
      "epoch": 510.79646017699116,
      "grad_norm": 2.8552796038638917e-07,
      "learning_rate": 8.594243743152705e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16541168,
      "step": 29115
    },
    {
      "epoch": 510.88495575221236,
      "grad_norm": 6.672252084172214e-07,
      "learning_rate": 8.58683712592438e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16543808,
      "step": 29120
    },
    {
      "epoch": 510.9734513274336,
      "grad_norm": 6.661179554612318e-07,
      "learning_rate": 8.579433039812037e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16546624,
      "step": 29125
    },
    {
      "epoch": 511.05309734513276,
      "grad_norm": 4.17854010947849e-07,
      "learning_rate": 8.572031485957466e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16548824,
      "step": 29130
    },
    {
      "epoch": 511.14159292035396,
      "grad_norm": 7.953205738431279e-08,
      "learning_rate": 8.564632465502084e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16551416,
      "step": 29135
    },
    {
      "epoch": 511.2300884955752,
      "grad_norm": 7.428311050716729e-07,
      "learning_rate": 8.557235979586928e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16554536,
      "step": 29140
    },
    {
      "epoch": 511.3185840707965,
      "grad_norm": 1.3474289062287426e-06,
      "learning_rate": 8.549842029352606e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16557032,
      "step": 29145
    },
    {
      "epoch": 511.4070796460177,
      "grad_norm": 1.2488542324717855e-06,
      "learning_rate": 8.542450615939376e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16559880,
      "step": 29150
    },
    {
      "epoch": 511.49557522123894,
      "grad_norm": 4.238449378135556e-07,
      "learning_rate": 8.535061740487082e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16562824,
      "step": 29155
    },
    {
      "epoch": 511.5840707964602,
      "grad_norm": 1.2043757351420936e-06,
      "learning_rate": 8.527675404135168e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16565800,
      "step": 29160
    },
    {
      "epoch": 511.6725663716814,
      "grad_norm": 6.497863864751707e-07,
      "learning_rate": 8.520291608022724e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16568408,
      "step": 29165
    },
    {
      "epoch": 511.76106194690266,
      "grad_norm": 3.088710798238026e-07,
      "learning_rate": 8.512910353288398e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16571496,
      "step": 29170
    },
    {
      "epoch": 511.8495575221239,
      "grad_norm": 3.192764950199489e-07,
      "learning_rate": 8.505531641070486e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16575064,
      "step": 29175
    },
    {
      "epoch": 511.9380530973451,
      "grad_norm": 7.979828637871833e-07,
      "learning_rate": 8.498155472506885e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16577848,
      "step": 29180
    },
    {
      "epoch": 512.0176991150443,
      "grad_norm": 5.56512361526984e-07,
      "learning_rate": 8.49078184873508e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16580344,
      "step": 29185
    },
    {
      "epoch": 512.1061946902655,
      "grad_norm": 6.576561304427742e-07,
      "learning_rate": 8.483410770892188e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16583176,
      "step": 29190
    },
    {
      "epoch": 512.1946902654868,
      "grad_norm": 1.9459416762401816e-06,
      "learning_rate": 8.476042240114909e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16586280,
      "step": 29195
    },
    {
      "epoch": 512.2831858407079,
      "grad_norm": 5.788293151454127e-07,
      "learning_rate": 8.468676257539568e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16588984,
      "step": 29200
    },
    {
      "epoch": 512.2831858407079,
      "eval_loss": 0.9239405989646912,
      "eval_runtime": 1.0615,
      "eval_samples_per_second": 23.551,
      "eval_steps_per_second": 12.247,
      "num_input_tokens_seen": 16588984,
      "step": 29200
    },
    {
      "epoch": 512.3716814159292,
      "grad_norm": 2.2662145227059227e-07,
      "learning_rate": 8.4613128243021e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16591880,
      "step": 29205
    },
    {
      "epoch": 512.4601769911504,
      "grad_norm": 5.132996534484846e-07,
      "learning_rate": 8.453951941538028e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16595352,
      "step": 29210
    },
    {
      "epoch": 512.5486725663717,
      "grad_norm": 1.4547205182680045e-06,
      "learning_rate": 8.446593610382495e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16597560,
      "step": 29215
    },
    {
      "epoch": 512.637168141593,
      "grad_norm": 4.127855106617062e-07,
      "learning_rate": 8.439237831970259e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16600440,
      "step": 29220
    },
    {
      "epoch": 512.7256637168142,
      "grad_norm": 9.2518794190255e-06,
      "learning_rate": 8.431884607435667e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16603464,
      "step": 29225
    },
    {
      "epoch": 512.8141592920354,
      "grad_norm": 1.4336677622850402e-07,
      "learning_rate": 8.424533937912665e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16606360,
      "step": 29230
    },
    {
      "epoch": 512.9026548672566,
      "grad_norm": 2.7307098093842797e-07,
      "learning_rate": 8.41718582453484e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16609224,
      "step": 29235
    },
    {
      "epoch": 512.9911504424779,
      "grad_norm": 2.3679579896906944e-07,
      "learning_rate": 8.409840268435346e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16612152,
      "step": 29240
    },
    {
      "epoch": 513.070796460177,
      "grad_norm": 1.2931330957144382e-06,
      "learning_rate": 8.402497270746976e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16614920,
      "step": 29245
    },
    {
      "epoch": 513.1592920353983,
      "grad_norm": 4.957957457918383e-07,
      "learning_rate": 8.395156832602095e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16617624,
      "step": 29250
    },
    {
      "epoch": 513.2477876106195,
      "grad_norm": 4.524523774307454e-06,
      "learning_rate": 8.387818955132707e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16620296,
      "step": 29255
    },
    {
      "epoch": 513.3362831858407,
      "grad_norm": 1.5950784870710777e-07,
      "learning_rate": 8.38048363947039e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16623352,
      "step": 29260
    },
    {
      "epoch": 513.4247787610619,
      "grad_norm": 1.7125263411799097e-06,
      "learning_rate": 8.373150886746351e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16626120,
      "step": 29265
    },
    {
      "epoch": 513.5132743362832,
      "grad_norm": 3.5529487263374904e-07,
      "learning_rate": 8.365820698091397e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16629112,
      "step": 29270
    },
    {
      "epoch": 513.6017699115044,
      "grad_norm": 5.008399739381275e-07,
      "learning_rate": 8.358493074635922e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16631800,
      "step": 29275
    },
    {
      "epoch": 513.6902654867257,
      "grad_norm": 7.885046215960756e-07,
      "learning_rate": 8.351168017509948e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16634680,
      "step": 29280
    },
    {
      "epoch": 513.7787610619469,
      "grad_norm": 2.9355246056184114e-07,
      "learning_rate": 8.343845527843094e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16637736,
      "step": 29285
    },
    {
      "epoch": 513.8672566371681,
      "grad_norm": 1.595967660250608e-07,
      "learning_rate": 8.336525606764566e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16640120,
      "step": 29290
    },
    {
      "epoch": 513.9557522123894,
      "grad_norm": 1.3341275462153135e-06,
      "learning_rate": 8.329208255403204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16643400,
      "step": 29295
    },
    {
      "epoch": 514.0353982300885,
      "grad_norm": 1.9478284229990095e-07,
      "learning_rate": 8.321893474887426e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16645576,
      "step": 29300
    },
    {
      "epoch": 514.1238938053098,
      "grad_norm": 8.399973125960969e-07,
      "learning_rate": 8.31458126634526e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16648184,
      "step": 29305
    },
    {
      "epoch": 514.212389380531,
      "grad_norm": 6.115697601671854e-07,
      "learning_rate": 8.30727163090435e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16650536,
      "step": 29310
    },
    {
      "epoch": 514.3008849557522,
      "grad_norm": 2.0872691663953447e-07,
      "learning_rate": 8.29996456969192e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16653352,
      "step": 29315
    },
    {
      "epoch": 514.3893805309734,
      "grad_norm": 1.1550930594239617e-06,
      "learning_rate": 8.292660083834818e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16656040,
      "step": 29320
    },
    {
      "epoch": 514.4778761061947,
      "grad_norm": 1.3901902775614872e-06,
      "learning_rate": 8.2853581744595e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16659160,
      "step": 29325
    },
    {
      "epoch": 514.566371681416,
      "grad_norm": 3.2156444262909645e-07,
      "learning_rate": 8.278058842691991e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16662312,
      "step": 29330
    },
    {
      "epoch": 514.6548672566372,
      "grad_norm": 4.736670575766766e-07,
      "learning_rate": 8.27076208965796e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16665432,
      "step": 29335
    },
    {
      "epoch": 514.7433628318585,
      "grad_norm": 7.225758054119069e-07,
      "learning_rate": 8.263467916482637e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16668440,
      "step": 29340
    },
    {
      "epoch": 514.8318584070796,
      "grad_norm": 4.363899734016741e-06,
      "learning_rate": 8.256176324290885e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16671416,
      "step": 29345
    },
    {
      "epoch": 514.9203539823009,
      "grad_norm": 5.387388955568895e-05,
      "learning_rate": 8.248887314207168e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16674584,
      "step": 29350
    },
    {
      "epoch": 515.0,
      "grad_norm": 1.4023554513187264e-06,
      "learning_rate": 8.24160088735553e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16676800,
      "step": 29355
    },
    {
      "epoch": 515.0884955752213,
      "grad_norm": 1.0583713674350292e-06,
      "learning_rate": 8.234317044859629e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16679280,
      "step": 29360
    },
    {
      "epoch": 515.1769911504425,
      "grad_norm": 1.5582891421672684e-07,
      "learning_rate": 8.227035787842744e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16681936,
      "step": 29365
    },
    {
      "epoch": 515.2654867256637,
      "grad_norm": 2.1462425081608671e-07,
      "learning_rate": 8.219757117427721e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16684608,
      "step": 29370
    },
    {
      "epoch": 515.3539823008849,
      "grad_norm": 1.0194433031074368e-07,
      "learning_rate": 8.212481034737014e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16688048,
      "step": 29375
    },
    {
      "epoch": 515.4424778761062,
      "grad_norm": 8.942789690991049e-07,
      "learning_rate": 8.205207540892707e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16691232,
      "step": 29380
    },
    {
      "epoch": 515.5309734513274,
      "grad_norm": 7.365633791778237e-06,
      "learning_rate": 8.197936637016442e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16693840,
      "step": 29385
    },
    {
      "epoch": 515.6194690265487,
      "grad_norm": 1.1361951237631729e-06,
      "learning_rate": 8.190668324229508e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16696720,
      "step": 29390
    },
    {
      "epoch": 515.70796460177,
      "grad_norm": 6.339534479593567e-07,
      "learning_rate": 8.183402603652749e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16699984,
      "step": 29395
    },
    {
      "epoch": 515.7964601769911,
      "grad_norm": 1.741921892062237e-06,
      "learning_rate": 8.176139476406635e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16702496,
      "step": 29400
    },
    {
      "epoch": 515.7964601769911,
      "eval_loss": 0.9102046489715576,
      "eval_runtime": 1.0629,
      "eval_samples_per_second": 23.52,
      "eval_steps_per_second": 12.231,
      "num_input_tokens_seen": 16702496,
      "step": 29400
    },
    {
      "epoch": 515.8849557522124,
      "grad_norm": 7.314197318919469e-07,
      "learning_rate": 8.16887894361125e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16705664,
      "step": 29405
    },
    {
      "epoch": 515.9734513274336,
      "grad_norm": 3.376747486072418e-07,
      "learning_rate": 8.161621006386233e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16708528,
      "step": 29410
    },
    {
      "epoch": 516.0530973451328,
      "grad_norm": 3.372342973761988e-07,
      "learning_rate": 8.154365665850869e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16710584,
      "step": 29415
    },
    {
      "epoch": 516.141592920354,
      "grad_norm": 1.1362328677932965e-06,
      "learning_rate": 8.147112923124005e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16713224,
      "step": 29420
    },
    {
      "epoch": 516.2300884955753,
      "grad_norm": 2.830643779816455e-06,
      "learning_rate": 8.13986277932412e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16716648,
      "step": 29425
    },
    {
      "epoch": 516.3185840707964,
      "grad_norm": 3.282446527919092e-07,
      "learning_rate": 8.132615235569277e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16719608,
      "step": 29430
    },
    {
      "epoch": 516.4070796460177,
      "grad_norm": 4.4576449909072835e-07,
      "learning_rate": 8.125370292977124e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16722216,
      "step": 29435
    },
    {
      "epoch": 516.4955752212389,
      "grad_norm": 5.243101668384043e-07,
      "learning_rate": 8.118127952664944e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16725176,
      "step": 29440
    },
    {
      "epoch": 516.5840707964602,
      "grad_norm": 2.8295826837165805e-07,
      "learning_rate": 8.110888215749574e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16727832,
      "step": 29445
    },
    {
      "epoch": 516.6725663716815,
      "grad_norm": 5.268682343739783e-07,
      "learning_rate": 8.10365108334749e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16730760,
      "step": 29450
    },
    {
      "epoch": 516.7610619469026,
      "grad_norm": 1.2228541663716896e-06,
      "learning_rate": 8.096416556574743e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16733416,
      "step": 29455
    },
    {
      "epoch": 516.8495575221239,
      "grad_norm": 3.275052620210772e-07,
      "learning_rate": 8.08918463654698e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16736632,
      "step": 29460
    },
    {
      "epoch": 516.9380530973451,
      "grad_norm": 1.1576256611078861e-06,
      "learning_rate": 8.081955324379458e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16739592,
      "step": 29465
    },
    {
      "epoch": 517.0176991150443,
      "grad_norm": 1.818203031689336e-06,
      "learning_rate": 8.074728621187039e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16741720,
      "step": 29470
    },
    {
      "epoch": 517.1061946902655,
      "grad_norm": 3.277597215856076e-07,
      "learning_rate": 8.067504528084158e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16744920,
      "step": 29475
    },
    {
      "epoch": 517.1946902654868,
      "grad_norm": 4.195331655409973e-07,
      "learning_rate": 8.060283046184861e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16747704,
      "step": 29480
    },
    {
      "epoch": 517.2831858407079,
      "grad_norm": 1.375132114844746e-06,
      "learning_rate": 8.053064176602806e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16750760,
      "step": 29485
    },
    {
      "epoch": 517.3716814159292,
      "grad_norm": 4.1433060005147127e-07,
      "learning_rate": 8.045847920451216e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16753672,
      "step": 29490
    },
    {
      "epoch": 517.4601769911504,
      "grad_norm": 1.1038705451937858e-06,
      "learning_rate": 8.038634278842944e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16756328,
      "step": 29495
    },
    {
      "epoch": 517.5486725663717,
      "grad_norm": 2.806770282859361e-07,
      "learning_rate": 8.031423252890408e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16758728,
      "step": 29500
    },
    {
      "epoch": 517.637168141593,
      "grad_norm": 3.798760644713184e-06,
      "learning_rate": 8.024214843705646e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16761656,
      "step": 29505
    },
    {
      "epoch": 517.7256637168142,
      "grad_norm": 2.0092676322747138e-07,
      "learning_rate": 8.017009052400295e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16764280,
      "step": 29510
    },
    {
      "epoch": 517.8141592920354,
      "grad_norm": 9.260010642719863e-07,
      "learning_rate": 8.00980588008557e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16767160,
      "step": 29515
    },
    {
      "epoch": 517.9026548672566,
      "grad_norm": 7.612411536683794e-07,
      "learning_rate": 8.002605327872282e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16770168,
      "step": 29520
    },
    {
      "epoch": 517.9911504424779,
      "grad_norm": 1.3396206668403465e-06,
      "learning_rate": 7.995407396870862e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16773192,
      "step": 29525
    },
    {
      "epoch": 518.070796460177,
      "grad_norm": 6.646975521107379e-07,
      "learning_rate": 7.988212088191307e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16775816,
      "step": 29530
    },
    {
      "epoch": 518.1592920353983,
      "grad_norm": 1.472098574595293e-06,
      "learning_rate": 7.98101940294324e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16778696,
      "step": 29535
    },
    {
      "epoch": 518.2477876106195,
      "grad_norm": 1.6585980233685405e-07,
      "learning_rate": 7.973829342235847e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16781720,
      "step": 29540
    },
    {
      "epoch": 518.3362831858407,
      "grad_norm": 6.989494636400195e-07,
      "learning_rate": 7.966641907177936e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16785320,
      "step": 29545
    },
    {
      "epoch": 518.4247787610619,
      "grad_norm": 2.2240269004214497e-07,
      "learning_rate": 7.959457098877901e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16788184,
      "step": 29550
    },
    {
      "epoch": 518.5132743362832,
      "grad_norm": 1.0039433107067453e-07,
      "learning_rate": 7.952274918443719e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16790488,
      "step": 29555
    },
    {
      "epoch": 518.6017699115044,
      "grad_norm": 2.53162170338328e-07,
      "learning_rate": 7.945095366982983e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16793080,
      "step": 29560
    },
    {
      "epoch": 518.6902654867257,
      "grad_norm": 7.95557934907265e-06,
      "learning_rate": 7.937918445602871e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16796024,
      "step": 29565
    },
    {
      "epoch": 518.7787610619469,
      "grad_norm": 5.526288191504136e-07,
      "learning_rate": 7.930744155410145e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16799032,
      "step": 29570
    },
    {
      "epoch": 518.8672566371681,
      "grad_norm": 2.604715518828016e-07,
      "learning_rate": 7.923572497511181e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16802152,
      "step": 29575
    },
    {
      "epoch": 518.9557522123894,
      "grad_norm": 2.1488209256403934e-07,
      "learning_rate": 7.916403473011927e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16804712,
      "step": 29580
    },
    {
      "epoch": 519.0353982300885,
      "grad_norm": 1.4719734053869615e-06,
      "learning_rate": 7.909237083017953e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16806992,
      "step": 29585
    },
    {
      "epoch": 519.1238938053098,
      "grad_norm": 1.5608146952672541e-07,
      "learning_rate": 7.902073328634389e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16810432,
      "step": 29590
    },
    {
      "epoch": 519.212389380531,
      "grad_norm": 5.988499083287024e-07,
      "learning_rate": 7.894912210965987e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16813504,
      "step": 29595
    },
    {
      "epoch": 519.3008849557522,
      "grad_norm": 1.0831198551386478e-06,
      "learning_rate": 7.887753731117075e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16816272,
      "step": 29600
    },
    {
      "epoch": 519.3008849557522,
      "eval_loss": 0.9127913117408752,
      "eval_runtime": 1.084,
      "eval_samples_per_second": 23.064,
      "eval_steps_per_second": 11.993,
      "num_input_tokens_seen": 16816272,
      "step": 29600
    },
    {
      "epoch": 519.3893805309734,
      "grad_norm": 1.9333181171532487e-06,
      "learning_rate": 7.880597890191587e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16818864,
      "step": 29605
    },
    {
      "epoch": 519.4778761061947,
      "grad_norm": 2.464781516664516e-07,
      "learning_rate": 7.873444689293036e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16821888,
      "step": 29610
    },
    {
      "epoch": 519.566371681416,
      "grad_norm": 3.1037916414788924e-07,
      "learning_rate": 7.866294129524548e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16824464,
      "step": 29615
    },
    {
      "epoch": 519.6548672566372,
      "grad_norm": 1.236087200595648e-06,
      "learning_rate": 7.859146211988811e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16827456,
      "step": 29620
    },
    {
      "epoch": 519.7433628318585,
      "grad_norm": 4.857160547544481e-06,
      "learning_rate": 7.852000937788134e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16830080,
      "step": 29625
    },
    {
      "epoch": 519.8318584070796,
      "grad_norm": 3.765586029658152e-07,
      "learning_rate": 7.844858308024416e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16832912,
      "step": 29630
    },
    {
      "epoch": 519.9203539823009,
      "grad_norm": 4.593340463543427e-07,
      "learning_rate": 7.837718323799122e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16835984,
      "step": 29635
    },
    {
      "epoch": 520.0,
      "grad_norm": 1.125559606407478e-06,
      "learning_rate": 7.83058098621334e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16838416,
      "step": 29640
    },
    {
      "epoch": 520.0884955752213,
      "grad_norm": 1.537888010716415e-06,
      "learning_rate": 7.823446296367739e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16841680,
      "step": 29645
    },
    {
      "epoch": 520.1769911504425,
      "grad_norm": 5.045512807555497e-07,
      "learning_rate": 7.81631425536257e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16844688,
      "step": 29650
    },
    {
      "epoch": 520.2654867256637,
      "grad_norm": 4.03419164740626e-07,
      "learning_rate": 7.809184864297689e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16847440,
      "step": 29655
    },
    {
      "epoch": 520.3539823008849,
      "grad_norm": 2.0651125964832318e-07,
      "learning_rate": 7.802058124272532e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16850000,
      "step": 29660
    },
    {
      "epoch": 520.4424778761062,
      "grad_norm": 5.578119726123987e-07,
      "learning_rate": 7.79493403638614e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16852880,
      "step": 29665
    },
    {
      "epoch": 520.5309734513274,
      "grad_norm": 4.085743228188221e-07,
      "learning_rate": 7.787812601737132e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16855152,
      "step": 29670
    },
    {
      "epoch": 520.6194690265487,
      "grad_norm": 8.029588798308396e-07,
      "learning_rate": 7.780693821423715e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16857888,
      "step": 29675
    },
    {
      "epoch": 520.70796460177,
      "grad_norm": 8.682266070536571e-07,
      "learning_rate": 7.773577696543705e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16860784,
      "step": 29680
    },
    {
      "epoch": 520.7964601769911,
      "grad_norm": 1.2160707285602257e-07,
      "learning_rate": 7.7664642281945e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16863808,
      "step": 29685
    },
    {
      "epoch": 520.8849557522124,
      "grad_norm": 5.876758564227202e-07,
      "learning_rate": 7.759353417473072e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16866656,
      "step": 29690
    },
    {
      "epoch": 520.9734513274336,
      "grad_norm": 5.45710292954027e-07,
      "learning_rate": 7.752245265476016e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16869920,
      "step": 29695
    },
    {
      "epoch": 521.0530973451328,
      "grad_norm": 1.296799609917798e-06,
      "learning_rate": 7.745139773299481e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16872592,
      "step": 29700
    },
    {
      "epoch": 521.141592920354,
      "grad_norm": 3.973701439008437e-07,
      "learning_rate": 7.738036942039232e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16875248,
      "step": 29705
    },
    {
      "epoch": 521.2300884955753,
      "grad_norm": 5.299504482536577e-05,
      "learning_rate": 7.73093677279062e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16877952,
      "step": 29710
    },
    {
      "epoch": 521.3185840707964,
      "grad_norm": 1.2300505147777585e-07,
      "learning_rate": 7.72383926664857e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16880848,
      "step": 29715
    },
    {
      "epoch": 521.4070796460177,
      "grad_norm": 1.7679748509635829e-07,
      "learning_rate": 7.716744424707606e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16883664,
      "step": 29720
    },
    {
      "epoch": 521.4955752212389,
      "grad_norm": 3.404639414839039e-07,
      "learning_rate": 7.709652248061858e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16886560,
      "step": 29725
    },
    {
      "epoch": 521.5840707964602,
      "grad_norm": 3.084642457906739e-06,
      "learning_rate": 7.702562737805017e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16889408,
      "step": 29730
    },
    {
      "epoch": 521.6725663716815,
      "grad_norm": 1.2662021617870778e-06,
      "learning_rate": 7.695475895030365e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16892192,
      "step": 29735
    },
    {
      "epoch": 521.7610619469026,
      "grad_norm": 5.28898453922011e-06,
      "learning_rate": 7.6883917208308e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16894688,
      "step": 29740
    },
    {
      "epoch": 521.8495575221239,
      "grad_norm": 7.46566968246043e-07,
      "learning_rate": 7.681310216298778e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16897888,
      "step": 29745
    },
    {
      "epoch": 521.9380530973451,
      "grad_norm": 4.985395776202495e-07,
      "learning_rate": 7.674231382526367e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16901152,
      "step": 29750
    },
    {
      "epoch": 522.0176991150443,
      "grad_norm": 5.65361176541046e-07,
      "learning_rate": 7.667155220605198e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16903504,
      "step": 29755
    },
    {
      "epoch": 522.1061946902655,
      "grad_norm": 2.9209456897660857e-07,
      "learning_rate": 7.660081731626515e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16905968,
      "step": 29760
    },
    {
      "epoch": 522.1946902654868,
      "grad_norm": 1.5399268704641145e-06,
      "learning_rate": 7.653010916681141e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16908688,
      "step": 29765
    },
    {
      "epoch": 522.2831858407079,
      "grad_norm": 1.7246641448309674e-07,
      "learning_rate": 7.645942776859472e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16911840,
      "step": 29770
    },
    {
      "epoch": 522.3716814159292,
      "grad_norm": 3.377322741471289e-07,
      "learning_rate": 7.63887731325152e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16915072,
      "step": 29775
    },
    {
      "epoch": 522.4601769911504,
      "grad_norm": 2.535106773393636e-07,
      "learning_rate": 7.63181452694685e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16917680,
      "step": 29780
    },
    {
      "epoch": 522.5486725663717,
      "grad_norm": 2.921683517342899e-07,
      "learning_rate": 7.624754419034644e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16920240,
      "step": 29785
    },
    {
      "epoch": 522.637168141593,
      "grad_norm": 1.4593591401990125e-07,
      "learning_rate": 7.6176969906036645e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16922992,
      "step": 29790
    },
    {
      "epoch": 522.7256637168142,
      "grad_norm": 2.0213212792441482e-07,
      "learning_rate": 7.610642242742242e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16925824,
      "step": 29795
    },
    {
      "epoch": 522.8141592920354,
      "grad_norm": 4.391057473185356e-07,
      "learning_rate": 7.603590176538322e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16929072,
      "step": 29800
    },
    {
      "epoch": 522.8141592920354,
      "eval_loss": 0.9138714671134949,
      "eval_runtime": 1.0637,
      "eval_samples_per_second": 23.502,
      "eval_steps_per_second": 12.221,
      "num_input_tokens_seen": 16929072,
      "step": 29800
    },
    {
      "epoch": 522.9026548672566,
      "grad_norm": 1.4462955277849687e-06,
      "learning_rate": 7.596540793079404e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16932176,
      "step": 29805
    },
    {
      "epoch": 522.9911504424779,
      "grad_norm": 4.91660159696039e-07,
      "learning_rate": 7.5894940934526125e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16934736,
      "step": 29810
    },
    {
      "epoch": 523.070796460177,
      "grad_norm": 1.981435616471572e-06,
      "learning_rate": 7.582450078744621e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16937232,
      "step": 29815
    },
    {
      "epoch": 523.1592920353983,
      "grad_norm": 1.1361997849235195e-06,
      "learning_rate": 7.575408750041707e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16940544,
      "step": 29820
    },
    {
      "epoch": 523.2477876106195,
      "grad_norm": 2.747283076587337e-07,
      "learning_rate": 7.568370108429732e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16943328,
      "step": 29825
    },
    {
      "epoch": 523.3362831858407,
      "grad_norm": 4.081752251750004e-07,
      "learning_rate": 7.561334154994154e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16946032,
      "step": 29830
    },
    {
      "epoch": 523.4247787610619,
      "grad_norm": 1.2164222198407515e-06,
      "learning_rate": 7.55430089081999e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16949232,
      "step": 29835
    },
    {
      "epoch": 523.5132743362832,
      "grad_norm": 3.551188285655371e-07,
      "learning_rate": 7.547270316991864e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16952080,
      "step": 29840
    },
    {
      "epoch": 523.6017699115044,
      "grad_norm": 4.935689048579661e-07,
      "learning_rate": 7.5402424345939884e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16955328,
      "step": 29845
    },
    {
      "epoch": 523.6902654867257,
      "grad_norm": 6.601727591259987e-07,
      "learning_rate": 7.533217244710133e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16957632,
      "step": 29850
    },
    {
      "epoch": 523.7787610619469,
      "grad_norm": 4.1711135168043256e-07,
      "learning_rate": 7.52619474842369e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16960688,
      "step": 29855
    },
    {
      "epoch": 523.8672566371681,
      "grad_norm": 1.6192795726510667e-07,
      "learning_rate": 7.519174946817597e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16963424,
      "step": 29860
    },
    {
      "epoch": 523.9557522123894,
      "grad_norm": 4.5629367377841845e-06,
      "learning_rate": 7.512157840974407e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16966576,
      "step": 29865
    },
    {
      "epoch": 524.0353982300885,
      "grad_norm": 2.7552238179850974e-07,
      "learning_rate": 7.5051434319762496e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16968544,
      "step": 29870
    },
    {
      "epoch": 524.1238938053098,
      "grad_norm": 4.950560992256214e-07,
      "learning_rate": 7.498131720904822e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16971152,
      "step": 29875
    },
    {
      "epoch": 524.212389380531,
      "grad_norm": 5.500389761436963e-07,
      "learning_rate": 7.491122708841433e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16973984,
      "step": 29880
    },
    {
      "epoch": 524.3008849557522,
      "grad_norm": 9.341923146166664e-08,
      "learning_rate": 7.4841163968669524e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16977440,
      "step": 29885
    },
    {
      "epoch": 524.3893805309734,
      "grad_norm": 3.5920900700148195e-05,
      "learning_rate": 7.4771127860618355e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16980368,
      "step": 29890
    },
    {
      "epoch": 524.4778761061947,
      "grad_norm": 7.73663487052545e-07,
      "learning_rate": 7.470111877506139e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16982768,
      "step": 29895
    },
    {
      "epoch": 524.566371681416,
      "grad_norm": 3.599724891500955e-07,
      "learning_rate": 7.463113672279479e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16985440,
      "step": 29900
    },
    {
      "epoch": 524.6548672566372,
      "grad_norm": 5.933436000304937e-07,
      "learning_rate": 7.456118171461071e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16988608,
      "step": 29905
    },
    {
      "epoch": 524.7433628318585,
      "grad_norm": 3.022520616013935e-07,
      "learning_rate": 7.449125376129721e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16991488,
      "step": 29910
    },
    {
      "epoch": 524.8318584070796,
      "grad_norm": 7.424566206282179e-07,
      "learning_rate": 7.442135287363788e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16994688,
      "step": 29915
    },
    {
      "epoch": 524.9203539823009,
      "grad_norm": 2.18327068068902e-06,
      "learning_rate": 7.435147906241247e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16997584,
      "step": 29920
    },
    {
      "epoch": 525.0,
      "grad_norm": 2.0772331481566653e-06,
      "learning_rate": 7.428163233839624e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 16999896,
      "step": 29925
    },
    {
      "epoch": 525.0884955752213,
      "grad_norm": 4.436708707089565e-07,
      "learning_rate": 7.4211812712360525e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17002184,
      "step": 29930
    },
    {
      "epoch": 525.1769911504425,
      "grad_norm": 1.093539049179526e-05,
      "learning_rate": 7.4142020195072464e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17004472,
      "step": 29935
    },
    {
      "epoch": 525.2654867256637,
      "grad_norm": 6.573648647645314e-07,
      "learning_rate": 7.407225479729479e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17007528,
      "step": 29940
    },
    {
      "epoch": 525.3539823008849,
      "grad_norm": 3.30162492900854e-07,
      "learning_rate": 7.400251652978632e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17010824,
      "step": 29945
    },
    {
      "epoch": 525.4424778761062,
      "grad_norm": 2.9651855584234e-06,
      "learning_rate": 7.393280540330147e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17013688,
      "step": 29950
    },
    {
      "epoch": 525.5309734513274,
      "grad_norm": 5.763504304923117e-07,
      "learning_rate": 7.386312142859069e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17016264,
      "step": 29955
    },
    {
      "epoch": 525.6194690265487,
      "grad_norm": 6.689342058052716e-07,
      "learning_rate": 7.379346461640008e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17019576,
      "step": 29960
    },
    {
      "epoch": 525.70796460177,
      "grad_norm": 7.869720093367505e-07,
      "learning_rate": 7.372383497747149e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17022472,
      "step": 29965
    },
    {
      "epoch": 525.7964601769911,
      "grad_norm": 1.8242490114062093e-06,
      "learning_rate": 7.3654232522542775e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17025880,
      "step": 29970
    },
    {
      "epoch": 525.8849557522124,
      "grad_norm": 5.468310064316029e-06,
      "learning_rate": 7.358465726234756e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17028856,
      "step": 29975
    },
    {
      "epoch": 525.9734513274336,
      "grad_norm": 2.6370287287136307e-07,
      "learning_rate": 7.351510920761512e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17031848,
      "step": 29980
    },
    {
      "epoch": 526.0530973451328,
      "grad_norm": 2.301992338971104e-07,
      "learning_rate": 7.344558836907067e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17034336,
      "step": 29985
    },
    {
      "epoch": 526.141592920354,
      "grad_norm": 4.4658747810899513e-07,
      "learning_rate": 7.3376094757435285e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17036992,
      "step": 29990
    },
    {
      "epoch": 526.2300884955753,
      "grad_norm": 3.619961717049591e-07,
      "learning_rate": 7.330662838342561e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17039888,
      "step": 29995
    },
    {
      "epoch": 526.3185840707964,
      "grad_norm": 1.1175318377354415e-06,
      "learning_rate": 7.323718925775438e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17043120,
      "step": 30000
    },
    {
      "epoch": 526.3185840707964,
      "eval_loss": 0.9153329133987427,
      "eval_runtime": 1.0677,
      "eval_samples_per_second": 23.415,
      "eval_steps_per_second": 12.176,
      "num_input_tokens_seen": 17043120,
      "step": 30000
    },
    {
      "epoch": 526.4070796460177,
      "grad_norm": 4.329345983933308e-07,
      "learning_rate": 7.316777739112985e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17046224,
      "step": 30005
    },
    {
      "epoch": 526.4955752212389,
      "grad_norm": 1.18929324344208e-06,
      "learning_rate": 7.309839279425626e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17049360,
      "step": 30010
    },
    {
      "epoch": 526.5840707964602,
      "grad_norm": 0.00011670442472677678,
      "learning_rate": 7.302903547783366e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17052352,
      "step": 30015
    },
    {
      "epoch": 526.6725663716815,
      "grad_norm": 5.48227546914859e-07,
      "learning_rate": 7.2959705452557644e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17055312,
      "step": 30020
    },
    {
      "epoch": 526.7610619469026,
      "grad_norm": 7.516409255003964e-07,
      "learning_rate": 7.289040272911996e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17057760,
      "step": 30025
    },
    {
      "epoch": 526.8495575221239,
      "grad_norm": 1.4064377182876342e-06,
      "learning_rate": 7.282112731820789e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17060432,
      "step": 30030
    },
    {
      "epoch": 526.9380530973451,
      "grad_norm": 3.4642783930394216e-07,
      "learning_rate": 7.275187923050447e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17063104,
      "step": 30035
    },
    {
      "epoch": 527.0176991150443,
      "grad_norm": 1.6492849397309328e-07,
      "learning_rate": 7.268265847668879e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17065312,
      "step": 30040
    },
    {
      "epoch": 527.1061946902655,
      "grad_norm": 7.602809546369826e-07,
      "learning_rate": 7.261346506743538e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17068096,
      "step": 30045
    },
    {
      "epoch": 527.1946902654868,
      "grad_norm": 4.42918832277428e-07,
      "learning_rate": 7.254429901341486e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17070832,
      "step": 30050
    },
    {
      "epoch": 527.2831858407079,
      "grad_norm": 2.729083519170672e-07,
      "learning_rate": 7.247516032529356e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17073680,
      "step": 30055
    },
    {
      "epoch": 527.3716814159292,
      "grad_norm": 1.5519300404776004e-06,
      "learning_rate": 7.240604901373338e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17076512,
      "step": 30060
    },
    {
      "epoch": 527.4601769911504,
      "grad_norm": 1.1379122042853851e-05,
      "learning_rate": 7.233696508939223e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17079632,
      "step": 30065
    },
    {
      "epoch": 527.5486725663717,
      "grad_norm": 1.9841502307826886e-06,
      "learning_rate": 7.226790856292376e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17082176,
      "step": 30070
    },
    {
      "epoch": 527.637168141593,
      "grad_norm": 3.493940710086463e-07,
      "learning_rate": 7.219887944497727e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17085872,
      "step": 30075
    },
    {
      "epoch": 527.7256637168142,
      "grad_norm": 9.431611402987983e-08,
      "learning_rate": 7.2129877746198e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17089216,
      "step": 30080
    },
    {
      "epoch": 527.8141592920354,
      "grad_norm": 4.241219357936643e-07,
      "learning_rate": 7.20609034772268e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17091968,
      "step": 30085
    },
    {
      "epoch": 527.9026548672566,
      "grad_norm": 1.0376672889833571e-06,
      "learning_rate": 7.19919566487004e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17094224,
      "step": 30090
    },
    {
      "epoch": 527.9911504424779,
      "grad_norm": 4.3127386106789345e-07,
      "learning_rate": 7.192303727125132e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17097264,
      "step": 30095
    },
    {
      "epoch": 528.070796460177,
      "grad_norm": 9.586980240783305e-07,
      "learning_rate": 7.185414535550777e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17099152,
      "step": 30100
    },
    {
      "epoch": 528.1592920353983,
      "grad_norm": 4.6271813403109263e-07,
      "learning_rate": 7.178528091209363e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17101872,
      "step": 30105
    },
    {
      "epoch": 528.2477876106195,
      "grad_norm": 1.9088863609795226e-06,
      "learning_rate": 7.171644395162888e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17105024,
      "step": 30110
    },
    {
      "epoch": 528.3362831858407,
      "grad_norm": 2.723175498431374e-07,
      "learning_rate": 7.164763448472881e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17107536,
      "step": 30115
    },
    {
      "epoch": 528.4247787610619,
      "grad_norm": 3.4290286521354574e-07,
      "learning_rate": 7.157885252200491e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17111008,
      "step": 30120
    },
    {
      "epoch": 528.5132743362832,
      "grad_norm": 3.2529712257201027e-07,
      "learning_rate": 7.151009807406403e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17113520,
      "step": 30125
    },
    {
      "epoch": 528.6017699115044,
      "grad_norm": 7.155813932513411e-07,
      "learning_rate": 7.144137115150909e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17116144,
      "step": 30130
    },
    {
      "epoch": 528.6902654867257,
      "grad_norm": 1.6333166286131018e-07,
      "learning_rate": 7.1372671764938725e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17118912,
      "step": 30135
    },
    {
      "epoch": 528.7787610619469,
      "grad_norm": 4.0135591916623525e-07,
      "learning_rate": 7.130399992494705e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17121584,
      "step": 30140
    },
    {
      "epoch": 528.8672566371681,
      "grad_norm": 1.3363193147597485e-06,
      "learning_rate": 7.123535564212419e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17124800,
      "step": 30145
    },
    {
      "epoch": 528.9557522123894,
      "grad_norm": 2.772997049760306e-07,
      "learning_rate": 7.116673892705611e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17127600,
      "step": 30150
    },
    {
      "epoch": 529.0353982300885,
      "grad_norm": 6.184783387652715e-07,
      "learning_rate": 7.109814979032415e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17130488,
      "step": 30155
    },
    {
      "epoch": 529.1238938053098,
      "grad_norm": 1.7294053122896003e-07,
      "learning_rate": 7.102958824250577e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17133512,
      "step": 30160
    },
    {
      "epoch": 529.212389380531,
      "grad_norm": 5.792583692709741e-07,
      "learning_rate": 7.096105429417393e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17135688,
      "step": 30165
    },
    {
      "epoch": 529.3008849557522,
      "grad_norm": 1.3818147408528603e-06,
      "learning_rate": 7.0892547955897506e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17138888,
      "step": 30170
    },
    {
      "epoch": 529.3893805309734,
      "grad_norm": 4.439007454948296e-07,
      "learning_rate": 7.0824069238241e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17141656,
      "step": 30175
    },
    {
      "epoch": 529.4778761061947,
      "grad_norm": 2.920998554145626e-07,
      "learning_rate": 7.075561815176462e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17144344,
      "step": 30180
    },
    {
      "epoch": 529.566371681416,
      "grad_norm": 1.7629413378017489e-06,
      "learning_rate": 7.068719470702445e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17147272,
      "step": 30185
    },
    {
      "epoch": 529.6548672566372,
      "grad_norm": 9.188149761030218e-07,
      "learning_rate": 7.061879891457229e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17150568,
      "step": 30190
    },
    {
      "epoch": 529.7433628318585,
      "grad_norm": 1.2493691770032456e-07,
      "learning_rate": 7.0550430784955515e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17153800,
      "step": 30195
    },
    {
      "epoch": 529.8318584070796,
      "grad_norm": 1.9185458199899585e-07,
      "learning_rate": 7.048209032871752e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17156344,
      "step": 30200
    },
    {
      "epoch": 529.8318584070796,
      "eval_loss": 0.9343230724334717,
      "eval_runtime": 1.0691,
      "eval_samples_per_second": 23.385,
      "eval_steps_per_second": 12.16,
      "num_input_tokens_seen": 17156344,
      "step": 30200
    },
    {
      "epoch": 529.9203539823009,
      "grad_norm": 1.2614459592441563e-06,
      "learning_rate": 7.0413777556397055e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17159240,
      "step": 30205
    },
    {
      "epoch": 530.0,
      "grad_norm": 2.9721800274273846e-06,
      "learning_rate": 7.0345492478528925e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17161240,
      "step": 30210
    },
    {
      "epoch": 530.0884955752213,
      "grad_norm": 1.834272893574962e-07,
      "learning_rate": 7.02772351056436e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17164104,
      "step": 30215
    },
    {
      "epoch": 530.1769911504425,
      "grad_norm": 4.826243866773439e-07,
      "learning_rate": 7.020900544826709e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17167256,
      "step": 30220
    },
    {
      "epoch": 530.2654867256637,
      "grad_norm": 2.2056558179883723e-07,
      "learning_rate": 7.014080351692134e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17170360,
      "step": 30225
    },
    {
      "epoch": 530.3539823008849,
      "grad_norm": 2.322718273717328e-06,
      "learning_rate": 7.0072629322124024e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17173128,
      "step": 30230
    },
    {
      "epoch": 530.4424778761062,
      "grad_norm": 1.3458890180118033e-07,
      "learning_rate": 7.000448287438827e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17176024,
      "step": 30235
    },
    {
      "epoch": 530.5309734513274,
      "grad_norm": 7.830775530237588e-07,
      "learning_rate": 6.993636418422331e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17178920,
      "step": 30240
    },
    {
      "epoch": 530.6194690265487,
      "grad_norm": 1.9181469212981028e-07,
      "learning_rate": 6.986827326213383e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17181768,
      "step": 30245
    },
    {
      "epoch": 530.70796460177,
      "grad_norm": 1.2283886690056534e-06,
      "learning_rate": 6.9800210118620205e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17184536,
      "step": 30250
    },
    {
      "epoch": 530.7964601769911,
      "grad_norm": 1.8080193058267469e-06,
      "learning_rate": 6.973217476417876e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17187192,
      "step": 30255
    },
    {
      "epoch": 530.8849557522124,
      "grad_norm": 1.6412658965236915e-07,
      "learning_rate": 6.96641672093013e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17190344,
      "step": 30260
    },
    {
      "epoch": 530.9734513274336,
      "grad_norm": 8.48650324769551e-06,
      "learning_rate": 6.95961874644755e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17192936,
      "step": 30265
    },
    {
      "epoch": 531.0530973451328,
      "grad_norm": 3.316517336315883e-07,
      "learning_rate": 6.952823554018476e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17195168,
      "step": 30270
    },
    {
      "epoch": 531.141592920354,
      "grad_norm": 1.3819253581459634e-06,
      "learning_rate": 6.946031144690798e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17198016,
      "step": 30275
    },
    {
      "epoch": 531.2300884955753,
      "grad_norm": 1.1958824188695871e-06,
      "learning_rate": 6.939241519512005e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17200832,
      "step": 30280
    },
    {
      "epoch": 531.3185840707964,
      "grad_norm": 1.08052006453363e-06,
      "learning_rate": 6.932454679529129e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17205024,
      "step": 30285
    },
    {
      "epoch": 531.4070796460177,
      "grad_norm": 1.0133929890798754e-06,
      "learning_rate": 6.925670625788791e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17207392,
      "step": 30290
    },
    {
      "epoch": 531.4955752212389,
      "grad_norm": 2.6785053250932833e-06,
      "learning_rate": 6.918889359337186e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17210176,
      "step": 30295
    },
    {
      "epoch": 531.5840707964602,
      "grad_norm": 4.0132582057594846e-07,
      "learning_rate": 6.912110881220058e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17212864,
      "step": 30300
    },
    {
      "epoch": 531.6725663716815,
      "grad_norm": 1.0883812819884042e-06,
      "learning_rate": 6.905335192482735e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17215360,
      "step": 30305
    },
    {
      "epoch": 531.7610619469026,
      "grad_norm": 9.57201905293914e-07,
      "learning_rate": 6.8985622941701275e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17218048,
      "step": 30310
    },
    {
      "epoch": 531.8495575221239,
      "grad_norm": 1.4416613112189225e-06,
      "learning_rate": 6.89179218732669e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17220688,
      "step": 30315
    },
    {
      "epoch": 531.9380530973451,
      "grad_norm": 9.782356755749788e-07,
      "learning_rate": 6.8850248729964595e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17224112,
      "step": 30320
    },
    {
      "epoch": 532.0176991150443,
      "grad_norm": 2.784724131288385e-07,
      "learning_rate": 6.8782603522230314e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17226920,
      "step": 30325
    },
    {
      "epoch": 532.1061946902655,
      "grad_norm": 3.35006490104206e-07,
      "learning_rate": 6.871498626049591e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17229416,
      "step": 30330
    },
    {
      "epoch": 532.1946902654868,
      "grad_norm": 1.0389501312602079e-06,
      "learning_rate": 6.8647396955188875e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17232344,
      "step": 30335
    },
    {
      "epoch": 532.2831858407079,
      "grad_norm": 1.9333380407715595e-07,
      "learning_rate": 6.857983561673218e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17235544,
      "step": 30340
    },
    {
      "epoch": 532.3716814159292,
      "grad_norm": 3.602740719088615e-07,
      "learning_rate": 6.851230225554467e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17238456,
      "step": 30345
    },
    {
      "epoch": 532.4601769911504,
      "grad_norm": 3.621143207510613e-07,
      "learning_rate": 6.8444796882040946e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17241512,
      "step": 30350
    },
    {
      "epoch": 532.5486725663717,
      "grad_norm": 2.425795742055925e-07,
      "learning_rate": 6.837731950663106e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17244456,
      "step": 30355
    },
    {
      "epoch": 532.637168141593,
      "grad_norm": 7.604571123920323e-07,
      "learning_rate": 6.830987013972098e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17246920,
      "step": 30360
    },
    {
      "epoch": 532.7256637168142,
      "grad_norm": 8.745345780880598e-08,
      "learning_rate": 6.82424487917121e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17249880,
      "step": 30365
    },
    {
      "epoch": 532.8141592920354,
      "grad_norm": 3.6269620977691375e-06,
      "learning_rate": 6.8175055473001735e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17252920,
      "step": 30370
    },
    {
      "epoch": 532.9026548672566,
      "grad_norm": 1.0218635679848376e-06,
      "learning_rate": 6.8107690193982855e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17255336,
      "step": 30375
    },
    {
      "epoch": 532.9911504424779,
      "grad_norm": 3.8725676176909474e-07,
      "learning_rate": 6.804035296504385e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17258136,
      "step": 30380
    },
    {
      "epoch": 533.070796460177,
      "grad_norm": 5.31294404026994e-07,
      "learning_rate": 6.797304379656916e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17260336,
      "step": 30385
    },
    {
      "epoch": 533.1592920353983,
      "grad_norm": 2.8442798338801367e-07,
      "learning_rate": 6.790576269893861e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17262688,
      "step": 30390
    },
    {
      "epoch": 533.2477876106195,
      "grad_norm": 2.794780300519051e-07,
      "learning_rate": 6.783850968252772e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17266016,
      "step": 30395
    },
    {
      "epoch": 533.3362831858407,
      "grad_norm": 2.6958062449011777e-07,
      "learning_rate": 6.777128475770789e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17268656,
      "step": 30400
    },
    {
      "epoch": 533.3362831858407,
      "eval_loss": 0.9050721526145935,
      "eval_runtime": 1.0628,
      "eval_samples_per_second": 23.522,
      "eval_steps_per_second": 12.232,
      "num_input_tokens_seen": 17268656,
      "step": 30400
    },
    {
      "epoch": 533.4247787610619,
      "grad_norm": 8.60675640979025e-07,
      "learning_rate": 6.77040879348459e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17271920,
      "step": 30405
    },
    {
      "epoch": 533.5132743362832,
      "grad_norm": 1.835924763327057e-07,
      "learning_rate": 6.763691922430443e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17274720,
      "step": 30410
    },
    {
      "epoch": 533.6017699115044,
      "grad_norm": 2.3372572286461946e-06,
      "learning_rate": 6.756977863644178e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17277680,
      "step": 30415
    },
    {
      "epoch": 533.6902654867257,
      "grad_norm": 1.3227523822934018e-06,
      "learning_rate": 6.7502666181611804e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17280192,
      "step": 30420
    },
    {
      "epoch": 533.7787610619469,
      "grad_norm": 1.996969558604178e-07,
      "learning_rate": 6.743558187016405e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17283296,
      "step": 30425
    },
    {
      "epoch": 533.8672566371681,
      "grad_norm": 1.7312173383743357e-07,
      "learning_rate": 6.7368525712443925e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17286688,
      "step": 30430
    },
    {
      "epoch": 533.9557522123894,
      "grad_norm": 1.8895884750236291e-06,
      "learning_rate": 6.7301497718792155e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17289952,
      "step": 30435
    },
    {
      "epoch": 534.0353982300885,
      "grad_norm": 1.1619221140790614e-06,
      "learning_rate": 6.723449789954544e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17292320,
      "step": 30440
    },
    {
      "epoch": 534.1238938053098,
      "grad_norm": 1.6158992366399616e-06,
      "learning_rate": 6.716752626503586e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17294800,
      "step": 30445
    },
    {
      "epoch": 534.212389380531,
      "grad_norm": 3.957016474487318e-07,
      "learning_rate": 6.710058282559131e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17297888,
      "step": 30450
    },
    {
      "epoch": 534.3008849557522,
      "grad_norm": 3.542621698215953e-07,
      "learning_rate": 6.703366759153545e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17300704,
      "step": 30455
    },
    {
      "epoch": 534.3893805309734,
      "grad_norm": 5.140670509717893e-06,
      "learning_rate": 6.6966780573187335e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17303120,
      "step": 30460
    },
    {
      "epoch": 534.4778761061947,
      "grad_norm": 5.699969278794015e-07,
      "learning_rate": 6.689992178086174e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17305888,
      "step": 30465
    },
    {
      "epoch": 534.566371681416,
      "grad_norm": 3.2293402796312876e-07,
      "learning_rate": 6.683309122486925e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17309120,
      "step": 30470
    },
    {
      "epoch": 534.6548672566372,
      "grad_norm": 1.960750637408637e-07,
      "learning_rate": 6.676628891551584e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17312464,
      "step": 30475
    },
    {
      "epoch": 534.7433628318585,
      "grad_norm": 3.564505277608987e-07,
      "learning_rate": 6.6699514863103385e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17315360,
      "step": 30480
    },
    {
      "epoch": 534.8318584070796,
      "grad_norm": 3.390274287085049e-06,
      "learning_rate": 6.663276907792921e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17318048,
      "step": 30485
    },
    {
      "epoch": 534.9203539823009,
      "grad_norm": 2.236115079767842e-07,
      "learning_rate": 6.656605157028634e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17320976,
      "step": 30490
    },
    {
      "epoch": 535.0,
      "grad_norm": 4.5044635044177994e-07,
      "learning_rate": 6.649936235046358e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17323544,
      "step": 30495
    },
    {
      "epoch": 535.0884955752213,
      "grad_norm": 2.8801321150240256e-07,
      "learning_rate": 6.643270142874508e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17326568,
      "step": 30500
    },
    {
      "epoch": 535.1769911504425,
      "grad_norm": 1.4993894410508801e-06,
      "learning_rate": 6.636606881541094e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17329160,
      "step": 30505
    },
    {
      "epoch": 535.2654867256637,
      "grad_norm": 2.2074489436363365e-07,
      "learning_rate": 6.629946452073662e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17331704,
      "step": 30510
    },
    {
      "epoch": 535.3539823008849,
      "grad_norm": 4.6802398401268874e-07,
      "learning_rate": 6.6232888554993375e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17334616,
      "step": 30515
    },
    {
      "epoch": 535.4424778761062,
      "grad_norm": 5.358174348657485e-07,
      "learning_rate": 6.616634092844817e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17337672,
      "step": 30520
    },
    {
      "epoch": 535.5309734513274,
      "grad_norm": 1.6995014107124007e-07,
      "learning_rate": 6.609982165136331e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17340328,
      "step": 30525
    },
    {
      "epoch": 535.6194690265487,
      "grad_norm": 1.8341162899560004e-07,
      "learning_rate": 6.603333073399706e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17343304,
      "step": 30530
    },
    {
      "epoch": 535.70796460177,
      "grad_norm": 5.941911922491272e-07,
      "learning_rate": 6.596686818660308e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17346472,
      "step": 30535
    },
    {
      "epoch": 535.7964601769911,
      "grad_norm": 3.1812288625587826e-07,
      "learning_rate": 6.590043401943066e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17349448,
      "step": 30540
    },
    {
      "epoch": 535.8849557522124,
      "grad_norm": 6.65332947846764e-07,
      "learning_rate": 6.583402824272494e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17352248,
      "step": 30545
    },
    {
      "epoch": 535.9734513274336,
      "grad_norm": 1.401288898250641e-07,
      "learning_rate": 6.576765086672634e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17355080,
      "step": 30550
    },
    {
      "epoch": 536.0530973451328,
      "grad_norm": 1.5280369325409993e-06,
      "learning_rate": 6.57013019016712e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17357600,
      "step": 30555
    },
    {
      "epoch": 536.141592920354,
      "grad_norm": 1.2923239012252452e-07,
      "learning_rate": 6.563498135779142e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17360256,
      "step": 30560
    },
    {
      "epoch": 536.2300884955753,
      "grad_norm": 4.285693648853339e-07,
      "learning_rate": 6.556868924531431e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17363392,
      "step": 30565
    },
    {
      "epoch": 536.3185840707964,
      "grad_norm": 3.3051423997676466e-07,
      "learning_rate": 6.550242557446304e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17366256,
      "step": 30570
    },
    {
      "epoch": 536.4070796460177,
      "grad_norm": 3.785175692883058e-07,
      "learning_rate": 6.543619035545634e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17369040,
      "step": 30575
    },
    {
      "epoch": 536.4955752212389,
      "grad_norm": 2.1634790755342692e-05,
      "learning_rate": 6.53699835985084e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17372240,
      "step": 30580
    },
    {
      "epoch": 536.5840707964602,
      "grad_norm": 3.7908912986495125e-07,
      "learning_rate": 6.530380531382927e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17375264,
      "step": 30585
    },
    {
      "epoch": 536.6725663716815,
      "grad_norm": 2.2543346744896553e-07,
      "learning_rate": 6.523765551162433e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17378016,
      "step": 30590
    },
    {
      "epoch": 536.7610619469026,
      "grad_norm": 6.214331369847059e-05,
      "learning_rate": 6.517153420209476e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17380928,
      "step": 30595
    },
    {
      "epoch": 536.8495575221239,
      "grad_norm": 1.1875417982309955e-07,
      "learning_rate": 6.510544139543739e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17383696,
      "step": 30600
    },
    {
      "epoch": 536.8495575221239,
      "eval_loss": 0.9374831318855286,
      "eval_runtime": 1.0623,
      "eval_samples_per_second": 23.535,
      "eval_steps_per_second": 12.238,
      "num_input_tokens_seen": 17383696,
      "step": 30600
    },
    {
      "epoch": 536.9380530973451,
      "grad_norm": 4.534462618721591e-07,
      "learning_rate": 6.503937710184452e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17386624,
      "step": 30605
    },
    {
      "epoch": 537.0176991150443,
      "grad_norm": 9.36962096602656e-05,
      "learning_rate": 6.4973341331503954e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17389112,
      "step": 30610
    },
    {
      "epoch": 537.1061946902655,
      "grad_norm": 8.029381888263742e-07,
      "learning_rate": 6.490733409459942e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17392024,
      "step": 30615
    },
    {
      "epoch": 537.1946902654868,
      "grad_norm": 3.088627806846489e-07,
      "learning_rate": 6.484135540130995e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17394344,
      "step": 30620
    },
    {
      "epoch": 537.2831858407079,
      "grad_norm": 1.6380915894842474e-06,
      "learning_rate": 6.4775405261810364e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17397416,
      "step": 30625
    },
    {
      "epoch": 537.3716814159292,
      "grad_norm": 1.9028757378691807e-05,
      "learning_rate": 6.470948368627092e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17400328,
      "step": 30630
    },
    {
      "epoch": 537.4601769911504,
      "grad_norm": 1.209439659533018e-07,
      "learning_rate": 6.464359068485756e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17403080,
      "step": 30635
    },
    {
      "epoch": 537.5486725663717,
      "grad_norm": 3.374952655121888e-07,
      "learning_rate": 6.457772626773195e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17405448,
      "step": 30640
    },
    {
      "epoch": 537.637168141593,
      "grad_norm": 7.728592095190834e-07,
      "learning_rate": 6.451189044505104e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17408312,
      "step": 30645
    },
    {
      "epoch": 537.7256637168142,
      "grad_norm": 3.3285823519690894e-07,
      "learning_rate": 6.44460832269676e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17410936,
      "step": 30650
    },
    {
      "epoch": 537.8141592920354,
      "grad_norm": 5.804321290270309e-07,
      "learning_rate": 6.438030462363001e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17413768,
      "step": 30655
    },
    {
      "epoch": 537.9026548672566,
      "grad_norm": 1.9621757019194774e-06,
      "learning_rate": 6.431455464518205e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17416952,
      "step": 30660
    },
    {
      "epoch": 537.9911504424779,
      "grad_norm": 1.6193709484468854e-07,
      "learning_rate": 6.424883330176326e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17419864,
      "step": 30665
    },
    {
      "epoch": 538.070796460177,
      "grad_norm": 1.077091269507946e-06,
      "learning_rate": 6.418314060350864e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17422224,
      "step": 30670
    },
    {
      "epoch": 538.1592920353983,
      "grad_norm": 1.703202627822975e-07,
      "learning_rate": 6.4117476560548895e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17424848,
      "step": 30675
    },
    {
      "epoch": 538.2477876106195,
      "grad_norm": 2.493090107691387e-07,
      "learning_rate": 6.405184118301016e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17427600,
      "step": 30680
    },
    {
      "epoch": 538.3362831858407,
      "grad_norm": 1.5021053059172118e-06,
      "learning_rate": 6.398623448101434e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17429824,
      "step": 30685
    },
    {
      "epoch": 538.4247787610619,
      "grad_norm": 1.0414011057946482e-06,
      "learning_rate": 6.392065646467871e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17431984,
      "step": 30690
    },
    {
      "epoch": 538.5132743362832,
      "grad_norm": 2.023995875788387e-06,
      "learning_rate": 6.385510714411632e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17435168,
      "step": 30695
    },
    {
      "epoch": 538.6017699115044,
      "grad_norm": 1.6584435513777862e-07,
      "learning_rate": 6.378958652943559e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17438368,
      "step": 30700
    },
    {
      "epoch": 538.6902654867257,
      "grad_norm": 5.752484867116436e-06,
      "learning_rate": 6.3724094630740776e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17441264,
      "step": 30705
    },
    {
      "epoch": 538.7787610619469,
      "grad_norm": 9.966339575839811e-08,
      "learning_rate": 6.365863145813136e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17444256,
      "step": 30710
    },
    {
      "epoch": 538.8672566371681,
      "grad_norm": 2.25833076683557e-07,
      "learning_rate": 6.359319702170269e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17446960,
      "step": 30715
    },
    {
      "epoch": 538.9557522123894,
      "grad_norm": 5.261840669845697e-07,
      "learning_rate": 6.352779133154566e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17450416,
      "step": 30720
    },
    {
      "epoch": 539.0353982300885,
      "grad_norm": 4.816202476831677e-07,
      "learning_rate": 6.346241439774648e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17453288,
      "step": 30725
    },
    {
      "epoch": 539.1238938053098,
      "grad_norm": 5.226646635492216e-07,
      "learning_rate": 6.339706623038716e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17456168,
      "step": 30730
    },
    {
      "epoch": 539.212389380531,
      "grad_norm": 4.3772951130449655e-07,
      "learning_rate": 6.333174683954532e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17459080,
      "step": 30735
    },
    {
      "epoch": 539.3008849557522,
      "grad_norm": 2.4204436499530857e-07,
      "learning_rate": 6.326645623529387e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17462104,
      "step": 30740
    },
    {
      "epoch": 539.3893805309734,
      "grad_norm": 6.357400366141519e-07,
      "learning_rate": 6.320119442770156e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17464792,
      "step": 30745
    },
    {
      "epoch": 539.4778761061947,
      "grad_norm": 3.4126867376471637e-07,
      "learning_rate": 6.313596142683254e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17467960,
      "step": 30750
    },
    {
      "epoch": 539.566371681416,
      "grad_norm": 2.1032523989106267e-07,
      "learning_rate": 6.307075724274647e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17470616,
      "step": 30755
    },
    {
      "epoch": 539.6548672566372,
      "grad_norm": 5.467801429404062e-07,
      "learning_rate": 6.300558188549882e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17473512,
      "step": 30760
    },
    {
      "epoch": 539.7433628318585,
      "grad_norm": 2.183815922762733e-06,
      "learning_rate": 6.29404353651403e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17475992,
      "step": 30765
    },
    {
      "epoch": 539.8318584070796,
      "grad_norm": 4.361876051461877e-07,
      "learning_rate": 6.287531769171737e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17479144,
      "step": 30770
    },
    {
      "epoch": 539.9203539823009,
      "grad_norm": 5.090253580419812e-06,
      "learning_rate": 6.2810228875272045e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17481544,
      "step": 30775
    },
    {
      "epoch": 540.0,
      "grad_norm": 1.0281875262307949e-07,
      "learning_rate": 6.274516892584179e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17483856,
      "step": 30780
    },
    {
      "epoch": 540.0884955752213,
      "grad_norm": 3.1065053462953074e-07,
      "learning_rate": 6.268013785345969e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17486816,
      "step": 30785
    },
    {
      "epoch": 540.1769911504425,
      "grad_norm": 1.355245160539198e-07,
      "learning_rate": 6.26151356681543e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17489920,
      "step": 30790
    },
    {
      "epoch": 540.2654867256637,
      "grad_norm": 2.516297570309689e-07,
      "learning_rate": 6.255016237994981e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17493120,
      "step": 30795
    },
    {
      "epoch": 540.3539823008849,
      "grad_norm": 1.3338560620468343e-06,
      "learning_rate": 6.248521799886603e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17495648,
      "step": 30800
    },
    {
      "epoch": 540.3539823008849,
      "eval_loss": 0.9451584219932556,
      "eval_runtime": 1.0696,
      "eval_samples_per_second": 23.373,
      "eval_steps_per_second": 12.154,
      "num_input_tokens_seen": 17495648,
      "step": 30800
    },
    {
      "epoch": 540.4424778761062,
      "grad_norm": 7.337231977544434e-07,
      "learning_rate": 6.242030253491798e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17498352,
      "step": 30805
    },
    {
      "epoch": 540.5309734513274,
      "grad_norm": 1.2681471162068192e-06,
      "learning_rate": 6.235541599811656e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17501120,
      "step": 30810
    },
    {
      "epoch": 540.6194690265487,
      "grad_norm": 1.2761539380790055e-07,
      "learning_rate": 6.229055839846814e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17504240,
      "step": 30815
    },
    {
      "epoch": 540.70796460177,
      "grad_norm": 6.931898951734183e-06,
      "learning_rate": 6.222572974597455e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17507136,
      "step": 30820
    },
    {
      "epoch": 540.7964601769911,
      "grad_norm": 4.4901310047862353e-07,
      "learning_rate": 6.216093005063306e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17509952,
      "step": 30825
    },
    {
      "epoch": 540.8849557522124,
      "grad_norm": 1.5932317865008372e-06,
      "learning_rate": 6.209615932243678e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17512608,
      "step": 30830
    },
    {
      "epoch": 540.9734513274336,
      "grad_norm": 3.3228512620553374e-05,
      "learning_rate": 6.203141757137399e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17515712,
      "step": 30835
    },
    {
      "epoch": 541.0530973451328,
      "grad_norm": 2.3174577563622734e-06,
      "learning_rate": 6.196670480742886e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17518920,
      "step": 30840
    },
    {
      "epoch": 541.141592920354,
      "grad_norm": 9.328845607115e-08,
      "learning_rate": 6.190202104058074e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17521688,
      "step": 30845
    },
    {
      "epoch": 541.2300884955753,
      "grad_norm": 5.83108601404092e-07,
      "learning_rate": 6.183736628080475e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17524584,
      "step": 30850
    },
    {
      "epoch": 541.3185840707964,
      "grad_norm": 8.493727818859043e-07,
      "learning_rate": 6.177274053807155e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17527096,
      "step": 30855
    },
    {
      "epoch": 541.4070796460177,
      "grad_norm": 4.016962122932455e-07,
      "learning_rate": 6.170814382234713e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17530072,
      "step": 30860
    },
    {
      "epoch": 541.4955752212389,
      "grad_norm": 1.10046130430419e-06,
      "learning_rate": 6.16435761435932e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17532920,
      "step": 30865
    },
    {
      "epoch": 541.5840707964602,
      "grad_norm": 2.289651206410781e-07,
      "learning_rate": 6.157903751176681e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17535288,
      "step": 30870
    },
    {
      "epoch": 541.6725663716815,
      "grad_norm": 1.0391906357654079e-07,
      "learning_rate": 6.151452793682066e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17538088,
      "step": 30875
    },
    {
      "epoch": 541.7610619469026,
      "grad_norm": 4.734177707632625e-07,
      "learning_rate": 6.145004742870305e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17541112,
      "step": 30880
    },
    {
      "epoch": 541.8495575221239,
      "grad_norm": 2.9727428341175255e-07,
      "learning_rate": 6.138559599735752e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17543800,
      "step": 30885
    },
    {
      "epoch": 541.9380530973451,
      "grad_norm": 5.112158873998851e-07,
      "learning_rate": 6.132117365272344e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17546808,
      "step": 30890
    },
    {
      "epoch": 542.0176991150443,
      "grad_norm": 7.695509651739485e-08,
      "learning_rate": 6.125678040473545e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17549200,
      "step": 30895
    },
    {
      "epoch": 542.1061946902655,
      "grad_norm": 2.1388943594047305e-07,
      "learning_rate": 6.1192416263323755e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17551856,
      "step": 30900
    },
    {
      "epoch": 542.1946902654868,
      "grad_norm": 2.8305754540269845e-07,
      "learning_rate": 6.112808123841424e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17554480,
      "step": 30905
    },
    {
      "epoch": 542.2831858407079,
      "grad_norm": 1.186347162729362e-06,
      "learning_rate": 6.106377533992805e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17556944,
      "step": 30910
    },
    {
      "epoch": 542.3716814159292,
      "grad_norm": 1.3503675688752992e-07,
      "learning_rate": 6.099949857778204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17559408,
      "step": 30915
    },
    {
      "epoch": 542.4601769911504,
      "grad_norm": 9.667932090451359e-07,
      "learning_rate": 6.093525096188852e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17562704,
      "step": 30920
    },
    {
      "epoch": 542.5486725663717,
      "grad_norm": 1.4957528549075505e-07,
      "learning_rate": 6.087103250215518e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17566128,
      "step": 30925
    },
    {
      "epoch": 542.637168141593,
      "grad_norm": 5.341211704035231e-07,
      "learning_rate": 6.080684320848537e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17569056,
      "step": 30930
    },
    {
      "epoch": 542.7256637168142,
      "grad_norm": 4.594941401592223e-06,
      "learning_rate": 6.074268309077794e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17571920,
      "step": 30935
    },
    {
      "epoch": 542.8141592920354,
      "grad_norm": 5.216189720158582e-07,
      "learning_rate": 6.067855215892709e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17574656,
      "step": 30940
    },
    {
      "epoch": 542.9026548672566,
      "grad_norm": 1.0103519798576599e-07,
      "learning_rate": 6.061445042282271e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17577936,
      "step": 30945
    },
    {
      "epoch": 542.9911504424779,
      "grad_norm": 4.3994759835186414e-07,
      "learning_rate": 6.055037789234999e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17580992,
      "step": 30950
    },
    {
      "epoch": 543.070796460177,
      "grad_norm": 8.445489925179572e-07,
      "learning_rate": 6.048633457738975e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17583632,
      "step": 30955
    },
    {
      "epoch": 543.1592920353983,
      "grad_norm": 2.7277685148874298e-05,
      "learning_rate": 6.042232048781837e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17586720,
      "step": 30960
    },
    {
      "epoch": 543.2477876106195,
      "grad_norm": 3.9844670141064853e-07,
      "learning_rate": 6.035833563350757e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17589520,
      "step": 30965
    },
    {
      "epoch": 543.3362831858407,
      "grad_norm": 1.0875073712668382e-06,
      "learning_rate": 6.0294380024324525e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17592240,
      "step": 30970
    },
    {
      "epoch": 543.4247787610619,
      "grad_norm": 8.70275471243076e-07,
      "learning_rate": 6.023045367013213e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17595472,
      "step": 30975
    },
    {
      "epoch": 543.5132743362832,
      "grad_norm": 1.437910555068811e-07,
      "learning_rate": 6.016655658078851e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17598656,
      "step": 30980
    },
    {
      "epoch": 543.6017699115044,
      "grad_norm": 7.224269893413293e-07,
      "learning_rate": 6.010268876614753e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17601600,
      "step": 30985
    },
    {
      "epoch": 543.6902654867257,
      "grad_norm": 1.4361480680236127e-06,
      "learning_rate": 6.0038850236058266e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17604112,
      "step": 30990
    },
    {
      "epoch": 543.7787610619469,
      "grad_norm": 8.729496130399639e-07,
      "learning_rate": 5.997504100036549e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17606848,
      "step": 30995
    },
    {
      "epoch": 543.8672566371681,
      "grad_norm": 4.837479536945466e-07,
      "learning_rate": 5.991126106890949e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17609616,
      "step": 31000
    },
    {
      "epoch": 543.8672566371681,
      "eval_loss": 0.9112905263900757,
      "eval_runtime": 1.0644,
      "eval_samples_per_second": 23.487,
      "eval_steps_per_second": 12.213,
      "num_input_tokens_seen": 17609616,
      "step": 31000
    },
    {
      "epoch": 543.9557522123894,
      "grad_norm": 4.0536602341489925e-07,
      "learning_rate": 5.984751045152576e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17612352,
      "step": 31005
    },
    {
      "epoch": 544.0353982300885,
      "grad_norm": 4.4635820017902006e-07,
      "learning_rate": 5.978378915804553e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17614896,
      "step": 31010
    },
    {
      "epoch": 544.1238938053098,
      "grad_norm": 1.6883035414139158e-07,
      "learning_rate": 5.972009719829547e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17617664,
      "step": 31015
    },
    {
      "epoch": 544.212389380531,
      "grad_norm": 3.3365413401043043e-06,
      "learning_rate": 5.965643458209755e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17620752,
      "step": 31020
    },
    {
      "epoch": 544.3008849557522,
      "grad_norm": 8.93589822226204e-05,
      "learning_rate": 5.95928013192695e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17623200,
      "step": 31025
    },
    {
      "epoch": 544.3893805309734,
      "grad_norm": 4.375417574919993e-07,
      "learning_rate": 5.952919741962423e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17625744,
      "step": 31030
    },
    {
      "epoch": 544.4778761061947,
      "grad_norm": 4.5822969241271494e-07,
      "learning_rate": 5.946562289297042e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17628736,
      "step": 31035
    },
    {
      "epoch": 544.566371681416,
      "grad_norm": 1.582354229867633e-07,
      "learning_rate": 5.9402077749111855e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17631328,
      "step": 31040
    },
    {
      "epoch": 544.6548672566372,
      "grad_norm": 2.9389440214799833e-07,
      "learning_rate": 5.933856199784821e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17634928,
      "step": 31045
    },
    {
      "epoch": 544.7433628318585,
      "grad_norm": 3.32429408445023e-07,
      "learning_rate": 5.927507564897419e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17637808,
      "step": 31050
    },
    {
      "epoch": 544.8318584070796,
      "grad_norm": 1.1134012112279379e-07,
      "learning_rate": 5.9211618712280395e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17640384,
      "step": 31055
    },
    {
      "epoch": 544.9203539823009,
      "grad_norm": 2.1347830170270754e-06,
      "learning_rate": 5.914819119755255e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17643376,
      "step": 31060
    },
    {
      "epoch": 545.0,
      "grad_norm": 8.983281674090904e-08,
      "learning_rate": 5.908479311457205e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17646072,
      "step": 31065
    },
    {
      "epoch": 545.0884955752213,
      "grad_norm": 2.171768983316724e-06,
      "learning_rate": 5.902142447311559e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17648888,
      "step": 31070
    },
    {
      "epoch": 545.1769911504425,
      "grad_norm": 8.651831961969947e-08,
      "learning_rate": 5.895808528295546e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17651992,
      "step": 31075
    },
    {
      "epoch": 545.2654867256637,
      "grad_norm": 3.721605992268451e-07,
      "learning_rate": 5.889477555385941e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17654792,
      "step": 31080
    },
    {
      "epoch": 545.3539823008849,
      "grad_norm": 4.826962140214164e-06,
      "learning_rate": 5.883149529559051e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17657592,
      "step": 31085
    },
    {
      "epoch": 545.4424778761062,
      "grad_norm": 1.7494613757662592e-07,
      "learning_rate": 5.876824451790738e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17660344,
      "step": 31090
    },
    {
      "epoch": 545.5309734513274,
      "grad_norm": 4.635094228433445e-07,
      "learning_rate": 5.87050232305642e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17663032,
      "step": 31095
    },
    {
      "epoch": 545.6194690265487,
      "grad_norm": 1.1710598357694835e-07,
      "learning_rate": 5.864183144331034e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17666024,
      "step": 31100
    },
    {
      "epoch": 545.70796460177,
      "grad_norm": 1.423275762135745e-06,
      "learning_rate": 5.857866916589089e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17669080,
      "step": 31105
    },
    {
      "epoch": 545.7964601769911,
      "grad_norm": 1.4061716058222373e-07,
      "learning_rate": 5.8515536408046216e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17672040,
      "step": 31110
    },
    {
      "epoch": 545.8849557522124,
      "grad_norm": 9.446276294511335e-07,
      "learning_rate": 5.845243317951208e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17675288,
      "step": 31115
    },
    {
      "epoch": 545.9734513274336,
      "grad_norm": 5.160122213965224e-07,
      "learning_rate": 5.838935949001997e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17678200,
      "step": 31120
    },
    {
      "epoch": 546.0530973451328,
      "grad_norm": 3.918212598819082e-07,
      "learning_rate": 5.8326315349296476e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17680416,
      "step": 31125
    },
    {
      "epoch": 546.141592920354,
      "grad_norm": 2.4770858431111265e-07,
      "learning_rate": 5.826330076706396e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17683328,
      "step": 31130
    },
    {
      "epoch": 546.2300884955753,
      "grad_norm": 8.377255289815366e-05,
      "learning_rate": 5.820031575303988e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17686576,
      "step": 31135
    },
    {
      "epoch": 546.3185840707964,
      "grad_norm": 5.180533548809763e-07,
      "learning_rate": 5.813736031693745e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17689184,
      "step": 31140
    },
    {
      "epoch": 546.4070796460177,
      "grad_norm": 1.514604548447096e-07,
      "learning_rate": 5.807443446846522e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17691776,
      "step": 31145
    },
    {
      "epoch": 546.4955752212389,
      "grad_norm": 8.850830681694788e-07,
      "learning_rate": 5.801153821732699e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17694752,
      "step": 31150
    },
    {
      "epoch": 546.5840707964602,
      "grad_norm": 2.4515003360647825e-07,
      "learning_rate": 5.794867157322229e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17697488,
      "step": 31155
    },
    {
      "epoch": 546.6725663716815,
      "grad_norm": 4.666698032451677e-07,
      "learning_rate": 5.788583454584593e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17700560,
      "step": 31160
    },
    {
      "epoch": 546.7610619469026,
      "grad_norm": 7.428754429383844e-07,
      "learning_rate": 5.7823027144888075e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17703584,
      "step": 31165
    },
    {
      "epoch": 546.8495575221239,
      "grad_norm": 4.6480809601234796e-07,
      "learning_rate": 5.776024938003455e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17706320,
      "step": 31170
    },
    {
      "epoch": 546.9380530973451,
      "grad_norm": 8.771768875703856e-07,
      "learning_rate": 5.7697501260966345e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17709136,
      "step": 31175
    },
    {
      "epoch": 547.0176991150443,
      "grad_norm": 7.157048571571067e-07,
      "learning_rate": 5.7634782797360145e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17712368,
      "step": 31180
    },
    {
      "epoch": 547.1061946902655,
      "grad_norm": 2.3171683949385624e-07,
      "learning_rate": 5.757209399888777e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17715392,
      "step": 31185
    },
    {
      "epoch": 547.1946902654868,
      "grad_norm": 9.828711711179494e-08,
      "learning_rate": 5.750943487521679e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17718640,
      "step": 31190
    },
    {
      "epoch": 547.2831858407079,
      "grad_norm": 4.4477633309725206e-07,
      "learning_rate": 5.744680543600986e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17721200,
      "step": 31195
    },
    {
      "epoch": 547.3716814159292,
      "grad_norm": 2.1881753298202966e-07,
      "learning_rate": 5.738420569092537e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17723600,
      "step": 31200
    },
    {
      "epoch": 547.3716814159292,
      "eval_loss": 0.9103177785873413,
      "eval_runtime": 1.08,
      "eval_samples_per_second": 23.149,
      "eval_steps_per_second": 12.038,
      "num_input_tokens_seen": 17723600,
      "step": 31200
    },
    {
      "epoch": 547.4601769911504,
      "grad_norm": 4.795837185156415e-07,
      "learning_rate": 5.732163564961684e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17726800,
      "step": 31205
    },
    {
      "epoch": 547.5486725663717,
      "grad_norm": 2.839052513081697e-07,
      "learning_rate": 5.725909532173354e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17729728,
      "step": 31210
    },
    {
      "epoch": 547.637168141593,
      "grad_norm": 3.0892115887581895e-07,
      "learning_rate": 5.719658471691977e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17732624,
      "step": 31215
    },
    {
      "epoch": 547.7256637168142,
      "grad_norm": 3.714760623552138e-07,
      "learning_rate": 5.71341038448156e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17735488,
      "step": 31220
    },
    {
      "epoch": 547.8141592920354,
      "grad_norm": 9.475194673314036e-08,
      "learning_rate": 5.707165271505635e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17737824,
      "step": 31225
    },
    {
      "epoch": 547.9026548672566,
      "grad_norm": 6.870089350741182e-07,
      "learning_rate": 5.700923133727271e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17740432,
      "step": 31230
    },
    {
      "epoch": 547.9911504424779,
      "grad_norm": 1.7465795565385633e-07,
      "learning_rate": 5.694683972109083e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17743584,
      "step": 31235
    },
    {
      "epoch": 548.070796460177,
      "grad_norm": 1.1196195373486262e-05,
      "learning_rate": 5.688447787613241e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17746264,
      "step": 31240
    },
    {
      "epoch": 548.1592920353983,
      "grad_norm": 2.31560920838092e-06,
      "learning_rate": 5.6822145812014285e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17748776,
      "step": 31245
    },
    {
      "epoch": 548.2477876106195,
      "grad_norm": 1.9534851958269428e-07,
      "learning_rate": 5.675984353834896e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17751256,
      "step": 31250
    },
    {
      "epoch": 548.3362831858407,
      "grad_norm": 1.1485666391308769e-06,
      "learning_rate": 5.66975710647441e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17754136,
      "step": 31255
    },
    {
      "epoch": 548.4247787610619,
      "grad_norm": 5.753890377491189e-07,
      "learning_rate": 5.663532840080304e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17757288,
      "step": 31260
    },
    {
      "epoch": 548.5132743362832,
      "grad_norm": 3.4828079265025735e-07,
      "learning_rate": 5.6573115556124325e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17760312,
      "step": 31265
    },
    {
      "epoch": 548.6017699115044,
      "grad_norm": 4.613324563251808e-06,
      "learning_rate": 5.651093254030185e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17762920,
      "step": 31270
    },
    {
      "epoch": 548.6902654867257,
      "grad_norm": 1.6819384995869768e-07,
      "learning_rate": 5.644877936292514e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17765608,
      "step": 31275
    },
    {
      "epoch": 548.7787610619469,
      "grad_norm": 1.1901488505827729e-07,
      "learning_rate": 5.638665603357901e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17768024,
      "step": 31280
    },
    {
      "epoch": 548.8672566371681,
      "grad_norm": 2.0738987416279997e-07,
      "learning_rate": 5.632456256184357e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17771368,
      "step": 31285
    },
    {
      "epoch": 548.9557522123894,
      "grad_norm": 2.1969211161376734e-07,
      "learning_rate": 5.626249895729452e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17774648,
      "step": 31290
    },
    {
      "epoch": 549.0353982300885,
      "grad_norm": 1.0480787295819027e-06,
      "learning_rate": 5.620046522950273e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17776816,
      "step": 31295
    },
    {
      "epoch": 549.1238938053098,
      "grad_norm": 2.3678178706632025e-07,
      "learning_rate": 5.613846138803464e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17779472,
      "step": 31300
    },
    {
      "epoch": 549.212389380531,
      "grad_norm": 2.055925705235495e-07,
      "learning_rate": 5.607648744245206e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17782336,
      "step": 31305
    },
    {
      "epoch": 549.3008849557522,
      "grad_norm": 1.1415554723726018e-07,
      "learning_rate": 5.601454340231207e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17784976,
      "step": 31310
    },
    {
      "epoch": 549.3893805309734,
      "grad_norm": 3.590561163946404e-07,
      "learning_rate": 5.595262927716724e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17787536,
      "step": 31315
    },
    {
      "epoch": 549.4778761061947,
      "grad_norm": 6.91046295742126e-07,
      "learning_rate": 5.589074507656561e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17790512,
      "step": 31320
    },
    {
      "epoch": 549.566371681416,
      "grad_norm": 8.078690143520362e-07,
      "learning_rate": 5.582889081005044e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17793200,
      "step": 31325
    },
    {
      "epoch": 549.6548672566372,
      "grad_norm": 6.692178430967033e-05,
      "learning_rate": 5.5767066487160316e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17796080,
      "step": 31330
    },
    {
      "epoch": 549.7433628318585,
      "grad_norm": 5.891910745958739e-07,
      "learning_rate": 5.570527211742949e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17798960,
      "step": 31335
    },
    {
      "epoch": 549.8318584070796,
      "grad_norm": 8.629925218883727e-08,
      "learning_rate": 5.564350771038731e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17801680,
      "step": 31340
    },
    {
      "epoch": 549.9203539823009,
      "grad_norm": 1.223024099772374e-07,
      "learning_rate": 5.558177327555875e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17805264,
      "step": 31345
    },
    {
      "epoch": 550.0,
      "grad_norm": 8.084730325208511e-07,
      "learning_rate": 5.552006882246388e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17807696,
      "step": 31350
    },
    {
      "epoch": 550.0884955752213,
      "grad_norm": 2.5354614763273275e-07,
      "learning_rate": 5.545839436061839e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17810464,
      "step": 31355
    },
    {
      "epoch": 550.1769911504425,
      "grad_norm": 2.7572525596042397e-07,
      "learning_rate": 5.539674989953331e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17813504,
      "step": 31360
    },
    {
      "epoch": 550.2654867256637,
      "grad_norm": 4.047718960009661e-07,
      "learning_rate": 5.533513544871488e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17816384,
      "step": 31365
    },
    {
      "epoch": 550.3539823008849,
      "grad_norm": 7.172785990405828e-05,
      "learning_rate": 5.527355101766493e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17819248,
      "step": 31370
    },
    {
      "epoch": 550.4424778761062,
      "grad_norm": 4.076464108493383e-07,
      "learning_rate": 5.521199661588044e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17822096,
      "step": 31375
    },
    {
      "epoch": 550.5309734513274,
      "grad_norm": 4.4224432826922566e-07,
      "learning_rate": 5.5150472252853944e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17825216,
      "step": 31380
    },
    {
      "epoch": 550.6194690265487,
      "grad_norm": 2.2804273669407849e-07,
      "learning_rate": 5.50889779380733e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17827856,
      "step": 31385
    },
    {
      "epoch": 550.70796460177,
      "grad_norm": 1.1618672033364419e-05,
      "learning_rate": 5.5027513681021605e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17830752,
      "step": 31390
    },
    {
      "epoch": 550.7964601769911,
      "grad_norm": 2.029921404300694e-07,
      "learning_rate": 5.4966079491177545e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17833584,
      "step": 31395
    },
    {
      "epoch": 550.8849557522124,
      "grad_norm": 3.21012294079992e-07,
      "learning_rate": 5.490467537801491e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17836576,
      "step": 31400
    },
    {
      "epoch": 550.8849557522124,
      "eval_loss": 0.8986397981643677,
      "eval_runtime": 1.0681,
      "eval_samples_per_second": 23.406,
      "eval_steps_per_second": 12.171,
      "num_input_tokens_seen": 17836576,
      "step": 31400
    },
    {
      "epoch": 550.9734513274336,
      "grad_norm": 6.006653734402789e-07,
      "learning_rate": 5.484330135100313e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17839360,
      "step": 31405
    },
    {
      "epoch": 551.0530973451328,
      "grad_norm": 1.534763896415825e-06,
      "learning_rate": 5.4781957419606785e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17841592,
      "step": 31410
    },
    {
      "epoch": 551.141592920354,
      "grad_norm": 5.6751218835415784e-06,
      "learning_rate": 5.472064359328577e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17844552,
      "step": 31415
    },
    {
      "epoch": 551.2300884955753,
      "grad_norm": 1.5572527445328888e-07,
      "learning_rate": 5.4659359881495565e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17847512,
      "step": 31420
    },
    {
      "epoch": 551.3185840707964,
      "grad_norm": 4.1236404513256275e-07,
      "learning_rate": 5.4598106293686916e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17850616,
      "step": 31425
    },
    {
      "epoch": 551.4070796460177,
      "grad_norm": 2.909201839429443e-07,
      "learning_rate": 5.45368828393058e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17853432,
      "step": 31430
    },
    {
      "epoch": 551.4955752212389,
      "grad_norm": 1.111987444346596e-07,
      "learning_rate": 5.44756895277937e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17856856,
      "step": 31435
    },
    {
      "epoch": 551.5840707964602,
      "grad_norm": 2.0085462892893702e-07,
      "learning_rate": 5.441452636858746e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17860120,
      "step": 31440
    },
    {
      "epoch": 551.6725663716815,
      "grad_norm": 5.300209977576742e-07,
      "learning_rate": 5.435339337111905e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17862840,
      "step": 31445
    },
    {
      "epoch": 551.7610619469026,
      "grad_norm": 1.1071161907239002e-06,
      "learning_rate": 5.42922905448161e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17865560,
      "step": 31450
    },
    {
      "epoch": 551.8495575221239,
      "grad_norm": 5.061963292973815e-07,
      "learning_rate": 5.423121789910129e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17868056,
      "step": 31455
    },
    {
      "epoch": 551.9380530973451,
      "grad_norm": 3.070182401643251e-07,
      "learning_rate": 5.417017544339287e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17870984,
      "step": 31460
    },
    {
      "epoch": 552.0176991150443,
      "grad_norm": 5.227651627137675e-07,
      "learning_rate": 5.410916318710443e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17873320,
      "step": 31465
    },
    {
      "epoch": 552.1061946902655,
      "grad_norm": 8.340625186065154e-07,
      "learning_rate": 5.404818113964466e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17876056,
      "step": 31470
    },
    {
      "epoch": 552.1946902654868,
      "grad_norm": 9.056664680429094e-07,
      "learning_rate": 5.398722931041792e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17879000,
      "step": 31475
    },
    {
      "epoch": 552.2831858407079,
      "grad_norm": 1.0708475173260013e-07,
      "learning_rate": 5.392630770882367e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17881992,
      "step": 31480
    },
    {
      "epoch": 552.3716814159292,
      "grad_norm": 3.7442848110913474e-07,
      "learning_rate": 5.3865416344256705e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17884488,
      "step": 31485
    },
    {
      "epoch": 552.4601769911504,
      "grad_norm": 1.6053226659096254e-07,
      "learning_rate": 5.380455522610742e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17887144,
      "step": 31490
    },
    {
      "epoch": 552.5486725663717,
      "grad_norm": 3.5860034586221445e-07,
      "learning_rate": 5.374372436376116e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17889704,
      "step": 31495
    },
    {
      "epoch": 552.637168141593,
      "grad_norm": 2.9796314038321725e-07,
      "learning_rate": 5.368292376659895e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17892952,
      "step": 31500
    },
    {
      "epoch": 552.7256637168142,
      "grad_norm": 5.97061386997666e-07,
      "learning_rate": 5.362215344399701e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17895880,
      "step": 31505
    },
    {
      "epoch": 552.8141592920354,
      "grad_norm": 2.7421870072430465e-06,
      "learning_rate": 5.356141340532678e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17898680,
      "step": 31510
    },
    {
      "epoch": 552.9026548672566,
      "grad_norm": 4.849426886721631e-07,
      "learning_rate": 5.350070365995522e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17902360,
      "step": 31515
    },
    {
      "epoch": 552.9911504424779,
      "grad_norm": 4.988560817764665e-07,
      "learning_rate": 5.344002421724459e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17904840,
      "step": 31520
    },
    {
      "epoch": 553.070796460177,
      "grad_norm": 3.1897531016511493e-07,
      "learning_rate": 5.337937508655228e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17907416,
      "step": 31525
    },
    {
      "epoch": 553.1592920353983,
      "grad_norm": 6.030947474755521e-07,
      "learning_rate": 5.331875627723126e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17910024,
      "step": 31530
    },
    {
      "epoch": 553.2477876106195,
      "grad_norm": 4.094029009138467e-07,
      "learning_rate": 5.325816779862963e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17913384,
      "step": 31535
    },
    {
      "epoch": 553.3362831858407,
      "grad_norm": 3.3274048405473877e-07,
      "learning_rate": 5.319760966009102e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17916520,
      "step": 31540
    },
    {
      "epoch": 553.4247787610619,
      "grad_norm": 9.390781201545906e-07,
      "learning_rate": 5.3137081870954096e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17919768,
      "step": 31545
    },
    {
      "epoch": 553.5132743362832,
      "grad_norm": 2.1312843045961927e-07,
      "learning_rate": 5.307658444055313e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17922728,
      "step": 31550
    },
    {
      "epoch": 553.6017699115044,
      "grad_norm": 9.890118235489354e-08,
      "learning_rate": 5.301611737821749e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17925240,
      "step": 31555
    },
    {
      "epoch": 553.6902654867257,
      "grad_norm": 1.564485046401387e-06,
      "learning_rate": 5.295568069327206e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17928088,
      "step": 31560
    },
    {
      "epoch": 553.7787610619469,
      "grad_norm": 9.96281528387044e-07,
      "learning_rate": 5.289527439503683e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17930728,
      "step": 31565
    },
    {
      "epoch": 553.8672566371681,
      "grad_norm": 9.584946383256465e-07,
      "learning_rate": 5.28348984928273e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17933432,
      "step": 31570
    },
    {
      "epoch": 553.9557522123894,
      "grad_norm": 2.0540454670481267e-07,
      "learning_rate": 5.27745529959541e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17936280,
      "step": 31575
    },
    {
      "epoch": 554.0353982300885,
      "grad_norm": 1.3442871704683057e-07,
      "learning_rate": 5.271423791372335e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17938712,
      "step": 31580
    },
    {
      "epoch": 554.1238938053098,
      "grad_norm": 1.2367522685963195e-06,
      "learning_rate": 5.26539532554364e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17941640,
      "step": 31585
    },
    {
      "epoch": 554.212389380531,
      "grad_norm": 1.8277169999691978e-07,
      "learning_rate": 5.25936990303898e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17944392,
      "step": 31590
    },
    {
      "epoch": 554.3008849557522,
      "grad_norm": 4.955993517796742e-07,
      "learning_rate": 5.253347524787555e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17947416,
      "step": 31595
    },
    {
      "epoch": 554.3893805309734,
      "grad_norm": 4.1672581119200913e-07,
      "learning_rate": 5.2473281917181035e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17949928,
      "step": 31600
    },
    {
      "epoch": 554.3893805309734,
      "eval_loss": 0.8947702646255493,
      "eval_runtime": 1.0775,
      "eval_samples_per_second": 23.201,
      "eval_steps_per_second": 12.065,
      "num_input_tokens_seen": 17949928,
      "step": 31600
    },
    {
      "epoch": 554.4778761061947,
      "grad_norm": 1.8788969669003563e-07,
      "learning_rate": 5.241311904758864e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17952648,
      "step": 31605
    },
    {
      "epoch": 554.566371681416,
      "grad_norm": 7.826726005077944e-07,
      "learning_rate": 5.23529866483764e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17955656,
      "step": 31610
    },
    {
      "epoch": 554.6548672566372,
      "grad_norm": 9.340217133058104e-08,
      "learning_rate": 5.229288472881732e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17958504,
      "step": 31615
    },
    {
      "epoch": 554.7433628318585,
      "grad_norm": 1.7169456612009526e-07,
      "learning_rate": 5.2232813298180025e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17961272,
      "step": 31620
    },
    {
      "epoch": 554.8318584070796,
      "grad_norm": 1.3054922192168306e-06,
      "learning_rate": 5.217277236572824e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17964104,
      "step": 31625
    },
    {
      "epoch": 554.9203539823009,
      "grad_norm": 2.2114519993010617e-07,
      "learning_rate": 5.211276194072093e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17967416,
      "step": 31630
    },
    {
      "epoch": 555.0,
      "grad_norm": 9.536659462128227e-08,
      "learning_rate": 5.205278203241254e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17969928,
      "step": 31635
    },
    {
      "epoch": 555.0884955752213,
      "grad_norm": 1.0199668167842901e-06,
      "learning_rate": 5.199283265005278e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17973176,
      "step": 31640
    },
    {
      "epoch": 555.1769911504425,
      "grad_norm": 3.1732070056023076e-06,
      "learning_rate": 5.193291380288648e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17976056,
      "step": 31645
    },
    {
      "epoch": 555.2654867256637,
      "grad_norm": 4.233666857089702e-07,
      "learning_rate": 5.1873025500153995e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17979336,
      "step": 31650
    },
    {
      "epoch": 555.3539823008849,
      "grad_norm": 1.670613443138791e-07,
      "learning_rate": 5.181316775109071e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17982168,
      "step": 31655
    },
    {
      "epoch": 555.4424778761062,
      "grad_norm": 5.873055783922609e-07,
      "learning_rate": 5.1753340564927564e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17985144,
      "step": 31660
    },
    {
      "epoch": 555.5309734513274,
      "grad_norm": 2.893857242725062e-07,
      "learning_rate": 5.169354395089068e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17988200,
      "step": 31665
    },
    {
      "epoch": 555.6194690265487,
      "grad_norm": 1.1307564307116991e-07,
      "learning_rate": 5.1633777918201346e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17990808,
      "step": 31670
    },
    {
      "epoch": 555.70796460177,
      "grad_norm": 3.274936375419202e-07,
      "learning_rate": 5.157404247607625e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17993320,
      "step": 31675
    },
    {
      "epoch": 555.7964601769911,
      "grad_norm": 1.8171505189457093e-07,
      "learning_rate": 5.1514337633727454e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17995992,
      "step": 31680
    },
    {
      "epoch": 555.8849557522124,
      "grad_norm": 2.6075150572069106e-07,
      "learning_rate": 5.145466340036206e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 17998856,
      "step": 31685
    },
    {
      "epoch": 555.9734513274336,
      "grad_norm": 4.044049546791939e-06,
      "learning_rate": 5.139501978518274e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18001784,
      "step": 31690
    },
    {
      "epoch": 556.0530973451328,
      "grad_norm": 3.1769053521202295e-07,
      "learning_rate": 5.133540679738716e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18003784,
      "step": 31695
    },
    {
      "epoch": 556.141592920354,
      "grad_norm": 1.4756873270016513e-06,
      "learning_rate": 5.127582444616838e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18007000,
      "step": 31700
    },
    {
      "epoch": 556.2300884955753,
      "grad_norm": 2.560168752552272e-07,
      "learning_rate": 5.121627274071486e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18009592,
      "step": 31705
    },
    {
      "epoch": 556.3185840707964,
      "grad_norm": 1.5112143501028186e-06,
      "learning_rate": 5.115675169021009e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18012728,
      "step": 31710
    },
    {
      "epoch": 556.4070796460177,
      "grad_norm": 3.389068297110498e-07,
      "learning_rate": 5.1097261303832994e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18015144,
      "step": 31715
    },
    {
      "epoch": 556.4955752212389,
      "grad_norm": 2.536945657993783e-07,
      "learning_rate": 5.103780159075788e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18018264,
      "step": 31720
    },
    {
      "epoch": 556.5840707964602,
      "grad_norm": 3.798239163188555e-07,
      "learning_rate": 5.0978372560154e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18020776,
      "step": 31725
    },
    {
      "epoch": 556.6725663716815,
      "grad_norm": 6.576405553460063e-08,
      "learning_rate": 5.091897422118619e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18024152,
      "step": 31730
    },
    {
      "epoch": 556.7610619469026,
      "grad_norm": 2.8319496436779446e-07,
      "learning_rate": 5.0859606583014305e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18026984,
      "step": 31735
    },
    {
      "epoch": 556.8495575221239,
      "grad_norm": 3.026150443474762e-05,
      "learning_rate": 5.080026965479365e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18029896,
      "step": 31740
    },
    {
      "epoch": 556.9380530973451,
      "grad_norm": 1.9332279066475166e-07,
      "learning_rate": 5.074096344567475e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18033000,
      "step": 31745
    },
    {
      "epoch": 557.0176991150443,
      "grad_norm": 3.6480892617873906e-07,
      "learning_rate": 5.0681687964803294e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18035296,
      "step": 31750
    },
    {
      "epoch": 557.1061946902655,
      "grad_norm": 8.41700824594227e-08,
      "learning_rate": 5.06224432213204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18038320,
      "step": 31755
    },
    {
      "epoch": 557.1946902654868,
      "grad_norm": 1.384458556685786e-07,
      "learning_rate": 5.056322922436224e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18041040,
      "step": 31760
    },
    {
      "epoch": 557.2831858407079,
      "grad_norm": 2.3183238795354555e-07,
      "learning_rate": 5.0504045983060465e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18044080,
      "step": 31765
    },
    {
      "epoch": 557.3716814159292,
      "grad_norm": 4.6397067308134865e-06,
      "learning_rate": 5.044489350654183e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18046656,
      "step": 31770
    },
    {
      "epoch": 557.4601769911504,
      "grad_norm": 4.2341565631431877e-07,
      "learning_rate": 5.038577180392831e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18049968,
      "step": 31775
    },
    {
      "epoch": 557.5486725663717,
      "grad_norm": 1.6529659774278116e-07,
      "learning_rate": 5.032668088433729e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18052640,
      "step": 31780
    },
    {
      "epoch": 557.637168141593,
      "grad_norm": 1.327048039456713e-06,
      "learning_rate": 5.02676207568814e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18055680,
      "step": 31785
    },
    {
      "epoch": 557.7256637168142,
      "grad_norm": 3.7250671880428854e-07,
      "learning_rate": 5.02085914306683e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18058944,
      "step": 31790
    },
    {
      "epoch": 557.8141592920354,
      "grad_norm": 7.773234642627358e-07,
      "learning_rate": 5.014959291480123e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18062080,
      "step": 31795
    },
    {
      "epoch": 557.9026548672566,
      "grad_norm": 2.1404724748208537e-07,
      "learning_rate": 5.009062521837835e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18064576,
      "step": 31800
    },
    {
      "epoch": 557.9026548672566,
      "eval_loss": 0.9036370515823364,
      "eval_runtime": 1.0706,
      "eval_samples_per_second": 23.351,
      "eval_steps_per_second": 12.143,
      "num_input_tokens_seen": 18064576,
      "step": 31800
    },
    {
      "epoch": 557.9911504424779,
      "grad_norm": 1.1661370535875903e-07,
      "learning_rate": 5.003168835049324e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18067184,
      "step": 31805
    },
    {
      "epoch": 558.070796460177,
      "grad_norm": 4.2359025087534974e-07,
      "learning_rate": 4.997278232023483e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18069656,
      "step": 31810
    },
    {
      "epoch": 558.1592920353983,
      "grad_norm": 1.557550319830625e-07,
      "learning_rate": 4.9913907136687036e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18072440,
      "step": 31815
    },
    {
      "epoch": 558.2477876106195,
      "grad_norm": 1.4580277252207452e-07,
      "learning_rate": 4.985506280892918e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18075960,
      "step": 31820
    },
    {
      "epoch": 558.3362831858407,
      "grad_norm": 1.5947397002946673e-07,
      "learning_rate": 4.979624934603589e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18079208,
      "step": 31825
    },
    {
      "epoch": 558.4247787610619,
      "grad_norm": 1.8146772617910756e-06,
      "learning_rate": 4.97374667570768e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18082024,
      "step": 31830
    },
    {
      "epoch": 558.5132743362832,
      "grad_norm": 5.757175358667155e-07,
      "learning_rate": 4.967871505111704e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18084456,
      "step": 31835
    },
    {
      "epoch": 558.6017699115044,
      "grad_norm": 2.9637576517416164e-05,
      "learning_rate": 4.961999423721686e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18087352,
      "step": 31840
    },
    {
      "epoch": 558.6902654867257,
      "grad_norm": 4.807002369489055e-07,
      "learning_rate": 4.956130432443159e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18090072,
      "step": 31845
    },
    {
      "epoch": 558.7787610619469,
      "grad_norm": 4.660026320379984e-07,
      "learning_rate": 4.950264532181215e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18093160,
      "step": 31850
    },
    {
      "epoch": 558.8672566371681,
      "grad_norm": 4.071911803293915e-07,
      "learning_rate": 4.944401723840433e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18095976,
      "step": 31855
    },
    {
      "epoch": 558.9557522123894,
      "grad_norm": 1.4862704347251565e-06,
      "learning_rate": 4.938542008324942e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18098824,
      "step": 31860
    },
    {
      "epoch": 559.0353982300885,
      "grad_norm": 3.525253760017222e-07,
      "learning_rate": 4.9326853865383855e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18101160,
      "step": 31865
    },
    {
      "epoch": 559.1238938053098,
      "grad_norm": 4.6730139047213015e-07,
      "learning_rate": 4.926831859383918e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18103528,
      "step": 31870
    },
    {
      "epoch": 559.212389380531,
      "grad_norm": 1.0434225714561762e-06,
      "learning_rate": 4.92098142776424e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18106248,
      "step": 31875
    },
    {
      "epoch": 559.3008849557522,
      "grad_norm": 5.201197836868232e-07,
      "learning_rate": 4.91513409258155e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18109000,
      "step": 31880
    },
    {
      "epoch": 559.3893805309734,
      "grad_norm": 5.545974204324011e-07,
      "learning_rate": 4.909289854737581e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18112120,
      "step": 31885
    },
    {
      "epoch": 559.4778761061947,
      "grad_norm": 1.4345800991577562e-07,
      "learning_rate": 4.903448715133602e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18115448,
      "step": 31890
    },
    {
      "epoch": 559.566371681416,
      "grad_norm": 3.571714728423103e-07,
      "learning_rate": 4.897610674670372e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18118536,
      "step": 31895
    },
    {
      "epoch": 559.6548672566372,
      "grad_norm": 6.389820441654592e-07,
      "learning_rate": 4.8917757342482e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18121528,
      "step": 31900
    },
    {
      "epoch": 559.7433628318585,
      "grad_norm": 2.3477505806113186e-07,
      "learning_rate": 4.885943894766909e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18124504,
      "step": 31905
    },
    {
      "epoch": 559.8318584070796,
      "grad_norm": 2.3978867602636456e-07,
      "learning_rate": 4.880115157125842e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18127512,
      "step": 31910
    },
    {
      "epoch": 559.9203539823009,
      "grad_norm": 3.219563495804323e-07,
      "learning_rate": 4.874289522223857e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18130072,
      "step": 31915
    },
    {
      "epoch": 560.0,
      "grad_norm": 1.463064336348907e-07,
      "learning_rate": 4.868466990959339e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18132392,
      "step": 31920
    },
    {
      "epoch": 560.0884955752213,
      "grad_norm": 1.4031395494384924e-06,
      "learning_rate": 4.8626475642301964e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18134904,
      "step": 31925
    },
    {
      "epoch": 560.1769911504425,
      "grad_norm": 1.7615639080759138e-07,
      "learning_rate": 4.856831242933871e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18138536,
      "step": 31930
    },
    {
      "epoch": 560.2654867256637,
      "grad_norm": 2.215562631135981e-07,
      "learning_rate": 4.851018027967294e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18141096,
      "step": 31935
    },
    {
      "epoch": 560.3539823008849,
      "grad_norm": 9.145364288087876e-07,
      "learning_rate": 4.845207920226946e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18143880,
      "step": 31940
    },
    {
      "epoch": 560.4424778761062,
      "grad_norm": 1.3054817316060507e-07,
      "learning_rate": 4.839400920608825e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18147144,
      "step": 31945
    },
    {
      "epoch": 560.5309734513274,
      "grad_norm": 3.576154767870321e-07,
      "learning_rate": 4.83359703000843e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18149880,
      "step": 31950
    },
    {
      "epoch": 560.6194690265487,
      "grad_norm": 8.105690199045057e-07,
      "learning_rate": 4.827796249320804e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18153160,
      "step": 31955
    },
    {
      "epoch": 560.70796460177,
      "grad_norm": 2.414540745121485e-07,
      "learning_rate": 4.82199857944049e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18155784,
      "step": 31960
    },
    {
      "epoch": 560.7964601769911,
      "grad_norm": 1.652938408369664e-06,
      "learning_rate": 4.8162040212615695e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18158344,
      "step": 31965
    },
    {
      "epoch": 560.8849557522124,
      "grad_norm": 1.2609650923423033e-07,
      "learning_rate": 4.810412575677639e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18161032,
      "step": 31970
    },
    {
      "epoch": 560.9734513274336,
      "grad_norm": 1.380358582991903e-07,
      "learning_rate": 4.804624243581801e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18163976,
      "step": 31975
    },
    {
      "epoch": 561.0530973451328,
      "grad_norm": 1.2853182624894544e-06,
      "learning_rate": 4.798839025866703e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18166088,
      "step": 31980
    },
    {
      "epoch": 561.141592920354,
      "grad_norm": 3.848068956813222e-07,
      "learning_rate": 4.793056923424491e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18168776,
      "step": 31985
    },
    {
      "epoch": 561.2300884955753,
      "grad_norm": 1.832476215213319e-07,
      "learning_rate": 4.78727793714683e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18171368,
      "step": 31990
    },
    {
      "epoch": 561.3185840707964,
      "grad_norm": 1.2646884215428145e-06,
      "learning_rate": 4.7815020679249285e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18174280,
      "step": 31995
    },
    {
      "epoch": 561.4070796460177,
      "grad_norm": 2.058490764511589e-07,
      "learning_rate": 4.775729316649483e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18177096,
      "step": 32000
    },
    {
      "epoch": 561.4070796460177,
      "eval_loss": 0.9058958888053894,
      "eval_runtime": 1.0628,
      "eval_samples_per_second": 23.522,
      "eval_steps_per_second": 12.232,
      "num_input_tokens_seen": 18177096,
      "step": 32000
    },
    {
      "epoch": 561.4955752212389,
      "grad_norm": 1.0760044233393273e-07,
      "learning_rate": 4.769959684210728e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18180024,
      "step": 32005
    },
    {
      "epoch": 561.5840707964602,
      "grad_norm": 2.4715404833841603e-07,
      "learning_rate": 4.764193171498426e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18182696,
      "step": 32010
    },
    {
      "epoch": 561.6725663716815,
      "grad_norm": 2.163504291274876e-07,
      "learning_rate": 4.75842977940183e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18185784,
      "step": 32015
    },
    {
      "epoch": 561.7610619469026,
      "grad_norm": 6.774319558644493e-07,
      "learning_rate": 4.752669508809729e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18188264,
      "step": 32020
    },
    {
      "epoch": 561.8495575221239,
      "grad_norm": 5.198140570428222e-06,
      "learning_rate": 4.746912360610445e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18191592,
      "step": 32025
    },
    {
      "epoch": 561.9380530973451,
      "grad_norm": 3.532100265601912e-07,
      "learning_rate": 4.741158335691781e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18194616,
      "step": 32030
    },
    {
      "epoch": 562.0176991150443,
      "grad_norm": 1.965997711295131e-07,
      "learning_rate": 4.7354074349410994e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18197592,
      "step": 32035
    },
    {
      "epoch": 562.1061946902655,
      "grad_norm": 6.354175638989545e-07,
      "learning_rate": 4.729659659245245e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18200152,
      "step": 32040
    },
    {
      "epoch": 562.1946902654868,
      "grad_norm": 5.14935891260393e-07,
      "learning_rate": 4.723915009490601e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18202856,
      "step": 32045
    },
    {
      "epoch": 562.2831858407079,
      "grad_norm": 1.488884322498052e-07,
      "learning_rate": 4.718173486563077e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18206424,
      "step": 32050
    },
    {
      "epoch": 562.3716814159292,
      "grad_norm": 2.2389944831502362e-07,
      "learning_rate": 4.71243509134808e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18208904,
      "step": 32055
    },
    {
      "epoch": 562.4601769911504,
      "grad_norm": 1.6806322378215555e-07,
      "learning_rate": 4.706699824730532e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18211496,
      "step": 32060
    },
    {
      "epoch": 562.5486725663717,
      "grad_norm": 4.405470463098027e-06,
      "learning_rate": 4.700967687594901e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18214104,
      "step": 32065
    },
    {
      "epoch": 562.637168141593,
      "grad_norm": 1.7111018735249672e-07,
      "learning_rate": 4.69523868082514e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18217016,
      "step": 32070
    },
    {
      "epoch": 562.7256637168142,
      "grad_norm": 6.31465468359238e-07,
      "learning_rate": 4.689512805304747e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18219688,
      "step": 32075
    },
    {
      "epoch": 562.8141592920354,
      "grad_norm": 5.780755145678995e-06,
      "learning_rate": 4.683790061916707e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18222680,
      "step": 32080
    },
    {
      "epoch": 562.9026548672566,
      "grad_norm": 3.049013628242392e-07,
      "learning_rate": 4.678070451543551e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18226072,
      "step": 32085
    },
    {
      "epoch": 562.9911504424779,
      "grad_norm": 1.6821581084514037e-05,
      "learning_rate": 4.6723539750673204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18228840,
      "step": 32090
    },
    {
      "epoch": 563.070796460177,
      "grad_norm": 1.9461156171018956e-06,
      "learning_rate": 4.666640633369551e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18231328,
      "step": 32095
    },
    {
      "epoch": 563.1592920353983,
      "grad_norm": 2.5644058609941567e-07,
      "learning_rate": 4.660930427331323e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18234000,
      "step": 32100
    },
    {
      "epoch": 563.2477876106195,
      "grad_norm": 4.4876233573631907e-07,
      "learning_rate": 4.6552233578332244e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18236352,
      "step": 32105
    },
    {
      "epoch": 563.3362831858407,
      "grad_norm": 5.5854194215498865e-05,
      "learning_rate": 4.649519425755347e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18239312,
      "step": 32110
    },
    {
      "epoch": 563.4247787610619,
      "grad_norm": 1.3111443308844173e-07,
      "learning_rate": 4.64381863197732e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18242208,
      "step": 32115
    },
    {
      "epoch": 563.5132743362832,
      "grad_norm": 5.230803594713507e-07,
      "learning_rate": 4.638120977378269e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18244624,
      "step": 32120
    },
    {
      "epoch": 563.6017699115044,
      "grad_norm": 3.691521328619274e-07,
      "learning_rate": 4.632426462836848e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18247712,
      "step": 32125
    },
    {
      "epoch": 563.6902654867257,
      "grad_norm": 3.600603690756543e-07,
      "learning_rate": 4.626735089231224e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18251168,
      "step": 32130
    },
    {
      "epoch": 563.7787610619469,
      "grad_norm": 1.2337935118011956e-07,
      "learning_rate": 4.621046857439068e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18254016,
      "step": 32135
    },
    {
      "epoch": 563.8672566371681,
      "grad_norm": 1.9632835801530746e-07,
      "learning_rate": 4.615361768337587e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18257392,
      "step": 32140
    },
    {
      "epoch": 563.9557522123894,
      "grad_norm": 2.0313724746756634e-07,
      "learning_rate": 4.6096798228034946e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18260528,
      "step": 32145
    },
    {
      "epoch": 564.0353982300885,
      "grad_norm": 6.442642188630998e-07,
      "learning_rate": 4.604001021713008e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18263056,
      "step": 32150
    },
    {
      "epoch": 564.1238938053098,
      "grad_norm": 3.086108108618646e-06,
      "learning_rate": 4.598325365941883e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18265872,
      "step": 32155
    },
    {
      "epoch": 564.212389380531,
      "grad_norm": 4.127653028263012e-07,
      "learning_rate": 4.5926528563653645e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18268544,
      "step": 32160
    },
    {
      "epoch": 564.3008849557522,
      "grad_norm": 1.329281076323241e-06,
      "learning_rate": 4.5869834938582295e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18271472,
      "step": 32165
    },
    {
      "epoch": 564.3893805309734,
      "grad_norm": 6.26573068984726e-07,
      "learning_rate": 4.581317279294772e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18274480,
      "step": 32170
    },
    {
      "epoch": 564.4778761061947,
      "grad_norm": 6.792174076508672e-07,
      "learning_rate": 4.57565421354878e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18277248,
      "step": 32175
    },
    {
      "epoch": 564.566371681416,
      "grad_norm": 2.1782759063171397e-07,
      "learning_rate": 4.569994297493579e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18279808,
      "step": 32180
    },
    {
      "epoch": 564.6548672566372,
      "grad_norm": 1.53212354803145e-07,
      "learning_rate": 4.564337532002002e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18282448,
      "step": 32185
    },
    {
      "epoch": 564.7433628318585,
      "grad_norm": 3.557278489552118e-07,
      "learning_rate": 4.55868391794638e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18285472,
      "step": 32190
    },
    {
      "epoch": 564.8318584070796,
      "grad_norm": 1.3674408592123655e-06,
      "learning_rate": 4.553033456198588e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18287728,
      "step": 32195
    },
    {
      "epoch": 564.9203539823009,
      "grad_norm": 2.1456382626183768e-07,
      "learning_rate": 4.54738614762999e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18290608,
      "step": 32200
    },
    {
      "epoch": 564.9203539823009,
      "eval_loss": 0.925896406173706,
      "eval_runtime": 1.0759,
      "eval_samples_per_second": 23.236,
      "eval_steps_per_second": 12.083,
      "num_input_tokens_seen": 18290608,
      "step": 32200
    },
    {
      "epoch": 565.0,
      "grad_norm": 7.649288136235555e-08,
      "learning_rate": 4.541741993111465e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18293712,
      "step": 32205
    },
    {
      "epoch": 565.0884955752213,
      "grad_norm": 1.1597228422033368e-06,
      "learning_rate": 4.536100993513423e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18296784,
      "step": 32210
    },
    {
      "epoch": 565.1769911504425,
      "grad_norm": 5.485850351760746e-07,
      "learning_rate": 4.530463149705768e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18299248,
      "step": 32215
    },
    {
      "epoch": 565.2654867256637,
      "grad_norm": 3.8363376120287285e-07,
      "learning_rate": 4.524828462557934e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18302000,
      "step": 32220
    },
    {
      "epoch": 565.3539823008849,
      "grad_norm": 9.574318937666249e-06,
      "learning_rate": 4.5191969329388625e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18304720,
      "step": 32225
    },
    {
      "epoch": 565.4424778761062,
      "grad_norm": 2.3465602794203733e-07,
      "learning_rate": 4.5135685617169965e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18307456,
      "step": 32230
    },
    {
      "epoch": 565.5309734513274,
      "grad_norm": 2.9234067255856644e-07,
      "learning_rate": 4.507943349760313e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18310592,
      "step": 32235
    },
    {
      "epoch": 565.6194690265487,
      "grad_norm": 3.79442070652658e-07,
      "learning_rate": 4.502321297936277e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18313648,
      "step": 32240
    },
    {
      "epoch": 565.70796460177,
      "grad_norm": 1.5627700804543565e-06,
      "learning_rate": 4.496702407111888e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18316544,
      "step": 32245
    },
    {
      "epoch": 565.7964601769911,
      "grad_norm": 2.2893611628660437e-07,
      "learning_rate": 4.491086678153653e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18319248,
      "step": 32250
    },
    {
      "epoch": 565.8849557522124,
      "grad_norm": 6.782221362300334e-07,
      "learning_rate": 4.485474111927579e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18322176,
      "step": 32255
    },
    {
      "epoch": 565.9734513274336,
      "grad_norm": 2.1894183532822353e-07,
      "learning_rate": 4.479864709299197e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18325200,
      "step": 32260
    },
    {
      "epoch": 566.0530973451328,
      "grad_norm": 2.591328609469201e-07,
      "learning_rate": 4.474258471133555e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18327664,
      "step": 32265
    },
    {
      "epoch": 566.141592920354,
      "grad_norm": 3.2188697218771267e-07,
      "learning_rate": 4.4686553982952014e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18330784,
      "step": 32270
    },
    {
      "epoch": 566.2300884955753,
      "grad_norm": 1.200112080823601e-07,
      "learning_rate": 4.463055491648191e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18333424,
      "step": 32275
    },
    {
      "epoch": 566.3185840707964,
      "grad_norm": 2.9959343805785466e-07,
      "learning_rate": 4.457458752056112e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18336512,
      "step": 32280
    },
    {
      "epoch": 566.4070796460177,
      "grad_norm": 5.712740858143661e-07,
      "learning_rate": 4.451865180382042e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18339616,
      "step": 32285
    },
    {
      "epoch": 566.4955752212389,
      "grad_norm": 7.624508384651563e-07,
      "learning_rate": 4.4462747774885936e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18342144,
      "step": 32290
    },
    {
      "epoch": 566.5840707964602,
      "grad_norm": 2.6627270699464134e-07,
      "learning_rate": 4.440687544237859e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18344944,
      "step": 32295
    },
    {
      "epoch": 566.6725663716815,
      "grad_norm": 8.360237870874698e-07,
      "learning_rate": 4.435103481491471e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18347792,
      "step": 32300
    },
    {
      "epoch": 566.7610619469026,
      "grad_norm": 6.62488560010388e-07,
      "learning_rate": 4.429522590110569e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18350816,
      "step": 32305
    },
    {
      "epoch": 566.8495575221239,
      "grad_norm": 4.813796294911299e-07,
      "learning_rate": 4.423944870955779e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18353712,
      "step": 32310
    },
    {
      "epoch": 566.9380530973451,
      "grad_norm": 2.560809093665739e-07,
      "learning_rate": 4.418370324887272e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18356640,
      "step": 32315
    },
    {
      "epoch": 567.0176991150443,
      "grad_norm": 3.2919524528551847e-07,
      "learning_rate": 4.412798952764699e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18358896,
      "step": 32320
    },
    {
      "epoch": 567.1061946902655,
      "grad_norm": 1.3702369017210003e-07,
      "learning_rate": 4.407230755447245e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18361824,
      "step": 32325
    },
    {
      "epoch": 567.1946902654868,
      "grad_norm": 2.894331032621267e-07,
      "learning_rate": 4.401665733793598e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18364736,
      "step": 32330
    },
    {
      "epoch": 567.2831858407079,
      "grad_norm": 5.318501621331961e-07,
      "learning_rate": 4.3961038886619425e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18367536,
      "step": 32335
    },
    {
      "epoch": 567.3716814159292,
      "grad_norm": 2.0647173641918926e-06,
      "learning_rate": 4.39054522091e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18370400,
      "step": 32340
    },
    {
      "epoch": 567.4601769911504,
      "grad_norm": 9.957681186278933e-07,
      "learning_rate": 4.384989731394979e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18373152,
      "step": 32345
    },
    {
      "epoch": 567.5486725663717,
      "grad_norm": 1.6454386013720068e-06,
      "learning_rate": 4.379437420973598e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18376144,
      "step": 32350
    },
    {
      "epoch": 567.637168141593,
      "grad_norm": 4.103922765352763e-05,
      "learning_rate": 4.373888290502107e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18378544,
      "step": 32355
    },
    {
      "epoch": 567.7256637168142,
      "grad_norm": 3.562586243788246e-07,
      "learning_rate": 4.36834234083624e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18381360,
      "step": 32360
    },
    {
      "epoch": 567.8141592920354,
      "grad_norm": 1.2040432295634673e-07,
      "learning_rate": 4.362799572831258e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18384032,
      "step": 32365
    },
    {
      "epoch": 567.9026548672566,
      "grad_norm": 9.540704013488721e-07,
      "learning_rate": 4.35725998734193e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18387872,
      "step": 32370
    },
    {
      "epoch": 567.9911504424779,
      "grad_norm": 4.59495453242198e-07,
      "learning_rate": 4.3517235852225195e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18390832,
      "step": 32375
    },
    {
      "epoch": 568.070796460177,
      "grad_norm": 7.665928478672868e-07,
      "learning_rate": 4.346190367326822e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18393512,
      "step": 32380
    },
    {
      "epoch": 568.1592920353983,
      "grad_norm": 3.158101549161074e-07,
      "learning_rate": 4.340660334508115e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18396328,
      "step": 32385
    },
    {
      "epoch": 568.2477876106195,
      "grad_norm": 7.646492008461792e-07,
      "learning_rate": 4.335133487619206e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18399416,
      "step": 32390
    },
    {
      "epoch": 568.3362831858407,
      "grad_norm": 5.54978498712444e-07,
      "learning_rate": 4.329609827512409e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18402056,
      "step": 32395
    },
    {
      "epoch": 568.4247787610619,
      "grad_norm": 2.245960502023081e-07,
      "learning_rate": 4.324089355039531e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18404648,
      "step": 32400
    },
    {
      "epoch": 568.4247787610619,
      "eval_loss": 0.9181575179100037,
      "eval_runtime": 1.0687,
      "eval_samples_per_second": 23.394,
      "eval_steps_per_second": 12.165,
      "num_input_tokens_seen": 18404648,
      "step": 32400
    },
    {
      "epoch": 568.5132743362832,
      "grad_norm": 1.008184256079403e-07,
      "learning_rate": 4.3185720710519075e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18407176,
      "step": 32405
    },
    {
      "epoch": 568.6017699115044,
      "grad_norm": 8.496380132783088e-07,
      "learning_rate": 4.3130579764003724e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18410408,
      "step": 32410
    },
    {
      "epoch": 568.6902654867257,
      "grad_norm": 5.618678073915362e-07,
      "learning_rate": 4.307547071935267e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18413048,
      "step": 32415
    },
    {
      "epoch": 568.7787610619469,
      "grad_norm": 3.4104023143299855e-06,
      "learning_rate": 4.302039358506435e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18416088,
      "step": 32420
    },
    {
      "epoch": 568.8672566371681,
      "grad_norm": 1.050637337129956e-07,
      "learning_rate": 4.296534836963245e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18418936,
      "step": 32425
    },
    {
      "epoch": 568.9557522123894,
      "grad_norm": 4.858597435486445e-07,
      "learning_rate": 4.291033508154555e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18421816,
      "step": 32430
    },
    {
      "epoch": 569.0353982300885,
      "grad_norm": 4.178425001555297e-07,
      "learning_rate": 4.285535372928748e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18424008,
      "step": 32435
    },
    {
      "epoch": 569.1238938053098,
      "grad_norm": 1.5403935549329617e-07,
      "learning_rate": 4.280040432133695e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18427272,
      "step": 32440
    },
    {
      "epoch": 569.212389380531,
      "grad_norm": 2.649959185418993e-07,
      "learning_rate": 4.274548686616789e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18429656,
      "step": 32445
    },
    {
      "epoch": 569.3008849557522,
      "grad_norm": 1.6569732963489514e-07,
      "learning_rate": 4.2690601372249364e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18432760,
      "step": 32450
    },
    {
      "epoch": 569.3893805309734,
      "grad_norm": 4.523363941189018e-07,
      "learning_rate": 4.263574784804525e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18435352,
      "step": 32455
    },
    {
      "epoch": 569.4778761061947,
      "grad_norm": 2.1245375592116034e-07,
      "learning_rate": 4.258092630201479e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18438264,
      "step": 32460
    },
    {
      "epoch": 569.566371681416,
      "grad_norm": 1.0606819245140287e-07,
      "learning_rate": 4.252613674261202e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18441176,
      "step": 32465
    },
    {
      "epoch": 569.6548672566372,
      "grad_norm": 1.7230328808182094e-07,
      "learning_rate": 4.2471379178286224e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18444120,
      "step": 32470
    },
    {
      "epoch": 569.7433628318585,
      "grad_norm": 1.0210251844000595e-07,
      "learning_rate": 4.241665361748181e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18447480,
      "step": 32475
    },
    {
      "epoch": 569.8318584070796,
      "grad_norm": 9.732004713214337e-08,
      "learning_rate": 4.2361960068637994e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18450216,
      "step": 32480
    },
    {
      "epoch": 569.9203539823009,
      "grad_norm": 2.3845478835937683e-07,
      "learning_rate": 4.230729854018933e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18452776,
      "step": 32485
    },
    {
      "epoch": 570.0,
      "grad_norm": 4.214995996676407e-08,
      "learning_rate": 4.225266904056521e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18455456,
      "step": 32490
    },
    {
      "epoch": 570.0884955752213,
      "grad_norm": 1.4090709044012328e-07,
      "learning_rate": 4.21980715781903e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18458320,
      "step": 32495
    },
    {
      "epoch": 570.1769911504425,
      "grad_norm": 1.2732972720641555e-07,
      "learning_rate": 4.214350616148416e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18461104,
      "step": 32500
    },
    {
      "epoch": 570.2654867256637,
      "grad_norm": 6.234653255887679e-07,
      "learning_rate": 4.20889727988614e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18464144,
      "step": 32505
    },
    {
      "epoch": 570.3539823008849,
      "grad_norm": 4.116244838314742e-07,
      "learning_rate": 4.20344714987318e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18466960,
      "step": 32510
    },
    {
      "epoch": 570.4424778761062,
      "grad_norm": 4.834372475670534e-07,
      "learning_rate": 4.198000226950022e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18469600,
      "step": 32515
    },
    {
      "epoch": 570.5309734513274,
      "grad_norm": 1.1581216341483014e-07,
      "learning_rate": 4.192556511956635e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18472368,
      "step": 32520
    },
    {
      "epoch": 570.6194690265487,
      "grad_norm": 5.895230401620211e-07,
      "learning_rate": 4.18711600573252e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18474832,
      "step": 32525
    },
    {
      "epoch": 570.70796460177,
      "grad_norm": 1.7011241197906202e-06,
      "learning_rate": 4.181678709116671e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18478032,
      "step": 32530
    },
    {
      "epoch": 570.7964601769911,
      "grad_norm": 2.6662692675927246e-07,
      "learning_rate": 4.1762446229475785e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18481072,
      "step": 32535
    },
    {
      "epoch": 570.8849557522124,
      "grad_norm": 1.8183509098435024e-07,
      "learning_rate": 4.17081374806326e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18484096,
      "step": 32540
    },
    {
      "epoch": 570.9734513274336,
      "grad_norm": 3.03910695720333e-07,
      "learning_rate": 4.165386085301212e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18487328,
      "step": 32545
    },
    {
      "epoch": 571.0530973451328,
      "grad_norm": 2.4044601332207094e-07,
      "learning_rate": 4.1599616354984525e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18490064,
      "step": 32550
    },
    {
      "epoch": 571.141592920354,
      "grad_norm": 1.2488261802445777e-07,
      "learning_rate": 4.154540399491508e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18492528,
      "step": 32555
    },
    {
      "epoch": 571.2300884955753,
      "grad_norm": 1.1863031801340185e-07,
      "learning_rate": 4.149122378116394e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18495648,
      "step": 32560
    },
    {
      "epoch": 571.3185840707964,
      "grad_norm": 2.6482800308258447e-07,
      "learning_rate": 4.14370757220863e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18498288,
      "step": 32565
    },
    {
      "epoch": 571.4070796460177,
      "grad_norm": 2.1782261683256365e-07,
      "learning_rate": 4.138295982603263e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18500960,
      "step": 32570
    },
    {
      "epoch": 571.4955752212389,
      "grad_norm": 9.880037765697125e-08,
      "learning_rate": 4.132887610134814e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18503616,
      "step": 32575
    },
    {
      "epoch": 571.5840707964602,
      "grad_norm": 5.863006435902207e-07,
      "learning_rate": 4.127482455637335e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18506528,
      "step": 32580
    },
    {
      "epoch": 571.6725663716815,
      "grad_norm": 1.2849947097492986e-06,
      "learning_rate": 4.1220805199443545e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18509696,
      "step": 32585
    },
    {
      "epoch": 571.7610619469026,
      "grad_norm": 2.7120023560200934e-07,
      "learning_rate": 4.116681803888925e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18512336,
      "step": 32590
    },
    {
      "epoch": 571.8495575221239,
      "grad_norm": 4.5652086555492133e-07,
      "learning_rate": 4.111286308303605e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18514880,
      "step": 32595
    },
    {
      "epoch": 571.9380530973451,
      "grad_norm": 1.5279606202511786e-07,
      "learning_rate": 4.105894034020433e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18517216,
      "step": 32600
    },
    {
      "epoch": 571.9380530973451,
      "eval_loss": 0.9213732481002808,
      "eval_runtime": 1.0737,
      "eval_samples_per_second": 23.284,
      "eval_steps_per_second": 12.108,
      "num_input_tokens_seen": 18517216,
      "step": 32600
    },
    {
      "epoch": 572.0176991150443,
      "grad_norm": 1.9635608339285682e-07,
      "learning_rate": 4.100504981870975e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18520328,
      "step": 32605
    },
    {
      "epoch": 572.1061946902655,
      "grad_norm": 9.234278053327216e-08,
      "learning_rate": 4.0951191526862915e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18523496,
      "step": 32610
    },
    {
      "epoch": 572.1946902654868,
      "grad_norm": 1.7754905456968118e-07,
      "learning_rate": 4.089736547296938e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18526248,
      "step": 32615
    },
    {
      "epoch": 572.2831858407079,
      "grad_norm": 5.48379284737166e-06,
      "learning_rate": 4.08435716653299e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18528968,
      "step": 32620
    },
    {
      "epoch": 572.3716814159292,
      "grad_norm": 1.343552753496624e-07,
      "learning_rate": 4.0789810112240005e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18532024,
      "step": 32625
    },
    {
      "epoch": 572.4601769911504,
      "grad_norm": 1.888599001631519e-07,
      "learning_rate": 4.073608082199057e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18534840,
      "step": 32630
    },
    {
      "epoch": 572.5486725663717,
      "grad_norm": 1.1977127201134863e-07,
      "learning_rate": 4.068238380286718e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18537848,
      "step": 32635
    },
    {
      "epoch": 572.637168141593,
      "grad_norm": 1.0304378292858019e-06,
      "learning_rate": 4.062871906315072e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18540392,
      "step": 32640
    },
    {
      "epoch": 572.7256637168142,
      "grad_norm": 1.5052502249091049e-06,
      "learning_rate": 4.057508661111686e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18543512,
      "step": 32645
    },
    {
      "epoch": 572.8141592920354,
      "grad_norm": 1.6909326916447753e-07,
      "learning_rate": 4.052148645503648e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18546328,
      "step": 32650
    },
    {
      "epoch": 572.9026548672566,
      "grad_norm": 1.7332888546661707e-07,
      "learning_rate": 4.046791860317531e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18549272,
      "step": 32655
    },
    {
      "epoch": 572.9911504424779,
      "grad_norm": 9.320534672951908e-07,
      "learning_rate": 4.041438306379431e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18552280,
      "step": 32660
    },
    {
      "epoch": 573.070796460177,
      "grad_norm": 1.5593971056659939e-06,
      "learning_rate": 4.036087984514916e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18554696,
      "step": 32665
    },
    {
      "epoch": 573.1592920353983,
      "grad_norm": 3.505867027797649e-07,
      "learning_rate": 4.030740895549084e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18557464,
      "step": 32670
    },
    {
      "epoch": 573.2477876106195,
      "grad_norm": 1.9962922692684515e-07,
      "learning_rate": 4.025397040306531e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18560728,
      "step": 32675
    },
    {
      "epoch": 573.3362831858407,
      "grad_norm": 1.1752215414162492e-06,
      "learning_rate": 4.0200564196113285e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18563304,
      "step": 32680
    },
    {
      "epoch": 573.4247787610619,
      "grad_norm": 3.00201179470605e-07,
      "learning_rate": 4.014719034287079e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18566248,
      "step": 32685
    },
    {
      "epoch": 573.5132743362832,
      "grad_norm": 2.0285906430217437e-05,
      "learning_rate": 4.0093848851568775e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18569528,
      "step": 32690
    },
    {
      "epoch": 573.6017699115044,
      "grad_norm": 2.5766055955500633e-07,
      "learning_rate": 4.004053973043304e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18571992,
      "step": 32695
    },
    {
      "epoch": 573.6902654867257,
      "grad_norm": 2.7797111101790506e-07,
      "learning_rate": 3.998726298768465e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18574296,
      "step": 32700
    },
    {
      "epoch": 573.7787610619469,
      "grad_norm": 4.6625655159004964e-07,
      "learning_rate": 3.99340186315395e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18577624,
      "step": 32705
    },
    {
      "epoch": 573.8672566371681,
      "grad_norm": 1.6460973029097659e-07,
      "learning_rate": 3.988080667020849e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18581016,
      "step": 32710
    },
    {
      "epoch": 573.9557522123894,
      "grad_norm": 2.966354770705948e-07,
      "learning_rate": 3.982762711189766e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18583576,
      "step": 32715
    },
    {
      "epoch": 574.0353982300885,
      "grad_norm": 1.4094050015955872e-07,
      "learning_rate": 3.977447996480785e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18585736,
      "step": 32720
    },
    {
      "epoch": 574.1238938053098,
      "grad_norm": 1.4749010688319686e-06,
      "learning_rate": 3.97213652371351e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18588888,
      "step": 32725
    },
    {
      "epoch": 574.212389380531,
      "grad_norm": 1.2773662660947593e-07,
      "learning_rate": 3.966828293707042e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18592088,
      "step": 32730
    },
    {
      "epoch": 574.3008849557522,
      "grad_norm": 3.000285175858153e-07,
      "learning_rate": 3.961523307279963e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18594728,
      "step": 32735
    },
    {
      "epoch": 574.3893805309734,
      "grad_norm": 4.2551914702926297e-07,
      "learning_rate": 3.956221565250382e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18598120,
      "step": 32740
    },
    {
      "epoch": 574.4778761061947,
      "grad_norm": 2.9415502922347514e-07,
      "learning_rate": 3.950923068435883e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18600968,
      "step": 32745
    },
    {
      "epoch": 574.566371681416,
      "grad_norm": 2.6094769509654725e-06,
      "learning_rate": 3.945627817653566e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18603944,
      "step": 32750
    },
    {
      "epoch": 574.6548672566372,
      "grad_norm": 1.2971841556463914e-07,
      "learning_rate": 3.9403358137200335e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18606520,
      "step": 32755
    },
    {
      "epoch": 574.7433628318585,
      "grad_norm": 8.620465337116912e-07,
      "learning_rate": 3.9350470574513605e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18609512,
      "step": 32760
    },
    {
      "epoch": 574.8318584070796,
      "grad_norm": 1.0819533144967863e-06,
      "learning_rate": 3.9297615496631525e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18612184,
      "step": 32765
    },
    {
      "epoch": 574.9203539823009,
      "grad_norm": 1.5097310779310646e-07,
      "learning_rate": 3.924479291170505e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18614952,
      "step": 32770
    },
    {
      "epoch": 575.0,
      "grad_norm": 7.101618848537328e-08,
      "learning_rate": 3.919200282788002e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18617264,
      "step": 32775
    },
    {
      "epoch": 575.0884955752213,
      "grad_norm": 3.0151511509757256e-07,
      "learning_rate": 3.913924525329726e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18619920,
      "step": 32780
    },
    {
      "epoch": 575.1769911504425,
      "grad_norm": 3.139487603220914e-07,
      "learning_rate": 3.908652019609279e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18622800,
      "step": 32785
    },
    {
      "epoch": 575.2654867256637,
      "grad_norm": 3.927625300548243e-07,
      "learning_rate": 3.9033827664397364e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18625872,
      "step": 32790
    },
    {
      "epoch": 575.3539823008849,
      "grad_norm": 2.9681505111511797e-06,
      "learning_rate": 3.898116766633694e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18628464,
      "step": 32795
    },
    {
      "epoch": 575.4424778761062,
      "grad_norm": 1.8953524261178245e-07,
      "learning_rate": 3.8928540210032225e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18631296,
      "step": 32800
    },
    {
      "epoch": 575.4424778761062,
      "eval_loss": 0.9141509532928467,
      "eval_runtime": 1.0596,
      "eval_samples_per_second": 23.595,
      "eval_steps_per_second": 12.269,
      "num_input_tokens_seen": 18631296,
      "step": 32800
    },
    {
      "epoch": 575.5309734513274,
      "grad_norm": 3.4373448443147936e-07,
      "learning_rate": 3.887594530359909e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18634368,
      "step": 32805
    },
    {
      "epoch": 575.6194690265487,
      "grad_norm": 5.6481740102753974e-06,
      "learning_rate": 3.88233829551484e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18637280,
      "step": 32810
    },
    {
      "epoch": 575.70796460177,
      "grad_norm": 1.5527962204942014e-07,
      "learning_rate": 3.877085317278581e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18640048,
      "step": 32815
    },
    {
      "epoch": 575.7964601769911,
      "grad_norm": 6.278119712987973e-07,
      "learning_rate": 3.87183559646122e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18642432,
      "step": 32820
    },
    {
      "epoch": 575.8849557522124,
      "grad_norm": 7.791866778461554e-07,
      "learning_rate": 3.866589133872317e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18645696,
      "step": 32825
    },
    {
      "epoch": 575.9734513274336,
      "grad_norm": 4.560820343613159e-07,
      "learning_rate": 3.861345930320948e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18649200,
      "step": 32830
    },
    {
      "epoch": 576.0530973451328,
      "grad_norm": 1.3676900323389418e-07,
      "learning_rate": 3.856105986615688e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18651216,
      "step": 32835
    },
    {
      "epoch": 576.141592920354,
      "grad_norm": 1.4747008663107408e-06,
      "learning_rate": 3.850869303564589e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18653952,
      "step": 32840
    },
    {
      "epoch": 576.2300884955753,
      "grad_norm": 2.1646471282110724e-07,
      "learning_rate": 3.845635881975226e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18656352,
      "step": 32845
    },
    {
      "epoch": 576.3185840707964,
      "grad_norm": 1.6181297723960597e-07,
      "learning_rate": 3.840405722654647e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18660192,
      "step": 32850
    },
    {
      "epoch": 576.4070796460177,
      "grad_norm": 9.790453248115227e-08,
      "learning_rate": 3.835178826409419e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18663600,
      "step": 32855
    },
    {
      "epoch": 576.4955752212389,
      "grad_norm": 1.5130843848965014e-06,
      "learning_rate": 3.8299551940455895e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18666096,
      "step": 32860
    },
    {
      "epoch": 576.5840707964602,
      "grad_norm": 3.73373325146531e-07,
      "learning_rate": 3.824734826368703e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18668576,
      "step": 32865
    },
    {
      "epoch": 576.6725663716815,
      "grad_norm": 1.100496334061063e-07,
      "learning_rate": 3.819517724183813e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18671504,
      "step": 32870
    },
    {
      "epoch": 576.7610619469026,
      "grad_norm": 2.8749047942255856e-07,
      "learning_rate": 3.8143038882954648e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18674448,
      "step": 32875
    },
    {
      "epoch": 576.8495575221239,
      "grad_norm": 3.0105817927505996e-07,
      "learning_rate": 3.8090933195076867e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18677680,
      "step": 32880
    },
    {
      "epoch": 576.9380530973451,
      "grad_norm": 1.8097603060596157e-07,
      "learning_rate": 3.8038860186240198e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18680368,
      "step": 32885
    },
    {
      "epoch": 577.0176991150443,
      "grad_norm": 1.2395733506309625e-07,
      "learning_rate": 3.7986819864475026e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18682512,
      "step": 32890
    },
    {
      "epoch": 577.1061946902655,
      "grad_norm": 1.8535571655320382e-07,
      "learning_rate": 3.793481223780651e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18685056,
      "step": 32895
    },
    {
      "epoch": 577.1946902654868,
      "grad_norm": 4.5672035753341333e-07,
      "learning_rate": 3.788283731425496e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18688352,
      "step": 32900
    },
    {
      "epoch": 577.2831858407079,
      "grad_norm": 1.5120697582915454e-07,
      "learning_rate": 3.7830895101835488e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18691200,
      "step": 32905
    },
    {
      "epoch": 577.3716814159292,
      "grad_norm": 2.358877821961869e-07,
      "learning_rate": 3.7778985608558274e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18693472,
      "step": 32910
    },
    {
      "epoch": 577.4601769911504,
      "grad_norm": 2.9676200483663706e-06,
      "learning_rate": 3.7727108842428443e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18696320,
      "step": 32915
    },
    {
      "epoch": 577.5486725663717,
      "grad_norm": 3.5368756812204083e-07,
      "learning_rate": 3.7675264811446065e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18699424,
      "step": 32920
    },
    {
      "epoch": 577.637168141593,
      "grad_norm": 2.813883384078508e-07,
      "learning_rate": 3.7623453523605994e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18702352,
      "step": 32925
    },
    {
      "epoch": 577.7256637168142,
      "grad_norm": 9.429593461618424e-08,
      "learning_rate": 3.757167498689834e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18705824,
      "step": 32930
    },
    {
      "epoch": 577.8141592920354,
      "grad_norm": 1.4272944781623664e-07,
      "learning_rate": 3.7519929209307914e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18708848,
      "step": 32935
    },
    {
      "epoch": 577.9026548672566,
      "grad_norm": 1.6013408412618446e-06,
      "learning_rate": 3.746821619881463e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18711600,
      "step": 32940
    },
    {
      "epoch": 577.9911504424779,
      "grad_norm": 1.9651608340609528e-07,
      "learning_rate": 3.74165359633932e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18714512,
      "step": 32945
    },
    {
      "epoch": 578.070796460177,
      "grad_norm": 1.652579442179558e-07,
      "learning_rate": 3.736488851101341e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18716952,
      "step": 32950
    },
    {
      "epoch": 578.1592920353983,
      "grad_norm": 3.7181749235060124e-07,
      "learning_rate": 3.7313273849640035e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18719576,
      "step": 32955
    },
    {
      "epoch": 578.2477876106195,
      "grad_norm": 9.65020376497705e-07,
      "learning_rate": 3.7261691987232533e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18722424,
      "step": 32960
    },
    {
      "epoch": 578.3362831858407,
      "grad_norm": 8.971848330929788e-08,
      "learning_rate": 3.7210142931745575e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18725784,
      "step": 32965
    },
    {
      "epoch": 578.4247787610619,
      "grad_norm": 9.774837650411428e-08,
      "learning_rate": 3.7158626691128712e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18728648,
      "step": 32970
    },
    {
      "epoch": 578.5132743362832,
      "grad_norm": 3.5670509532792494e-06,
      "learning_rate": 3.710714327332629e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18731080,
      "step": 32975
    },
    {
      "epoch": 578.6017699115044,
      "grad_norm": 1.8976639637457993e-07,
      "learning_rate": 3.7055692686277815e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18733704,
      "step": 32980
    },
    {
      "epoch": 578.6902654867257,
      "grad_norm": 2.2259541765379254e-07,
      "learning_rate": 3.70042749379175e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18736664,
      "step": 32985
    },
    {
      "epoch": 578.7787610619469,
      "grad_norm": 1.783682535005937e-07,
      "learning_rate": 3.6952890036174693e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18739928,
      "step": 32990
    },
    {
      "epoch": 578.8672566371681,
      "grad_norm": 4.009395979664987e-07,
      "learning_rate": 3.690153798897353e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18742552,
      "step": 32995
    },
    {
      "epoch": 578.9557522123894,
      "grad_norm": 1.5775427755215787e-06,
      "learning_rate": 3.6850218804233225e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18745416,
      "step": 33000
    },
    {
      "epoch": 578.9557522123894,
      "eval_loss": 0.9105704426765442,
      "eval_runtime": 1.0659,
      "eval_samples_per_second": 23.455,
      "eval_steps_per_second": 12.197,
      "num_input_tokens_seen": 18745416,
      "step": 33000
    },
    {
      "epoch": 579.0353982300885,
      "grad_norm": 2.245268575506998e-07,
      "learning_rate": 3.679893248986779e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18747704,
      "step": 33005
    },
    {
      "epoch": 579.1238938053098,
      "grad_norm": 9.087485750569613e-07,
      "learning_rate": 3.6747679053786147e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18750712,
      "step": 33010
    },
    {
      "epoch": 579.212389380531,
      "grad_norm": 9.471472139921389e-07,
      "learning_rate": 3.669645850389228e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18753192,
      "step": 33015
    },
    {
      "epoch": 579.3008849557522,
      "grad_norm": 9.763257935446745e-08,
      "learning_rate": 3.664527084808514e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18756168,
      "step": 33020
    },
    {
      "epoch": 579.3893805309734,
      "grad_norm": 2.0852716886565759e-07,
      "learning_rate": 3.6594116094258337e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18759176,
      "step": 33025
    },
    {
      "epoch": 579.4778761061947,
      "grad_norm": 3.385984541637299e-07,
      "learning_rate": 3.6542994250300665e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18761576,
      "step": 33030
    },
    {
      "epoch": 579.566371681416,
      "grad_norm": 8.494398429093053e-08,
      "learning_rate": 3.6491905324095825e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18764488,
      "step": 33035
    },
    {
      "epoch": 579.6548672566372,
      "grad_norm": 3.5098796047350334e-07,
      "learning_rate": 3.644084932352221e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18767128,
      "step": 33040
    },
    {
      "epoch": 579.7433628318585,
      "grad_norm": 5.884103515541028e-08,
      "learning_rate": 3.6389826256453457e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18770376,
      "step": 33045
    },
    {
      "epoch": 579.8318584070796,
      "grad_norm": 3.201683284714818e-06,
      "learning_rate": 3.633883613075781e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18773064,
      "step": 33050
    },
    {
      "epoch": 579.9203539823009,
      "grad_norm": 1.87345889912649e-07,
      "learning_rate": 3.6287878954298693e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18776232,
      "step": 33055
    },
    {
      "epoch": 580.0,
      "grad_norm": 2.7386050760469516e-07,
      "learning_rate": 3.6236954734934354e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18778768,
      "step": 33060
    },
    {
      "epoch": 580.0884955752213,
      "grad_norm": 1.4115863677943707e-06,
      "learning_rate": 3.618606348051784e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18781232,
      "step": 33065
    },
    {
      "epoch": 580.1769911504425,
      "grad_norm": 9.77439640337252e-08,
      "learning_rate": 3.6135205198897376e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18784032,
      "step": 33070
    },
    {
      "epoch": 580.2654867256637,
      "grad_norm": 2.294895438126332e-07,
      "learning_rate": 3.6084379897915854e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18786496,
      "step": 33075
    },
    {
      "epoch": 580.3539823008849,
      "grad_norm": 9.191255401219678e-08,
      "learning_rate": 3.6033587585411115e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18789728,
      "step": 33080
    },
    {
      "epoch": 580.4424778761062,
      "grad_norm": 1.5477110082429135e-07,
      "learning_rate": 3.5982828269216117e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18792496,
      "step": 33085
    },
    {
      "epoch": 580.5309734513274,
      "grad_norm": 2.433699251014332e-07,
      "learning_rate": 3.593210195715843e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18795040,
      "step": 33090
    },
    {
      "epoch": 580.6194690265487,
      "grad_norm": 2.153011280370265e-07,
      "learning_rate": 3.5881408657060773e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18798128,
      "step": 33095
    },
    {
      "epoch": 580.70796460177,
      "grad_norm": 9.749262517289026e-07,
      "learning_rate": 3.583074837674075e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18801024,
      "step": 33100
    },
    {
      "epoch": 580.7964601769911,
      "grad_norm": 5.732774752686964e-07,
      "learning_rate": 3.578012112401069e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18803984,
      "step": 33105
    },
    {
      "epoch": 580.8849557522124,
      "grad_norm": 1.7522673090297758e-07,
      "learning_rate": 3.5729526906677996e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18806832,
      "step": 33110
    },
    {
      "epoch": 580.9734513274336,
      "grad_norm": 5.189575858821627e-06,
      "learning_rate": 3.5678965732545007e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18809840,
      "step": 33115
    },
    {
      "epoch": 581.0530973451328,
      "grad_norm": 1.958260327228345e-07,
      "learning_rate": 3.562843760940876e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18812080,
      "step": 33120
    },
    {
      "epoch": 581.141592920354,
      "grad_norm": 3.9863721212896053e-07,
      "learning_rate": 3.5577942545061473e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18814976,
      "step": 33125
    },
    {
      "epoch": 581.2300884955753,
      "grad_norm": 1.916204837471014e-06,
      "learning_rate": 3.5527480547289967e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18817808,
      "step": 33130
    },
    {
      "epoch": 581.3185840707964,
      "grad_norm": 2.2504593744088197e-07,
      "learning_rate": 3.547705162387624e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18820512,
      "step": 33135
    },
    {
      "epoch": 581.4070796460177,
      "grad_norm": 6.766325100215909e-07,
      "learning_rate": 3.542665578259699e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18823232,
      "step": 33140
    },
    {
      "epoch": 581.4955752212389,
      "grad_norm": 4.4983528368902626e-07,
      "learning_rate": 3.5376293031223945e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18825856,
      "step": 33145
    },
    {
      "epoch": 581.5840707964602,
      "grad_norm": 1.9070452594860399e-07,
      "learning_rate": 3.5325963377523614e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18828608,
      "step": 33150
    },
    {
      "epoch": 581.6725663716815,
      "grad_norm": 1.8717148009272933e-07,
      "learning_rate": 3.5275666829257536e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18831824,
      "step": 33155
    },
    {
      "epoch": 581.7610619469026,
      "grad_norm": 5.694519131793641e-07,
      "learning_rate": 3.5225403394181955e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18834416,
      "step": 33160
    },
    {
      "epoch": 581.8495575221239,
      "grad_norm": 1.203057308885036e-06,
      "learning_rate": 3.517517308004828e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18837776,
      "step": 33165
    },
    {
      "epoch": 581.9380530973451,
      "grad_norm": 1.6267786406842788e-07,
      "learning_rate": 3.512497589460251e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18840480,
      "step": 33170
    },
    {
      "epoch": 582.0176991150443,
      "grad_norm": 1.33930214474276e-07,
      "learning_rate": 3.5074811845585727e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18843368,
      "step": 33175
    },
    {
      "epoch": 582.1061946902655,
      "grad_norm": 3.560132313396025e-07,
      "learning_rate": 3.5024680940733937e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18846072,
      "step": 33180
    },
    {
      "epoch": 582.1946902654868,
      "grad_norm": 2.4821898136906384e-07,
      "learning_rate": 3.4974583187777852e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18849240,
      "step": 33185
    },
    {
      "epoch": 582.2831858407079,
      "grad_norm": 2.2301007618352742e-07,
      "learning_rate": 3.4924518594443204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18851896,
      "step": 33190
    },
    {
      "epoch": 582.3716814159292,
      "grad_norm": 1.3060778769613535e-07,
      "learning_rate": 3.4874487168450682e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18855000,
      "step": 33195
    },
    {
      "epoch": 582.4601769911504,
      "grad_norm": 2.9337203955037694e-07,
      "learning_rate": 3.482448891751558e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18857896,
      "step": 33200
    },
    {
      "epoch": 582.4601769911504,
      "eval_loss": 0.9186511635780334,
      "eval_runtime": 1.0625,
      "eval_samples_per_second": 23.529,
      "eval_steps_per_second": 12.235,
      "num_input_tokens_seen": 18857896,
      "step": 33200
    },
    {
      "epoch": 582.5486725663717,
      "grad_norm": 1.228459893809486e-07,
      "learning_rate": 3.477452384934843e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18860696,
      "step": 33205
    },
    {
      "epoch": 582.637168141593,
      "grad_norm": 4.771854378304852e-07,
      "learning_rate": 3.472459197165434e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18863448,
      "step": 33210
    },
    {
      "epoch": 582.7256637168142,
      "grad_norm": 2.680663442333753e-07,
      "learning_rate": 3.4674693292133518e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18866344,
      "step": 33215
    },
    {
      "epoch": 582.8141592920354,
      "grad_norm": 2.182077878387645e-07,
      "learning_rate": 3.4624827818480977e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18869256,
      "step": 33220
    },
    {
      "epoch": 582.9026548672566,
      "grad_norm": 4.091806147243915e-07,
      "learning_rate": 3.4574995558386474e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18871912,
      "step": 33225
    },
    {
      "epoch": 582.9911504424779,
      "grad_norm": 9.170756243292999e-07,
      "learning_rate": 3.452519651953487e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18874824,
      "step": 33230
    },
    {
      "epoch": 583.070796460177,
      "grad_norm": 1.8767302378819295e-07,
      "learning_rate": 3.447543070960585e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18877168,
      "step": 33235
    },
    {
      "epoch": 583.1592920353983,
      "grad_norm": 1.146066438195703e-06,
      "learning_rate": 3.4425698136273778e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18879680,
      "step": 33240
    },
    {
      "epoch": 583.2477876106195,
      "grad_norm": 8.125660144742142e-08,
      "learning_rate": 3.437599880720821e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18882672,
      "step": 33245
    },
    {
      "epoch": 583.3362831858407,
      "grad_norm": 1.2531720585684525e-06,
      "learning_rate": 3.4326332730073267e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18885328,
      "step": 33250
    },
    {
      "epoch": 583.4247787610619,
      "grad_norm": 5.512005145646981e-07,
      "learning_rate": 3.427669991252813e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18888608,
      "step": 33255
    },
    {
      "epoch": 583.5132743362832,
      "grad_norm": 3.5603599712885625e-07,
      "learning_rate": 3.42271003622269e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18891328,
      "step": 33260
    },
    {
      "epoch": 583.6017699115044,
      "grad_norm": 1.3486300076692714e-07,
      "learning_rate": 3.4177534086818286e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18893664,
      "step": 33265
    },
    {
      "epoch": 583.6902654867257,
      "grad_norm": 3.0935601103010413e-07,
      "learning_rate": 3.412800109394612e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18896768,
      "step": 33270
    },
    {
      "epoch": 583.7787610619469,
      "grad_norm": 1.7938426026375964e-05,
      "learning_rate": 3.4078501391249044e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18899664,
      "step": 33275
    },
    {
      "epoch": 583.8672566371681,
      "grad_norm": 3.4964418205163383e-07,
      "learning_rate": 3.4029034986360453e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18902896,
      "step": 33280
    },
    {
      "epoch": 583.9557522123894,
      "grad_norm": 2.702123822473368e-07,
      "learning_rate": 3.397960188690877e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18906096,
      "step": 33285
    },
    {
      "epoch": 584.0353982300885,
      "grad_norm": 1.8658023748230335e-07,
      "learning_rate": 3.393020210051717e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18908552,
      "step": 33290
    },
    {
      "epoch": 584.1238938053098,
      "grad_norm": 1.2483612010782963e-07,
      "learning_rate": 3.3880835634803655e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18911352,
      "step": 33295
    },
    {
      "epoch": 584.212389380531,
      "grad_norm": 6.317712291092903e-07,
      "learning_rate": 3.383150249738126e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18914264,
      "step": 33300
    },
    {
      "epoch": 584.3008849557522,
      "grad_norm": 1.6241087053003866e-07,
      "learning_rate": 3.3782202695857663e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18916792,
      "step": 33305
    },
    {
      "epoch": 584.3893805309734,
      "grad_norm": 4.4662013465313066e-07,
      "learning_rate": 3.373293623783558e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18919640,
      "step": 33310
    },
    {
      "epoch": 584.4778761061947,
      "grad_norm": 4.023184203560959e-07,
      "learning_rate": 3.368370313091257e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18922456,
      "step": 33315
    },
    {
      "epoch": 584.566371681416,
      "grad_norm": 3.9333971812993695e-07,
      "learning_rate": 3.363450338268087e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18925608,
      "step": 33320
    },
    {
      "epoch": 584.6548672566372,
      "grad_norm": 3.1311767543229507e-06,
      "learning_rate": 3.358533700072783e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18928568,
      "step": 33325
    },
    {
      "epoch": 584.7433628318585,
      "grad_norm": 2.1086466972519702e-07,
      "learning_rate": 3.3536203992635377e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18931592,
      "step": 33330
    },
    {
      "epoch": 584.8318584070796,
      "grad_norm": 5.656637540596421e-07,
      "learning_rate": 3.348710436598057e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18933800,
      "step": 33335
    },
    {
      "epoch": 584.9203539823009,
      "grad_norm": 3.273764605182805e-06,
      "learning_rate": 3.3438038128335155e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18936952,
      "step": 33340
    },
    {
      "epoch": 585.0,
      "grad_norm": 2.878094846892054e-07,
      "learning_rate": 3.338900528726571e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18939504,
      "step": 33345
    },
    {
      "epoch": 585.0884955752213,
      "grad_norm": 1.5495528771225509e-07,
      "learning_rate": 3.3340005850333812e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18942224,
      "step": 33350
    },
    {
      "epoch": 585.1769911504425,
      "grad_norm": 1.91774688573787e-07,
      "learning_rate": 3.329103982509568e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18944976,
      "step": 33355
    },
    {
      "epoch": 585.2654867256637,
      "grad_norm": 4.372987405076856e-06,
      "learning_rate": 3.324210721910259e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18947680,
      "step": 33360
    },
    {
      "epoch": 585.3539823008849,
      "grad_norm": 1.555274877773627e-07,
      "learning_rate": 3.319320803990053e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18950320,
      "step": 33365
    },
    {
      "epoch": 585.4424778761062,
      "grad_norm": 9.147615855908953e-07,
      "learning_rate": 3.3144342295030274e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18953200,
      "step": 33370
    },
    {
      "epoch": 585.5309734513274,
      "grad_norm": 5.436976948658412e-07,
      "learning_rate": 3.309550999202765e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18955760,
      "step": 33375
    },
    {
      "epoch": 585.6194690265487,
      "grad_norm": 1.5256126744134235e-06,
      "learning_rate": 3.3046711138423197e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18959216,
      "step": 33380
    },
    {
      "epoch": 585.70796460177,
      "grad_norm": 2.006346448979457e-07,
      "learning_rate": 3.2997945741742255e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18962608,
      "step": 33385
    },
    {
      "epoch": 585.7964601769911,
      "grad_norm": 1.5566411093459465e-07,
      "learning_rate": 3.2949213809505082e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18966064,
      "step": 33390
    },
    {
      "epoch": 585.8849557522124,
      "grad_norm": 1.0847437437178087e-07,
      "learning_rate": 3.2900515349226834e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18968752,
      "step": 33395
    },
    {
      "epoch": 585.9734513274336,
      "grad_norm": 2.430478218684584e-07,
      "learning_rate": 3.285185036841731e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18971344,
      "step": 33400
    },
    {
      "epoch": 585.9734513274336,
      "eval_loss": 0.9217712879180908,
      "eval_runtime": 1.0639,
      "eval_samples_per_second": 23.498,
      "eval_steps_per_second": 12.219,
      "num_input_tokens_seen": 18971344,
      "step": 33400
    },
    {
      "epoch": 586.0530973451328,
      "grad_norm": 2.4741635229474923e-07,
      "learning_rate": 3.2803218874581377e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18973736,
      "step": 33405
    },
    {
      "epoch": 586.141592920354,
      "grad_norm": 1.6670988145506271e-07,
      "learning_rate": 3.2754620875218494e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18976392,
      "step": 33410
    },
    {
      "epoch": 586.2300884955753,
      "grad_norm": 2.9518454880417266e-07,
      "learning_rate": 3.2706056377823146e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18979576,
      "step": 33415
    },
    {
      "epoch": 586.3185840707964,
      "grad_norm": 1.2206685369164916e-06,
      "learning_rate": 3.2657525389884647e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18982136,
      "step": 33420
    },
    {
      "epoch": 586.4070796460177,
      "grad_norm": 6.734982207490248e-07,
      "learning_rate": 3.260902791888698e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18985288,
      "step": 33425
    },
    {
      "epoch": 586.4955752212389,
      "grad_norm": 1.2950587802151858e-07,
      "learning_rate": 3.2560563972309166e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18987864,
      "step": 33430
    },
    {
      "epoch": 586.5840707964602,
      "grad_norm": 6.386666200342006e-07,
      "learning_rate": 3.251213355762489e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18991144,
      "step": 33435
    },
    {
      "epoch": 586.6725663716815,
      "grad_norm": 3.651375379831734e-07,
      "learning_rate": 3.2463736682302707e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18993976,
      "step": 33440
    },
    {
      "epoch": 586.7610619469026,
      "grad_norm": 1.3952080735180061e-06,
      "learning_rate": 3.2415373353806124e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18997032,
      "step": 33445
    },
    {
      "epoch": 586.8495575221239,
      "grad_norm": 3.2813161965350446e-07,
      "learning_rate": 3.236704357959322e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 18999736,
      "step": 33450
    },
    {
      "epoch": 586.9380530973451,
      "grad_norm": 1.778837344090789e-07,
      "learning_rate": 3.2318747367117154e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19002936,
      "step": 33455
    },
    {
      "epoch": 587.0176991150443,
      "grad_norm": 1.7204371260959306e-07,
      "learning_rate": 3.227048472382585e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19005400,
      "step": 33460
    },
    {
      "epoch": 587.1061946902655,
      "grad_norm": 3.1410331757797394e-07,
      "learning_rate": 3.2222255657161915e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19007928,
      "step": 33465
    },
    {
      "epoch": 587.1946902654868,
      "grad_norm": 1.7330623904854292e-06,
      "learning_rate": 3.2174060174562924e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19010600,
      "step": 33470
    },
    {
      "epoch": 587.2831858407079,
      "grad_norm": 8.289311637099672e-08,
      "learning_rate": 3.2125898283461298e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19014456,
      "step": 33475
    },
    {
      "epoch": 587.3716814159292,
      "grad_norm": 7.497345109186426e-07,
      "learning_rate": 3.207776999128406e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19016920,
      "step": 33480
    },
    {
      "epoch": 587.4601769911504,
      "grad_norm": 2.7884297537639213e-07,
      "learning_rate": 3.202967530545331e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19019432,
      "step": 33485
    },
    {
      "epoch": 587.5486725663717,
      "grad_norm": 9.852776372554217e-08,
      "learning_rate": 3.1981614233385778e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19022376,
      "step": 33490
    },
    {
      "epoch": 587.637168141593,
      "grad_norm": 5.16110389980895e-07,
      "learning_rate": 3.1933586782493115e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19025416,
      "step": 33495
    },
    {
      "epoch": 587.7256637168142,
      "grad_norm": 4.83430085296277e-07,
      "learning_rate": 3.188559296018184e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19028360,
      "step": 33500
    },
    {
      "epoch": 587.8141592920354,
      "grad_norm": 3.581956207199255e-07,
      "learning_rate": 3.1837632773853098e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19031432,
      "step": 33505
    },
    {
      "epoch": 587.9026548672566,
      "grad_norm": 6.61936439882993e-07,
      "learning_rate": 3.178970623090294e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19034456,
      "step": 33510
    },
    {
      "epoch": 587.9911504424779,
      "grad_norm": 1.7617609046283178e-05,
      "learning_rate": 3.174181333872234e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19037032,
      "step": 33515
    },
    {
      "epoch": 588.070796460177,
      "grad_norm": 3.085055197971087e-07,
      "learning_rate": 3.169395410469686e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19039232,
      "step": 33520
    },
    {
      "epoch": 588.1592920353983,
      "grad_norm": 4.450958783763781e-07,
      "learning_rate": 3.164612853620713e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19042032,
      "step": 33525
    },
    {
      "epoch": 588.2477876106195,
      "grad_norm": 3.249995188525645e-07,
      "learning_rate": 3.1598336640628333e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19045184,
      "step": 33530
    },
    {
      "epoch": 588.3362831858407,
      "grad_norm": 1.5466470415503863e-07,
      "learning_rate": 3.155057842533063e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19048288,
      "step": 33535
    },
    {
      "epoch": 588.4247787610619,
      "grad_norm": 1.3615672287414782e-06,
      "learning_rate": 3.1502853897678984e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19051520,
      "step": 33540
    },
    {
      "epoch": 588.5132743362832,
      "grad_norm": 1.3586598868187139e-07,
      "learning_rate": 3.1455163065033017e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19054576,
      "step": 33545
    },
    {
      "epoch": 588.6017699115044,
      "grad_norm": 1.4559563510374574e-07,
      "learning_rate": 3.140750593474734e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19057552,
      "step": 33550
    },
    {
      "epoch": 588.6902654867257,
      "grad_norm": 3.350789370415441e-07,
      "learning_rate": 3.1359882514171294e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19060400,
      "step": 33555
    },
    {
      "epoch": 588.7787610619469,
      "grad_norm": 8.503251791580624e-08,
      "learning_rate": 3.1312292810648903e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19062864,
      "step": 33560
    },
    {
      "epoch": 588.8672566371681,
      "grad_norm": 1.5684000800320064e-06,
      "learning_rate": 3.1264736831519204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19065840,
      "step": 33565
    },
    {
      "epoch": 588.9557522123894,
      "grad_norm": 2.181593856676045e-07,
      "learning_rate": 3.1217214584115863e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19068432,
      "step": 33570
    },
    {
      "epoch": 589.0353982300885,
      "grad_norm": 1.36839119591059e-07,
      "learning_rate": 3.116972607576746e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19070672,
      "step": 33575
    },
    {
      "epoch": 589.1238938053098,
      "grad_norm": 8.210864166358078e-07,
      "learning_rate": 3.1122271313797303e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19073664,
      "step": 33580
    },
    {
      "epoch": 589.212389380531,
      "grad_norm": 6.884483809699304e-07,
      "learning_rate": 3.107485030552343e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19076400,
      "step": 33585
    },
    {
      "epoch": 589.3008849557522,
      "grad_norm": 1.1352096151995283e-07,
      "learning_rate": 3.1027463058258848e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19079264,
      "step": 33590
    },
    {
      "epoch": 589.3893805309734,
      "grad_norm": 2.555169373863464e-07,
      "learning_rate": 3.0980109579311273e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19081904,
      "step": 33595
    },
    {
      "epoch": 589.4778761061947,
      "grad_norm": 8.960328159446362e-07,
      "learning_rate": 3.093278987598314e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19085248,
      "step": 33600
    },
    {
      "epoch": 589.4778761061947,
      "eval_loss": 0.923610270023346,
      "eval_runtime": 1.0614,
      "eval_samples_per_second": 23.553,
      "eval_steps_per_second": 12.248,
      "num_input_tokens_seen": 19085248,
      "step": 33600
    },
    {
      "epoch": 589.566371681416,
      "grad_norm": 4.4841925728178467e-07,
      "learning_rate": 3.0885503955571826e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19087968,
      "step": 33605
    },
    {
      "epoch": 589.6548672566372,
      "grad_norm": 3.8916846278880257e-07,
      "learning_rate": 3.0838251825369313e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19090672,
      "step": 33610
    },
    {
      "epoch": 589.7433628318585,
      "grad_norm": 3.083779915868945e-07,
      "learning_rate": 3.0791033492662517e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19093360,
      "step": 33615
    },
    {
      "epoch": 589.8318584070796,
      "grad_norm": 2.9224180252640508e-05,
      "learning_rate": 3.0743848964733203e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19096592,
      "step": 33620
    },
    {
      "epoch": 589.9203539823009,
      "grad_norm": 4.796153234565281e-07,
      "learning_rate": 3.0696698248857625e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19099872,
      "step": 33625
    },
    {
      "epoch": 590.0,
      "grad_norm": 3.519659941275677e-08,
      "learning_rate": 3.0649581352307192e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19101944,
      "step": 33630
    },
    {
      "epoch": 590.0884955752213,
      "grad_norm": 1.5105788975233736e-07,
      "learning_rate": 3.060249828234776e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19105160,
      "step": 33635
    },
    {
      "epoch": 590.1769911504425,
      "grad_norm": 2.0946846746028314e-07,
      "learning_rate": 3.055544904624025e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19108088,
      "step": 33640
    },
    {
      "epoch": 590.2654867256637,
      "grad_norm": 2.0736604255944258e-07,
      "learning_rate": 3.050843365124026e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19111032,
      "step": 33645
    },
    {
      "epoch": 590.3539823008849,
      "grad_norm": 1.558070607643458e-06,
      "learning_rate": 3.0461452104598083e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19113704,
      "step": 33650
    },
    {
      "epoch": 590.4424778761062,
      "grad_norm": 3.013341824953386e-07,
      "learning_rate": 3.0414504413558836e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19116456,
      "step": 33655
    },
    {
      "epoch": 590.5309734513274,
      "grad_norm": 4.061064373672707e-07,
      "learning_rate": 3.0367590585362564e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19119416,
      "step": 33660
    },
    {
      "epoch": 590.6194690265487,
      "grad_norm": 7.75464101820944e-08,
      "learning_rate": 3.0320710627243813e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19122520,
      "step": 33665
    },
    {
      "epoch": 590.70796460177,
      "grad_norm": 5.59034606339992e-06,
      "learning_rate": 3.027386454643222e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19125240,
      "step": 33670
    },
    {
      "epoch": 590.7964601769911,
      "grad_norm": 1.708429095970132e-07,
      "learning_rate": 3.0227052350151914e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19127912,
      "step": 33675
    },
    {
      "epoch": 590.8849557522124,
      "grad_norm": 4.748980018121074e-07,
      "learning_rate": 3.0180274045621957e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19130696,
      "step": 33680
    },
    {
      "epoch": 590.9734513274336,
      "grad_norm": 3.833264372588019e-07,
      "learning_rate": 3.013352964005625e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19133816,
      "step": 33685
    },
    {
      "epoch": 591.0530973451328,
      "grad_norm": 1.570100067738167e-07,
      "learning_rate": 3.0086819140663218e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19136248,
      "step": 33690
    },
    {
      "epoch": 591.141592920354,
      "grad_norm": 2.138994830147567e-07,
      "learning_rate": 3.0040142554646265e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19138872,
      "step": 33695
    },
    {
      "epoch": 591.2300884955753,
      "grad_norm": 1.7805473362386692e-07,
      "learning_rate": 2.999349988920361e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19141784,
      "step": 33700
    },
    {
      "epoch": 591.3185840707964,
      "grad_norm": 1.3703015611099545e-07,
      "learning_rate": 2.994689115152796e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19144808,
      "step": 33705
    },
    {
      "epoch": 591.4070796460177,
      "grad_norm": 1.0378257542242864e-07,
      "learning_rate": 2.9900316348807105e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19147336,
      "step": 33710
    },
    {
      "epoch": 591.4955752212389,
      "grad_norm": 5.138491019351932e-07,
      "learning_rate": 2.985377548822338e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19149992,
      "step": 33715
    },
    {
      "epoch": 591.5840707964602,
      "grad_norm": 4.278184349004732e-07,
      "learning_rate": 2.980726857695404e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19153048,
      "step": 33720
    },
    {
      "epoch": 591.6725663716815,
      "grad_norm": 2.560815630658908e-07,
      "learning_rate": 2.9760795622171017e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19156120,
      "step": 33725
    },
    {
      "epoch": 591.7610619469026,
      "grad_norm": 2.42369679881449e-07,
      "learning_rate": 2.971435663104094e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19159192,
      "step": 33730
    },
    {
      "epoch": 591.8495575221239,
      "grad_norm": 2.4411735921603395e-07,
      "learning_rate": 2.9667951610725385e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19162104,
      "step": 33735
    },
    {
      "epoch": 591.9380530973451,
      "grad_norm": 1.189397380585433e-07,
      "learning_rate": 2.9621580568380575e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19165496,
      "step": 33740
    },
    {
      "epoch": 592.0176991150443,
      "grad_norm": 9.980081472349411e-08,
      "learning_rate": 2.9575243511157453e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19168032,
      "step": 33745
    },
    {
      "epoch": 592.1061946902655,
      "grad_norm": 5.07698928231548e-07,
      "learning_rate": 2.952894044620186e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19170752,
      "step": 33750
    },
    {
      "epoch": 592.1946902654868,
      "grad_norm": 3.705596611780493e-07,
      "learning_rate": 2.948267138065419e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19173712,
      "step": 33755
    },
    {
      "epoch": 592.2831858407079,
      "grad_norm": 2.1303509356584982e-07,
      "learning_rate": 2.943643632164983e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19176624,
      "step": 33760
    },
    {
      "epoch": 592.3716814159292,
      "grad_norm": 6.407588898582617e-06,
      "learning_rate": 2.939023527631879e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19179296,
      "step": 33765
    },
    {
      "epoch": 592.4601769911504,
      "grad_norm": 1.7134745178282174e-07,
      "learning_rate": 2.934406825178576e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19182064,
      "step": 33770
    },
    {
      "epoch": 592.5486725663717,
      "grad_norm": 1.399624807163491e-06,
      "learning_rate": 2.9297935255170357e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19184848,
      "step": 33775
    },
    {
      "epoch": 592.637168141593,
      "grad_norm": 1.1234389774017473e-07,
      "learning_rate": 2.925183629358691e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19187488,
      "step": 33780
    },
    {
      "epoch": 592.7256637168142,
      "grad_norm": 1.9213017310448777e-07,
      "learning_rate": 2.9205771374144346e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19190832,
      "step": 33785
    },
    {
      "epoch": 592.8141592920354,
      "grad_norm": 1.8329134832129057e-07,
      "learning_rate": 2.915974050394657e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19193536,
      "step": 33790
    },
    {
      "epoch": 592.9026548672566,
      "grad_norm": 2.1037047304162115e-07,
      "learning_rate": 2.9113743690092067e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19196176,
      "step": 33795
    },
    {
      "epoch": 592.9911504424779,
      "grad_norm": 3.19986639851777e-07,
      "learning_rate": 2.906778093967402e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19199136,
      "step": 33800
    },
    {
      "epoch": 592.9911504424779,
      "eval_loss": 0.9060526490211487,
      "eval_runtime": 1.0631,
      "eval_samples_per_second": 23.516,
      "eval_steps_per_second": 12.228,
      "num_input_tokens_seen": 19199136,
      "step": 33800
    },
    {
      "epoch": 593.070796460177,
      "grad_norm": 2.4860392500158923e-07,
      "learning_rate": 2.9021852259780656e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19201480,
      "step": 33805
    },
    {
      "epoch": 593.1592920353983,
      "grad_norm": 5.185256100048719e-07,
      "learning_rate": 2.8975957657494583e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19204632,
      "step": 33810
    },
    {
      "epoch": 593.2477876106195,
      "grad_norm": 7.853183774386707e-07,
      "learning_rate": 2.8930097139893417e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19207288,
      "step": 33815
    },
    {
      "epoch": 593.3362831858407,
      "grad_norm": 2.1948251571757282e-07,
      "learning_rate": 2.888427071404945e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19210072,
      "step": 33820
    },
    {
      "epoch": 593.4247787610619,
      "grad_norm": 1.2975107210877468e-07,
      "learning_rate": 2.8838478387029606e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19212712,
      "step": 33825
    },
    {
      "epoch": 593.5132743362832,
      "grad_norm": 3.7456553059200814e-07,
      "learning_rate": 2.8792720165895737e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19215864,
      "step": 33830
    },
    {
      "epoch": 593.6017699115044,
      "grad_norm": 7.989905981276024e-08,
      "learning_rate": 2.874699605770423e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19218776,
      "step": 33835
    },
    {
      "epoch": 593.6902654867257,
      "grad_norm": 1.8388891476206481e-07,
      "learning_rate": 2.8701306069506383e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19221544,
      "step": 33840
    },
    {
      "epoch": 593.7787610619469,
      "grad_norm": 1.157796418738144e-06,
      "learning_rate": 2.8655650208348178e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19224920,
      "step": 33845
    },
    {
      "epoch": 593.8672566371681,
      "grad_norm": 1.0863941213301587e-07,
      "learning_rate": 2.8610028481270257e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19227400,
      "step": 33850
    },
    {
      "epoch": 593.9557522123894,
      "grad_norm": 2.764295459201094e-07,
      "learning_rate": 2.856444089530813e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19230360,
      "step": 33855
    },
    {
      "epoch": 594.0353982300885,
      "grad_norm": 1.1659165011224104e-06,
      "learning_rate": 2.8518887457491955e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19232576,
      "step": 33860
    },
    {
      "epoch": 594.1238938053098,
      "grad_norm": 2.3592609466049907e-07,
      "learning_rate": 2.8473368174846666e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19235184,
      "step": 33865
    },
    {
      "epoch": 594.212389380531,
      "grad_norm": 1.3475777222993202e-06,
      "learning_rate": 2.842788305439184e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19237568,
      "step": 33870
    },
    {
      "epoch": 594.3008849557522,
      "grad_norm": 9.124087796408276e-08,
      "learning_rate": 2.8382432103141925e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19240160,
      "step": 33875
    },
    {
      "epoch": 594.3893805309734,
      "grad_norm": 1.700249612213156e-07,
      "learning_rate": 2.833701532810598e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19242960,
      "step": 33880
    },
    {
      "epoch": 594.4778761061947,
      "grad_norm": 2.2404108790397004e-07,
      "learning_rate": 2.8291632736287877e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19245824,
      "step": 33885
    },
    {
      "epoch": 594.566371681416,
      "grad_norm": 7.259010459392812e-08,
      "learning_rate": 2.824628433468615e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19248544,
      "step": 33890
    },
    {
      "epoch": 594.6548672566372,
      "grad_norm": 1.8059975559481245e-07,
      "learning_rate": 2.8200970130294073e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19252112,
      "step": 33895
    },
    {
      "epoch": 594.7433628318585,
      "grad_norm": 7.193208602984669e-06,
      "learning_rate": 2.8155690130099775e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19255504,
      "step": 33900
    },
    {
      "epoch": 594.8318584070796,
      "grad_norm": 2.2184661929713911e-07,
      "learning_rate": 2.8110444341085895e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19258432,
      "step": 33905
    },
    {
      "epoch": 594.9203539823009,
      "grad_norm": 3.5151802535438037e-07,
      "learning_rate": 2.806523277022996e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19262000,
      "step": 33910
    },
    {
      "epoch": 595.0,
      "grad_norm": 3.358956348620268e-07,
      "learning_rate": 2.802005542450409e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19264208,
      "step": 33915
    },
    {
      "epoch": 595.0884955752213,
      "grad_norm": 1.965603217968237e-07,
      "learning_rate": 2.797491231087526e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19266768,
      "step": 33920
    },
    {
      "epoch": 595.1769911504425,
      "grad_norm": 2.6924024609797925e-07,
      "learning_rate": 2.7929803436305137e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19269536,
      "step": 33925
    },
    {
      "epoch": 595.2654867256637,
      "grad_norm": 1.1485263229360498e-07,
      "learning_rate": 2.788472880774998e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19272336,
      "step": 33930
    },
    {
      "epoch": 595.3539823008849,
      "grad_norm": 2.600687594167539e-07,
      "learning_rate": 2.7839688432160977e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19275168,
      "step": 33935
    },
    {
      "epoch": 595.4424778761062,
      "grad_norm": 1.6372329980640643e-07,
      "learning_rate": 2.779468231648383e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19278048,
      "step": 33940
    },
    {
      "epoch": 595.5309734513274,
      "grad_norm": 2.416441020614002e-07,
      "learning_rate": 2.774971046765906e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19280864,
      "step": 33945
    },
    {
      "epoch": 595.6194690265487,
      "grad_norm": 3.7198603877186542e-06,
      "learning_rate": 2.770477289262194e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19283920,
      "step": 33950
    },
    {
      "epoch": 595.70796460177,
      "grad_norm": 2.0736844419388945e-07,
      "learning_rate": 2.765986959830233e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19287024,
      "step": 33955
    },
    {
      "epoch": 595.7964601769911,
      "grad_norm": 1.483507503508008e-07,
      "learning_rate": 2.761500059162492e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19289616,
      "step": 33960
    },
    {
      "epoch": 595.8849557522124,
      "grad_norm": 1.293070539531982e-07,
      "learning_rate": 2.757016587950914e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19292528,
      "step": 33965
    },
    {
      "epoch": 595.9734513274336,
      "grad_norm": 2.75030174634594e-07,
      "learning_rate": 2.752536546886897e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19295648,
      "step": 33970
    },
    {
      "epoch": 596.0530973451328,
      "grad_norm": 1.259570154843459e-07,
      "learning_rate": 2.7480599366613234e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19297952,
      "step": 33975
    },
    {
      "epoch": 596.141592920354,
      "grad_norm": 1.9343772805768822e-07,
      "learning_rate": 2.7435867579645473e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19300784,
      "step": 33980
    },
    {
      "epoch": 596.2300884955753,
      "grad_norm": 2.2314266345802025e-07,
      "learning_rate": 2.739117011486378e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19303312,
      "step": 33985
    },
    {
      "epoch": 596.3185840707964,
      "grad_norm": 6.768777893739752e-08,
      "learning_rate": 2.7346506979161216e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19306544,
      "step": 33990
    },
    {
      "epoch": 596.4070796460177,
      "grad_norm": 2.523427156120306e-07,
      "learning_rate": 2.7301878179425227e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19308912,
      "step": 33995
    },
    {
      "epoch": 596.4955752212389,
      "grad_norm": 5.561665261666349e-07,
      "learning_rate": 2.7257283722538244e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19311344,
      "step": 34000
    },
    {
      "epoch": 596.4955752212389,
      "eval_loss": 0.8945153951644897,
      "eval_runtime": 1.0716,
      "eval_samples_per_second": 23.33,
      "eval_steps_per_second": 12.132,
      "num_input_tokens_seen": 19311344,
      "step": 34000
    },
    {
      "epoch": 596.5840707964602,
      "grad_norm": 1.7664319784671534e-07,
      "learning_rate": 2.7212723615377326e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19313920,
      "step": 34005
    },
    {
      "epoch": 596.6725663716815,
      "grad_norm": 8.651538223602984e-07,
      "learning_rate": 2.7168197864814145e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19317408,
      "step": 34010
    },
    {
      "epoch": 596.7610619469026,
      "grad_norm": 2.912424292844662e-07,
      "learning_rate": 2.712370647771509e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19320352,
      "step": 34015
    },
    {
      "epoch": 596.8495575221239,
      "grad_norm": 1.1182503101281327e-07,
      "learning_rate": 2.707924946094137e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19323360,
      "step": 34020
    },
    {
      "epoch": 596.9380530973451,
      "grad_norm": 1.9817979080016812e-07,
      "learning_rate": 2.7034826821348723e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19326640,
      "step": 34025
    },
    {
      "epoch": 597.0176991150443,
      "grad_norm": 1.3007271491005667e-06,
      "learning_rate": 2.6990438565787786e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19328992,
      "step": 34030
    },
    {
      "epoch": 597.1061946902655,
      "grad_norm": 4.282685040379874e-06,
      "learning_rate": 2.6946084701103714e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19331680,
      "step": 34035
    },
    {
      "epoch": 597.1946902654868,
      "grad_norm": 2.656299784575822e-07,
      "learning_rate": 2.6901765234136428e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19334352,
      "step": 34040
    },
    {
      "epoch": 597.2831858407079,
      "grad_norm": 4.287186357032624e-07,
      "learning_rate": 2.685748017172063e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19337600,
      "step": 34045
    },
    {
      "epoch": 597.3716814159292,
      "grad_norm": 2.527561093756958e-07,
      "learning_rate": 2.681322952068549e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19340288,
      "step": 34050
    },
    {
      "epoch": 597.4601769911504,
      "grad_norm": 1.4483212851246208e-07,
      "learning_rate": 2.6769013287855137e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19343552,
      "step": 34055
    },
    {
      "epoch": 597.5486725663717,
      "grad_norm": 1.3328448744687194e-07,
      "learning_rate": 2.6724831480048286e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19346272,
      "step": 34060
    },
    {
      "epoch": 597.637168141593,
      "grad_norm": 2.1534199845518742e-07,
      "learning_rate": 2.66806841040782e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19349104,
      "step": 34065
    },
    {
      "epoch": 597.7256637168142,
      "grad_norm": 1.3851790470198466e-07,
      "learning_rate": 2.6636571166753083e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19352032,
      "step": 34070
    },
    {
      "epoch": 597.8141592920354,
      "grad_norm": 1.0259770277798452e-07,
      "learning_rate": 2.6592492674875598e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19354688,
      "step": 34075
    },
    {
      "epoch": 597.9026548672566,
      "grad_norm": 1.1392372698537656e-06,
      "learning_rate": 2.6548448635243305e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19357392,
      "step": 34080
    },
    {
      "epoch": 597.9911504424779,
      "grad_norm": 1.1407626487880407e-07,
      "learning_rate": 2.650443905464828e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19360480,
      "step": 34085
    },
    {
      "epoch": 598.070796460177,
      "grad_norm": 1.2455579678771755e-07,
      "learning_rate": 2.646046393987739e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19362624,
      "step": 34090
    },
    {
      "epoch": 598.1592920353983,
      "grad_norm": 3.51911438656316e-07,
      "learning_rate": 2.64165232977121e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19365200,
      "step": 34095
    },
    {
      "epoch": 598.2477876106195,
      "grad_norm": 2.753582748482586e-06,
      "learning_rate": 2.6372617134928695e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19368704,
      "step": 34100
    },
    {
      "epoch": 598.3362831858407,
      "grad_norm": 2.1719182541346527e-07,
      "learning_rate": 2.6328745458297943e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19371456,
      "step": 34105
    },
    {
      "epoch": 598.4247787610619,
      "grad_norm": 3.9594519307684095e-07,
      "learning_rate": 2.6284908274585546e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19374720,
      "step": 34110
    },
    {
      "epoch": 598.5132743362832,
      "grad_norm": 2.5517641688566073e-07,
      "learning_rate": 2.6241105590551595e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19377728,
      "step": 34115
    },
    {
      "epoch": 598.6017699115044,
      "grad_norm": 2.0537217437777144e-07,
      "learning_rate": 2.6197337412951105e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19380768,
      "step": 34120
    },
    {
      "epoch": 598.6902654867257,
      "grad_norm": 2.479042962022504e-07,
      "learning_rate": 2.6153603748533705e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19383360,
      "step": 34125
    },
    {
      "epoch": 598.7787610619469,
      "grad_norm": 4.902178147858649e-07,
      "learning_rate": 2.6109904604043585e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19386320,
      "step": 34130
    },
    {
      "epoch": 598.8672566371681,
      "grad_norm": 1.1886924767168239e-05,
      "learning_rate": 2.6066239986219765e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19389280,
      "step": 34135
    },
    {
      "epoch": 598.9557522123894,
      "grad_norm": 4.3070911488030106e-05,
      "learning_rate": 2.602260990179592e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19392048,
      "step": 34140
    },
    {
      "epoch": 599.0353982300885,
      "grad_norm": 2.9435116744025436e-07,
      "learning_rate": 2.5979014357500248e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19394200,
      "step": 34145
    },
    {
      "epoch": 599.1238938053098,
      "grad_norm": 2.3243514135629084e-07,
      "learning_rate": 2.5935453360055844e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19396904,
      "step": 34150
    },
    {
      "epoch": 599.212389380531,
      "grad_norm": 1.895810868290937e-07,
      "learning_rate": 2.5891926916180283e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19399688,
      "step": 34155
    },
    {
      "epoch": 599.3008849557522,
      "grad_norm": 1.1687883727518056e-07,
      "learning_rate": 2.5848435032585883e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19402568,
      "step": 34160
    },
    {
      "epoch": 599.3893805309734,
      "grad_norm": 9.107460385848754e-08,
      "learning_rate": 2.58049777159797e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19405336,
      "step": 34165
    },
    {
      "epoch": 599.4778761061947,
      "grad_norm": 7.254139291035244e-06,
      "learning_rate": 2.576155497306332e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19408136,
      "step": 34170
    },
    {
      "epoch": 599.566371681416,
      "grad_norm": 2.0671355116519408e-07,
      "learning_rate": 2.57181668105331e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19410840,
      "step": 34175
    },
    {
      "epoch": 599.6548672566372,
      "grad_norm": 1.8207924767921213e-06,
      "learning_rate": 2.567481323508014e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19413720,
      "step": 34180
    },
    {
      "epoch": 599.7433628318585,
      "grad_norm": 4.714585770670965e-07,
      "learning_rate": 2.5631494253389954e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19416952,
      "step": 34185
    },
    {
      "epoch": 599.8318584070796,
      "grad_norm": 1.9583498556130507e-07,
      "learning_rate": 2.5588209872142997e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19419912,
      "step": 34190
    },
    {
      "epoch": 599.9203539823009,
      "grad_norm": 1.240030798044245e-07,
      "learning_rate": 2.5544960098014186e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19423032,
      "step": 34195
    },
    {
      "epoch": 600.0,
      "grad_norm": 5.480513465272452e-08,
      "learning_rate": 2.550174493767318e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19425472,
      "step": 34200
    },
    {
      "epoch": 600.0,
      "eval_loss": 0.897942304611206,
      "eval_runtime": 1.065,
      "eval_samples_per_second": 23.474,
      "eval_steps_per_second": 12.207,
      "num_input_tokens_seen": 19425472,
      "step": 34200
    },
    {
      "epoch": 600.0884955752213,
      "grad_norm": 2.7440066787676187e-06,
      "learning_rate": 2.545856439778438e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19428848,
      "step": 34205
    },
    {
      "epoch": 600.1769911504425,
      "grad_norm": 8.234206347879081e-07,
      "learning_rate": 2.541541848500667e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19432096,
      "step": 34210
    },
    {
      "epoch": 600.2654867256637,
      "grad_norm": 1.963293243534281e-07,
      "learning_rate": 2.5372307205993733e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19434656,
      "step": 34215
    },
    {
      "epoch": 600.3539823008849,
      "grad_norm": 3.8373516986212053e-07,
      "learning_rate": 2.5329230567393917e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19437280,
      "step": 34220
    },
    {
      "epoch": 600.4424778761062,
      "grad_norm": 3.9717576783004915e-07,
      "learning_rate": 2.5286188575850164e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19440832,
      "step": 34225
    },
    {
      "epoch": 600.5309734513274,
      "grad_norm": 4.569035354506923e-06,
      "learning_rate": 2.5243181237999984e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19443712,
      "step": 34230
    },
    {
      "epoch": 600.6194690265487,
      "grad_norm": 2.125335640812409e-07,
      "learning_rate": 2.520020856047578e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19446624,
      "step": 34235
    },
    {
      "epoch": 600.70796460177,
      "grad_norm": 4.354102145498473e-07,
      "learning_rate": 2.515727054990438e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19449424,
      "step": 34240
    },
    {
      "epoch": 600.7964601769911,
      "grad_norm": 2.1765941937701427e-07,
      "learning_rate": 2.511436721290747e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19452176,
      "step": 34245
    },
    {
      "epoch": 600.8849557522124,
      "grad_norm": 1.8498535325761623e-07,
      "learning_rate": 2.5071498556101164e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19454512,
      "step": 34250
    },
    {
      "epoch": 600.9734513274336,
      "grad_norm": 1.436885952443845e-07,
      "learning_rate": 2.5028664586096485e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19457248,
      "step": 34255
    },
    {
      "epoch": 601.0530973451328,
      "grad_norm": 9.350559793119828e-08,
      "learning_rate": 2.498586530949881e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19459592,
      "step": 34260
    },
    {
      "epoch": 601.141592920354,
      "grad_norm": 1.0840744124607227e-07,
      "learning_rate": 2.4943100732908427e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19462248,
      "step": 34265
    },
    {
      "epoch": 601.2300884955753,
      "grad_norm": 1.5388400242954958e-07,
      "learning_rate": 2.4900370862920188e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19465224,
      "step": 34270
    },
    {
      "epoch": 601.3185840707964,
      "grad_norm": 2.823421141329163e-07,
      "learning_rate": 2.4857675706123518e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19468040,
      "step": 34275
    },
    {
      "epoch": 601.4070796460177,
      "grad_norm": 1.6009967396257707e-07,
      "learning_rate": 2.4815015269102543e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19470920,
      "step": 34280
    },
    {
      "epoch": 601.4955752212389,
      "grad_norm": 1.0708864550679209e-07,
      "learning_rate": 2.477238955843611e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19473928,
      "step": 34285
    },
    {
      "epoch": 601.5840707964602,
      "grad_norm": 2.5467659270361764e-07,
      "learning_rate": 2.4729798580697573e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19477480,
      "step": 34290
    },
    {
      "epoch": 601.6725663716815,
      "grad_norm": 2.1369145031258086e-07,
      "learning_rate": 2.4687242342455034e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19479848,
      "step": 34295
    },
    {
      "epoch": 601.7610619469026,
      "grad_norm": 7.014772904767597e-07,
      "learning_rate": 2.4644720850271196e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19482648,
      "step": 34300
    },
    {
      "epoch": 601.8495575221239,
      "grad_norm": 1.4536490766658972e-07,
      "learning_rate": 2.4602234110703364e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19485432,
      "step": 34305
    },
    {
      "epoch": 601.9380530973451,
      "grad_norm": 3.6017425486534194e-07,
      "learning_rate": 2.4559782130303576e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19488392,
      "step": 34310
    },
    {
      "epoch": 602.0176991150443,
      "grad_norm": 1.3620458503282862e-06,
      "learning_rate": 2.451736491561843e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19490824,
      "step": 34315
    },
    {
      "epoch": 602.1061946902655,
      "grad_norm": 5.65203947644477e-07,
      "learning_rate": 2.4474982473189163e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19493352,
      "step": 34320
    },
    {
      "epoch": 602.1946902654868,
      "grad_norm": 1.4135279968741088e-07,
      "learning_rate": 2.4432634809551796e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19496392,
      "step": 34325
    },
    {
      "epoch": 602.2831858407079,
      "grad_norm": 8.453847613054677e-07,
      "learning_rate": 2.439032193123675e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19499208,
      "step": 34330
    },
    {
      "epoch": 602.3716814159292,
      "grad_norm": 7.525930101337508e-08,
      "learning_rate": 2.4348043844769297e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19502056,
      "step": 34335
    },
    {
      "epoch": 602.4601769911504,
      "grad_norm": 9.380091228194942e-07,
      "learning_rate": 2.4305800556669146e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19504536,
      "step": 34340
    },
    {
      "epoch": 602.5486725663717,
      "grad_norm": 5.063511707703583e-07,
      "learning_rate": 2.426359207345083e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19507144,
      "step": 34345
    },
    {
      "epoch": 602.637168141593,
      "grad_norm": 2.323129137948854e-07,
      "learning_rate": 2.4221418401623396e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19510280,
      "step": 34350
    },
    {
      "epoch": 602.7256637168142,
      "grad_norm": 2.2821393486083252e-07,
      "learning_rate": 2.4179279547690557e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19513240,
      "step": 34355
    },
    {
      "epoch": 602.8141592920354,
      "grad_norm": 1.3870408110960852e-07,
      "learning_rate": 2.413717551815062e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19516184,
      "step": 34360
    },
    {
      "epoch": 602.9026548672566,
      "grad_norm": 1.616557767647464e-07,
      "learning_rate": 2.409510631949666e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19519640,
      "step": 34365
    },
    {
      "epoch": 602.9911504424779,
      "grad_norm": 1.393820411976776e-06,
      "learning_rate": 2.405307195821618e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19522104,
      "step": 34370
    },
    {
      "epoch": 603.070796460177,
      "grad_norm": 6.971429229452042e-07,
      "learning_rate": 2.4011072440791372e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19524120,
      "step": 34375
    },
    {
      "epoch": 603.1592920353983,
      "grad_norm": 1.0819370999115563e-07,
      "learning_rate": 2.3969107773699233e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19526824,
      "step": 34380
    },
    {
      "epoch": 603.2477876106195,
      "grad_norm": 1.5274677878096554e-07,
      "learning_rate": 2.3927177963411096e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19529816,
      "step": 34385
    },
    {
      "epoch": 603.3362831858407,
      "grad_norm": 1.4888210841945693e-07,
      "learning_rate": 2.3885283016393144e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19532840,
      "step": 34390
    },
    {
      "epoch": 603.4247787610619,
      "grad_norm": 1.0080423606950717e-07,
      "learning_rate": 2.3843422939106076e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19536040,
      "step": 34395
    },
    {
      "epoch": 603.5132743362832,
      "grad_norm": 1.4298943540325126e-07,
      "learning_rate": 2.380159773800525e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19539112,
      "step": 34400
    },
    {
      "epoch": 603.5132743362832,
      "eval_loss": 0.9249515533447266,
      "eval_runtime": 1.0671,
      "eval_samples_per_second": 23.429,
      "eval_steps_per_second": 12.183,
      "num_input_tokens_seen": 19539112,
      "step": 34400
    },
    {
      "epoch": 603.6017699115044,
      "grad_norm": 9.928406541348522e-08,
      "learning_rate": 2.3759807419540675e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19541960,
      "step": 34405
    },
    {
      "epoch": 603.6902654867257,
      "grad_norm": 1.1091515261796303e-05,
      "learning_rate": 2.3718051990156835e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19544568,
      "step": 34410
    },
    {
      "epoch": 603.7787610619469,
      "grad_norm": 1.9755511004859727e-07,
      "learning_rate": 2.367633145629311e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19547272,
      "step": 34415
    },
    {
      "epoch": 603.8672566371681,
      "grad_norm": 1.3344929072900413e-07,
      "learning_rate": 2.363464582438316e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19550296,
      "step": 34420
    },
    {
      "epoch": 603.9557522123894,
      "grad_norm": 1.252870902135328e-06,
      "learning_rate": 2.3592995100855526e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19553336,
      "step": 34425
    },
    {
      "epoch": 604.0353982300885,
      "grad_norm": 1.4972812323321705e-07,
      "learning_rate": 2.3551379292133273e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19555984,
      "step": 34430
    },
    {
      "epoch": 604.1238938053098,
      "grad_norm": 2.2943304145428556e-07,
      "learning_rate": 2.3509798404634047e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19559088,
      "step": 34435
    },
    {
      "epoch": 604.212389380531,
      "grad_norm": 1.928644906001864e-07,
      "learning_rate": 2.346825244477019e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19562400,
      "step": 34440
    },
    {
      "epoch": 604.3008849557522,
      "grad_norm": 1.5789980523095437e-07,
      "learning_rate": 2.3426741418948545e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19565104,
      "step": 34445
    },
    {
      "epoch": 604.3893805309734,
      "grad_norm": 1.5724333479738561e-06,
      "learning_rate": 2.3385265333570715e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19567440,
      "step": 34450
    },
    {
      "epoch": 604.4778761061947,
      "grad_norm": 1.1286286394351919e-07,
      "learning_rate": 2.334382419503278e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19570256,
      "step": 34455
    },
    {
      "epoch": 604.566371681416,
      "grad_norm": 1.115302325160883e-06,
      "learning_rate": 2.3302418009725465e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19572928,
      "step": 34460
    },
    {
      "epoch": 604.6548672566372,
      "grad_norm": 3.9290017639359576e-07,
      "learning_rate": 2.326104678403415e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19575920,
      "step": 34465
    },
    {
      "epoch": 604.7433628318585,
      "grad_norm": 2.722108263242262e-07,
      "learning_rate": 2.321971052433883e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19579440,
      "step": 34470
    },
    {
      "epoch": 604.8318584070796,
      "grad_norm": 2.1022585769969737e-06,
      "learning_rate": 2.3178409237014004e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19581968,
      "step": 34475
    },
    {
      "epoch": 604.9203539823009,
      "grad_norm": 1.5389380791930307e-07,
      "learning_rate": 2.313714292842889e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19584656,
      "step": 34480
    },
    {
      "epoch": 605.0,
      "grad_norm": 1.1710324798741567e-07,
      "learning_rate": 2.309591160494734e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19587152,
      "step": 34485
    },
    {
      "epoch": 605.0884955752213,
      "grad_norm": 1.948044854316322e-07,
      "learning_rate": 2.305471527292763e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19589984,
      "step": 34490
    },
    {
      "epoch": 605.1769911504425,
      "grad_norm": 8.199060630431632e-08,
      "learning_rate": 2.3013553938722817e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19592784,
      "step": 34495
    },
    {
      "epoch": 605.2654867256637,
      "grad_norm": 6.565578587469645e-07,
      "learning_rate": 2.297242760868043e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19595488,
      "step": 34500
    },
    {
      "epoch": 605.3539823008849,
      "grad_norm": 4.677811773490248e-07,
      "learning_rate": 2.2931336289142735e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19597968,
      "step": 34505
    },
    {
      "epoch": 605.4424778761062,
      "grad_norm": 1.3548289246045897e-07,
      "learning_rate": 2.289027998644655e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19600800,
      "step": 34510
    },
    {
      "epoch": 605.5309734513274,
      "grad_norm": 1.4845564066945371e-07,
      "learning_rate": 2.2849258706923228e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19604080,
      "step": 34515
    },
    {
      "epoch": 605.6194690265487,
      "grad_norm": 1.9634940429114067e-07,
      "learning_rate": 2.2808272456898705e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19607456,
      "step": 34520
    },
    {
      "epoch": 605.70796460177,
      "grad_norm": 2.1214032130956184e-07,
      "learning_rate": 2.2767321242693707e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19610304,
      "step": 34525
    },
    {
      "epoch": 605.7964601769911,
      "grad_norm": 2.582256115601922e-07,
      "learning_rate": 2.272640507062329e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19612976,
      "step": 34530
    },
    {
      "epoch": 605.8849557522124,
      "grad_norm": 1.239920663920202e-07,
      "learning_rate": 2.2685523946997382e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19616224,
      "step": 34535
    },
    {
      "epoch": 605.9734513274336,
      "grad_norm": 2.3655124437027553e-07,
      "learning_rate": 2.2644677878120245e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19618912,
      "step": 34540
    },
    {
      "epoch": 606.0530973451328,
      "grad_norm": 8.626215617368871e-07,
      "learning_rate": 2.2603866870290897e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19621232,
      "step": 34545
    },
    {
      "epoch": 606.141592920354,
      "grad_norm": 2.684761852833617e-07,
      "learning_rate": 2.256309092980294e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19623680,
      "step": 34550
    },
    {
      "epoch": 606.2300884955753,
      "grad_norm": 9.037692905167205e-08,
      "learning_rate": 2.252235006294448e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19626768,
      "step": 34555
    },
    {
      "epoch": 606.3185840707964,
      "grad_norm": 9.608640993974404e-08,
      "learning_rate": 2.2481644275998333e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19630224,
      "step": 34560
    },
    {
      "epoch": 606.4070796460177,
      "grad_norm": 3.8381708122869895e-07,
      "learning_rate": 2.2440973575241832e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19633392,
      "step": 34565
    },
    {
      "epoch": 606.4955752212389,
      "grad_norm": 3.252619933391543e-07,
      "learning_rate": 2.240033796694685e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19635872,
      "step": 34570
    },
    {
      "epoch": 606.5840707964602,
      "grad_norm": 5.9243919992013616e-08,
      "learning_rate": 2.235973745737999e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19638784,
      "step": 34575
    },
    {
      "epoch": 606.6725663716815,
      "grad_norm": 5.084175427327864e-07,
      "learning_rate": 2.2319172052802263e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19641376,
      "step": 34580
    },
    {
      "epoch": 606.7610619469026,
      "grad_norm": 1.7610223324027174e-07,
      "learning_rate": 2.2278641759469477e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19644176,
      "step": 34585
    },
    {
      "epoch": 606.8495575221239,
      "grad_norm": 3.338927285767568e-07,
      "learning_rate": 2.2238146583631825e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19647680,
      "step": 34590
    },
    {
      "epoch": 606.9380530973451,
      "grad_norm": 1.7360339654715062e-07,
      "learning_rate": 2.2197686531534256e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19650336,
      "step": 34595
    },
    {
      "epoch": 607.0176991150443,
      "grad_norm": 3.684982630147715e-07,
      "learning_rate": 2.2157261609416087e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19652392,
      "step": 34600
    },
    {
      "epoch": 607.0176991150443,
      "eval_loss": 0.9027122855186462,
      "eval_runtime": 1.0674,
      "eval_samples_per_second": 23.422,
      "eval_steps_per_second": 12.179,
      "num_input_tokens_seen": 19652392,
      "step": 34600
    },
    {
      "epoch": 607.1061946902655,
      "grad_norm": 1.1709221325872932e-07,
      "learning_rate": 2.211687182351149e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19654904,
      "step": 34605
    },
    {
      "epoch": 607.1946902654868,
      "grad_norm": 1.2774515312230506e-07,
      "learning_rate": 2.2076517180048993e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19657784,
      "step": 34610
    },
    {
      "epoch": 607.2831858407079,
      "grad_norm": 3.370805075064709e-07,
      "learning_rate": 2.2036197685251834e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19660856,
      "step": 34615
    },
    {
      "epoch": 607.3716814159292,
      "grad_norm": 1.3616266869576066e-07,
      "learning_rate": 2.199591334533771e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19663528,
      "step": 34620
    },
    {
      "epoch": 607.4601769911504,
      "grad_norm": 3.2764160096121486e-07,
      "learning_rate": 2.1955664166519036e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19666168,
      "step": 34625
    },
    {
      "epoch": 607.5486725663717,
      "grad_norm": 1.334509988737409e-06,
      "learning_rate": 2.1915450155002793e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19669096,
      "step": 34630
    },
    {
      "epoch": 607.637168141593,
      "grad_norm": 1.4400632153410697e-07,
      "learning_rate": 2.187527131699038e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19672424,
      "step": 34635
    },
    {
      "epoch": 607.7256637168142,
      "grad_norm": 2.531120912863116e-07,
      "learning_rate": 2.18351276586779e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19675128,
      "step": 34640
    },
    {
      "epoch": 607.8141592920354,
      "grad_norm": 2.1803047900448291e-07,
      "learning_rate": 2.1795019186256092e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19677944,
      "step": 34645
    },
    {
      "epoch": 607.9026548672566,
      "grad_norm": 1.1061084848051905e-07,
      "learning_rate": 2.1754945905910094e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19681160,
      "step": 34650
    },
    {
      "epoch": 607.9911504424779,
      "grad_norm": 1.7455681700084824e-07,
      "learning_rate": 2.171490782381977e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19684296,
      "step": 34655
    },
    {
      "epoch": 608.070796460177,
      "grad_norm": 3.335392193548614e-07,
      "learning_rate": 2.1674904946159425e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19686560,
      "step": 34660
    },
    {
      "epoch": 608.1592920353983,
      "grad_norm": 9.509388121387019e-08,
      "learning_rate": 2.16349372790981e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19689328,
      "step": 34665
    },
    {
      "epoch": 608.2477876106195,
      "grad_norm": 1.728694627445293e-07,
      "learning_rate": 2.159500482879928e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19692768,
      "step": 34670
    },
    {
      "epoch": 608.3362831858407,
      "grad_norm": 3.373763775016414e-07,
      "learning_rate": 2.155510760142096e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19695840,
      "step": 34675
    },
    {
      "epoch": 608.4247787610619,
      "grad_norm": 8.387191883230116e-07,
      "learning_rate": 2.151524560311588e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19699120,
      "step": 34680
    },
    {
      "epoch": 608.5132743362832,
      "grad_norm": 7.859359385520293e-08,
      "learning_rate": 2.147541884003129e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19701616,
      "step": 34685
    },
    {
      "epoch": 608.6017699115044,
      "grad_norm": 1.0702778752147424e-07,
      "learning_rate": 2.1435627318308895e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19704512,
      "step": 34690
    },
    {
      "epoch": 608.6902654867257,
      "grad_norm": 4.791701826434291e-07,
      "learning_rate": 2.139587104408511e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19706896,
      "step": 34695
    },
    {
      "epoch": 608.7787610619469,
      "grad_norm": 1.2239901536759135e-07,
      "learning_rate": 2.1356150023490783e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19709456,
      "step": 34700
    },
    {
      "epoch": 608.8672566371681,
      "grad_norm": 1.4004490367369726e-07,
      "learning_rate": 2.1316464262651464e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19712480,
      "step": 34705
    },
    {
      "epoch": 608.9557522123894,
      "grad_norm": 8.515398519648443e-08,
      "learning_rate": 2.1276813767687224e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19715312,
      "step": 34710
    },
    {
      "epoch": 609.0353982300885,
      "grad_norm": 1.7905530569350958e-07,
      "learning_rate": 2.123719854471254e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19717576,
      "step": 34715
    },
    {
      "epoch": 609.1238938053098,
      "grad_norm": 2.1925987425674975e-07,
      "learning_rate": 2.119761859983668e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19720744,
      "step": 34720
    },
    {
      "epoch": 609.212389380531,
      "grad_norm": 1.4087731869949494e-07,
      "learning_rate": 2.1158073939163386e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19723784,
      "step": 34725
    },
    {
      "epoch": 609.3008849557522,
      "grad_norm": 2.6980129064213543e-07,
      "learning_rate": 2.111856456879088e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19726920,
      "step": 34730
    },
    {
      "epoch": 609.3893805309734,
      "grad_norm": 7.732626983170121e-08,
      "learning_rate": 2.1079090494811993e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19730088,
      "step": 34735
    },
    {
      "epoch": 609.4778761061947,
      "grad_norm": 2.4664009856678604e-07,
      "learning_rate": 2.103965172331418e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19733432,
      "step": 34740
    },
    {
      "epoch": 609.566371681416,
      "grad_norm": 2.0845718040618522e-07,
      "learning_rate": 2.100024826037933e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19736040,
      "step": 34745
    },
    {
      "epoch": 609.6548672566372,
      "grad_norm": 1.6668072078118712e-07,
      "learning_rate": 2.0960880112084027e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19738776,
      "step": 34750
    },
    {
      "epoch": 609.7433628318585,
      "grad_norm": 1.229045892614522e-06,
      "learning_rate": 2.092154728449927e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19741384,
      "step": 34755
    },
    {
      "epoch": 609.8318584070796,
      "grad_norm": 9.935174460906637e-08,
      "learning_rate": 2.0882249783690687e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19744232,
      "step": 34760
    },
    {
      "epoch": 609.9203539823009,
      "grad_norm": 2.559678762281692e-07,
      "learning_rate": 2.084298761571851e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19746936,
      "step": 34765
    },
    {
      "epoch": 610.0,
      "grad_norm": 7.626579190400662e-08,
      "learning_rate": 2.080376078663737e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19749320,
      "step": 34770
    },
    {
      "epoch": 610.0884955752213,
      "grad_norm": 3.3424166190343385e-07,
      "learning_rate": 2.0764569302496593e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19752776,
      "step": 34775
    },
    {
      "epoch": 610.1769911504425,
      "grad_norm": 4.4226646878087195e-07,
      "learning_rate": 2.0725413169339957e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19755656,
      "step": 34780
    },
    {
      "epoch": 610.2654867256637,
      "grad_norm": 3.6160611216473626e-07,
      "learning_rate": 2.068629239320588e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19758344,
      "step": 34785
    },
    {
      "epoch": 610.3539823008849,
      "grad_norm": 1.9389125327506918e-07,
      "learning_rate": 2.064720698012726e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19761240,
      "step": 34790
    },
    {
      "epoch": 610.4424778761062,
      "grad_norm": 3.602206959385512e-07,
      "learning_rate": 2.0608156936131522e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19764248,
      "step": 34795
    },
    {
      "epoch": 610.5309734513274,
      "grad_norm": 2.488406778411445e-07,
      "learning_rate": 2.056914226724074e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19766904,
      "step": 34800
    },
    {
      "epoch": 610.5309734513274,
      "eval_loss": 0.9086984395980835,
      "eval_runtime": 1.0694,
      "eval_samples_per_second": 23.378,
      "eval_steps_per_second": 12.157,
      "num_input_tokens_seen": 19766904,
      "step": 34800
    },
    {
      "epoch": 610.6194690265487,
      "grad_norm": 1.7723748442222131e-06,
      "learning_rate": 2.0530162979471385e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19769528,
      "step": 34805
    },
    {
      "epoch": 610.70796460177,
      "grad_norm": 2.886319805384119e-07,
      "learning_rate": 2.0491219078834667e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19772296,
      "step": 34810
    },
    {
      "epoch": 610.7964601769911,
      "grad_norm": 1.7865754387003108e-07,
      "learning_rate": 2.045231057133612e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19775096,
      "step": 34815
    },
    {
      "epoch": 610.8849557522124,
      "grad_norm": 3.7526496043938096e-07,
      "learning_rate": 2.0413437462975944e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19777944,
      "step": 34820
    },
    {
      "epoch": 610.9734513274336,
      "grad_norm": 1.3040053659096884e-07,
      "learning_rate": 2.0374599759748843e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19780600,
      "step": 34825
    },
    {
      "epoch": 611.0530973451328,
      "grad_norm": 1.1248619102843804e-06,
      "learning_rate": 2.033579746764419e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19783248,
      "step": 34830
    },
    {
      "epoch": 611.141592920354,
      "grad_norm": 8.021039121786089e-08,
      "learning_rate": 2.029703059264565e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19786496,
      "step": 34835
    },
    {
      "epoch": 611.2300884955753,
      "grad_norm": 8.162937206179777e-07,
      "learning_rate": 2.02582991407316e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19789584,
      "step": 34840
    },
    {
      "epoch": 611.3185840707964,
      "grad_norm": 9.648250198779351e-08,
      "learning_rate": 2.0219603117874992e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19792288,
      "step": 34845
    },
    {
      "epoch": 611.4070796460177,
      "grad_norm": 1.478193667026062e-06,
      "learning_rate": 2.0180942530043156e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19794864,
      "step": 34850
    },
    {
      "epoch": 611.4955752212389,
      "grad_norm": 1.2481901023875253e-07,
      "learning_rate": 2.0142317383198107e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19797264,
      "step": 34855
    },
    {
      "epoch": 611.5840707964602,
      "grad_norm": 1.0887777790458131e-07,
      "learning_rate": 2.0103727683296243e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19800448,
      "step": 34860
    },
    {
      "epoch": 611.6725663716815,
      "grad_norm": 2.2692020706926996e-07,
      "learning_rate": 2.0065173436288636e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19803584,
      "step": 34865
    },
    {
      "epoch": 611.7610619469026,
      "grad_norm": 1.6508440126017376e-07,
      "learning_rate": 2.002665464812087e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19806336,
      "step": 34870
    },
    {
      "epoch": 611.8495575221239,
      "grad_norm": 2.2462002391421265e-07,
      "learning_rate": 1.998817132473291e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19809456,
      "step": 34875
    },
    {
      "epoch": 611.9380530973451,
      "grad_norm": 6.400614438462071e-08,
      "learning_rate": 1.9949723472059507e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19812656,
      "step": 34880
    },
    {
      "epoch": 612.0176991150443,
      "grad_norm": 3.1515656928604585e-07,
      "learning_rate": 1.9911311096029726e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19814672,
      "step": 34885
    },
    {
      "epoch": 612.1061946902655,
      "grad_norm": 2.1532419225422927e-07,
      "learning_rate": 1.9872934202567224e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19817376,
      "step": 34890
    },
    {
      "epoch": 612.1946902654868,
      "grad_norm": 3.9784899286132713e-07,
      "learning_rate": 1.9834592797590257e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19820384,
      "step": 34895
    },
    {
      "epoch": 612.2831858407079,
      "grad_norm": 1.8108872268385312e-07,
      "learning_rate": 1.979628688701149e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19822992,
      "step": 34900
    },
    {
      "epoch": 612.3716814159292,
      "grad_norm": 1.3845300372850033e-07,
      "learning_rate": 1.9758016476738193e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19825552,
      "step": 34905
    },
    {
      "epoch": 612.4601769911504,
      "grad_norm": 5.777013143415388e-07,
      "learning_rate": 1.971978157267221e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19828320,
      "step": 34910
    },
    {
      "epoch": 612.5486725663717,
      "grad_norm": 1.339676828138181e-06,
      "learning_rate": 1.968158218070973e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19832016,
      "step": 34915
    },
    {
      "epoch": 612.637168141593,
      "grad_norm": 3.5722982261177094e-07,
      "learning_rate": 1.9643418306741682e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19835632,
      "step": 34920
    },
    {
      "epoch": 612.7256637168142,
      "grad_norm": 1.4699201074108714e-06,
      "learning_rate": 1.9605289956653337e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19838384,
      "step": 34925
    },
    {
      "epoch": 612.8141592920354,
      "grad_norm": 1.1729643745184148e-07,
      "learning_rate": 1.9567197136324626e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19840784,
      "step": 34930
    },
    {
      "epoch": 612.9026548672566,
      "grad_norm": 1.2557774198285188e-07,
      "learning_rate": 1.9529139851629935e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19843472,
      "step": 34935
    },
    {
      "epoch": 612.9911504424779,
      "grad_norm": 1.5417596159750246e-06,
      "learning_rate": 1.949111810843812e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19846304,
      "step": 34940
    },
    {
      "epoch": 613.070796460177,
      "grad_norm": 1.9620708258116792e-07,
      "learning_rate": 1.9453131912612694e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19848624,
      "step": 34945
    },
    {
      "epoch": 613.1592920353983,
      "grad_norm": 1.1814953495559166e-06,
      "learning_rate": 1.941518127001149e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19851312,
      "step": 34950
    },
    {
      "epoch": 613.2477876106195,
      "grad_norm": 1.2894973622223915e-07,
      "learning_rate": 1.9377266186487107e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19854032,
      "step": 34955
    },
    {
      "epoch": 613.3362831858407,
      "grad_norm": 1.8575069304915814e-07,
      "learning_rate": 1.9339386667886483e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19856832,
      "step": 34960
    },
    {
      "epoch": 613.4247787610619,
      "grad_norm": 2.45122151909527e-07,
      "learning_rate": 1.9301542720051024e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19859984,
      "step": 34965
    },
    {
      "epoch": 613.5132743362832,
      "grad_norm": 2.4964279532468936e-07,
      "learning_rate": 1.926373434881684e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19863152,
      "step": 34970
    },
    {
      "epoch": 613.6017699115044,
      "grad_norm": 1.6488229448441416e-07,
      "learning_rate": 1.9225961560014468e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19865824,
      "step": 34975
    },
    {
      "epoch": 613.6902654867257,
      "grad_norm": 1.0489907253941055e-05,
      "learning_rate": 1.918822435946885e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19868768,
      "step": 34980
    },
    {
      "epoch": 613.7787610619469,
      "grad_norm": 7.182684953477292e-07,
      "learning_rate": 1.915052275299961e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19871184,
      "step": 34985
    },
    {
      "epoch": 613.8672566371681,
      "grad_norm": 5.829169253956934e-07,
      "learning_rate": 1.9112856746420854e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19874480,
      "step": 34990
    },
    {
      "epoch": 613.9557522123894,
      "grad_norm": 1.0098479208409117e-07,
      "learning_rate": 1.907522634554104e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19877808,
      "step": 34995
    },
    {
      "epoch": 614.0353982300885,
      "grad_norm": 7.65711689609816e-08,
      "learning_rate": 1.9037631556163337e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19879808,
      "step": 35000
    },
    {
      "epoch": 614.0353982300885,
      "eval_loss": 0.8934006094932556,
      "eval_runtime": 1.0638,
      "eval_samples_per_second": 23.501,
      "eval_steps_per_second": 12.221,
      "num_input_tokens_seen": 19879808,
      "step": 35000
    },
    {
      "epoch": 614.1238938053098,
      "grad_norm": 1.2652932923629123e-07,
      "learning_rate": 1.9000072384085272e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19882944,
      "step": 35005
    },
    {
      "epoch": 614.212389380531,
      "grad_norm": 1.0809133499378731e-07,
      "learning_rate": 1.8962548835098987e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19885712,
      "step": 35010
    },
    {
      "epoch": 614.3008849557522,
      "grad_norm": 1.4391276863534586e-06,
      "learning_rate": 1.8925060914991077e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19889200,
      "step": 35015
    },
    {
      "epoch": 614.3893805309734,
      "grad_norm": 1.832834328752142e-07,
      "learning_rate": 1.888760862954264e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19892128,
      "step": 35020
    },
    {
      "epoch": 614.4778761061947,
      "grad_norm": 1.6508828082351101e-07,
      "learning_rate": 1.8850191984529309e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19894768,
      "step": 35025
    },
    {
      "epoch": 614.566371681416,
      "grad_norm": 3.178290342020773e-07,
      "learning_rate": 1.8812810985721186e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19897520,
      "step": 35030
    },
    {
      "epoch": 614.6548672566372,
      "grad_norm": 1.9910902437914046e-07,
      "learning_rate": 1.8775465638882856e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19899888,
      "step": 35035
    },
    {
      "epoch": 614.7433628318585,
      "grad_norm": 6.10500521247559e-08,
      "learning_rate": 1.8738155949773517e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19903024,
      "step": 35040
    },
    {
      "epoch": 614.8318584070796,
      "grad_norm": 3.9529615492028825e-07,
      "learning_rate": 1.8700881924146707e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19905856,
      "step": 35045
    },
    {
      "epoch": 614.9203539823009,
      "grad_norm": 4.7668035563219746e-07,
      "learning_rate": 1.8663643567750577e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19908624,
      "step": 35050
    },
    {
      "epoch": 615.0,
      "grad_norm": 3.4932887160721293e-07,
      "learning_rate": 1.8626440886327813e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19911632,
      "step": 35055
    },
    {
      "epoch": 615.0884955752213,
      "grad_norm": 2.0403552980496897e-07,
      "learning_rate": 1.8589273885615432e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19914528,
      "step": 35060
    },
    {
      "epoch": 615.1769911504425,
      "grad_norm": 1.3429557554900384e-07,
      "learning_rate": 1.8552142571345133e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19917424,
      "step": 35065
    },
    {
      "epoch": 615.2654867256637,
      "grad_norm": 1.8112228872269043e-07,
      "learning_rate": 1.8515046949243025e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19919952,
      "step": 35070
    },
    {
      "epoch": 615.3539823008849,
      "grad_norm": 1.0695755037204435e-07,
      "learning_rate": 1.8477987025029674e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19922608,
      "step": 35075
    },
    {
      "epoch": 615.4424778761062,
      "grad_norm": 1.2611187116817746e-07,
      "learning_rate": 1.8440962804420232e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19925584,
      "step": 35080
    },
    {
      "epoch": 615.5309734513274,
      "grad_norm": 1.3830830880579015e-07,
      "learning_rate": 1.8403974293124265e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19928576,
      "step": 35085
    },
    {
      "epoch": 615.6194690265487,
      "grad_norm": 1.5239036201819545e-06,
      "learning_rate": 1.8367021496845854e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19931808,
      "step": 35090
    },
    {
      "epoch": 615.70796460177,
      "grad_norm": 9.640497466989473e-08,
      "learning_rate": 1.8330104421283662e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19934528,
      "step": 35095
    },
    {
      "epoch": 615.7964601769911,
      "grad_norm": 1.2629757293325383e-05,
      "learning_rate": 1.8293223072130717e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19937040,
      "step": 35100
    },
    {
      "epoch": 615.8849557522124,
      "grad_norm": 8.415912589043728e-08,
      "learning_rate": 1.8256377455074525e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19939840,
      "step": 35105
    },
    {
      "epoch": 615.9734513274336,
      "grad_norm": 1.3139164423137117e-07,
      "learning_rate": 1.8219567575797263e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19943088,
      "step": 35110
    },
    {
      "epoch": 616.0530973451328,
      "grad_norm": 1.0388584996690042e-06,
      "learning_rate": 1.8182793439975365e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19945608,
      "step": 35115
    },
    {
      "epoch": 616.141592920354,
      "grad_norm": 2.2861675574858964e-07,
      "learning_rate": 1.8146055053279958e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19949160,
      "step": 35120
    },
    {
      "epoch": 616.2300884955753,
      "grad_norm": 1.563117990599494e-07,
      "learning_rate": 1.8109352421376486e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19951752,
      "step": 35125
    },
    {
      "epoch": 616.3185840707964,
      "grad_norm": 1.149434112335257e-07,
      "learning_rate": 1.8072685549924972e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19954360,
      "step": 35130
    },
    {
      "epoch": 616.4070796460177,
      "grad_norm": 2.2251171571952e-07,
      "learning_rate": 1.8036054444579982e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19957512,
      "step": 35135
    },
    {
      "epoch": 616.4955752212389,
      "grad_norm": 7.164440063434085e-08,
      "learning_rate": 1.7999459110990407e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19959912,
      "step": 35140
    },
    {
      "epoch": 616.5840707964602,
      "grad_norm": 2.0000764777705626e-07,
      "learning_rate": 1.7962899554799712e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19962712,
      "step": 35145
    },
    {
      "epoch": 616.6725663716815,
      "grad_norm": 3.2438181278848788e-06,
      "learning_rate": 1.7926375781645937e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19965896,
      "step": 35150
    },
    {
      "epoch": 616.7610619469026,
      "grad_norm": 1.1771676611260773e-07,
      "learning_rate": 1.7889887797161359e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19968424,
      "step": 35155
    },
    {
      "epoch": 616.8495575221239,
      "grad_norm": 1.4475629939170176e-07,
      "learning_rate": 1.7853435606973028e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19971368,
      "step": 35160
    },
    {
      "epoch": 616.9380530973451,
      "grad_norm": 4.070879811024497e-07,
      "learning_rate": 1.781701921670223e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19974552,
      "step": 35165
    },
    {
      "epoch": 617.0176991150443,
      "grad_norm": 9.936676548250034e-08,
      "learning_rate": 1.7780638631964886e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19976784,
      "step": 35170
    },
    {
      "epoch": 617.1061946902655,
      "grad_norm": 1.4377710044755077e-07,
      "learning_rate": 1.7744293858371314e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19979376,
      "step": 35175
    },
    {
      "epoch": 617.1946902654868,
      "grad_norm": 1.4136398931441363e-06,
      "learning_rate": 1.770798490152631e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19982160,
      "step": 35180
    },
    {
      "epoch": 617.2831858407079,
      "grad_norm": 3.1649875609218725e-07,
      "learning_rate": 1.767171176702917e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19985040,
      "step": 35185
    },
    {
      "epoch": 617.3716814159292,
      "grad_norm": 2.1351608836539526e-07,
      "learning_rate": 1.7635474460473755e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19987680,
      "step": 35190
    },
    {
      "epoch": 617.4601769911504,
      "grad_norm": 1.747040982991166e-07,
      "learning_rate": 1.7599272987448206e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19990416,
      "step": 35195
    },
    {
      "epoch": 617.5486725663717,
      "grad_norm": 2.6146804543714097e-07,
      "learning_rate": 1.7563107353535362e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19993952,
      "step": 35200
    },
    {
      "epoch": 617.5486725663717,
      "eval_loss": 0.9040440917015076,
      "eval_runtime": 1.064,
      "eval_samples_per_second": 23.497,
      "eval_steps_per_second": 12.218,
      "num_input_tokens_seen": 19993952,
      "step": 35200
    },
    {
      "epoch": 617.637168141593,
      "grad_norm": 3.776313519665564e-07,
      "learning_rate": 1.7526977564312263e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 19997328,
      "step": 35205
    },
    {
      "epoch": 617.7256637168142,
      "grad_norm": 1.2225073930949293e-07,
      "learning_rate": 1.7490883625350701e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20000320,
      "step": 35210
    },
    {
      "epoch": 617.8141592920354,
      "grad_norm": 9.351442713523284e-07,
      "learning_rate": 1.7454825542216807e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20003296,
      "step": 35215
    },
    {
      "epoch": 617.9026548672566,
      "grad_norm": 1.6351222598132154e-07,
      "learning_rate": 1.7418803320471105e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20006144,
      "step": 35220
    },
    {
      "epoch": 617.9911504424779,
      "grad_norm": 2.830544190146611e-07,
      "learning_rate": 1.7382816965668737e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20008832,
      "step": 35225
    },
    {
      "epoch": 618.070796460177,
      "grad_norm": 1.462086743231339e-06,
      "learning_rate": 1.7346866483359285e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20011784,
      "step": 35230
    },
    {
      "epoch": 618.1592920353983,
      "grad_norm": 2.3901034751361294e-07,
      "learning_rate": 1.7310951879086657e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20014936,
      "step": 35235
    },
    {
      "epoch": 618.2477876106195,
      "grad_norm": 2.037483000094653e-06,
      "learning_rate": 1.7275073158389471e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20017608,
      "step": 35240
    },
    {
      "epoch": 618.3362831858407,
      "grad_norm": 1.679079133509731e-07,
      "learning_rate": 1.723923032680061e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20020616,
      "step": 35245
    },
    {
      "epoch": 618.4247787610619,
      "grad_norm": 1.9903640691154578e-07,
      "learning_rate": 1.7203423389847428e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20023352,
      "step": 35250
    },
    {
      "epoch": 618.5132743362832,
      "grad_norm": 2.2260709897636843e-07,
      "learning_rate": 1.7167652353051928e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20025912,
      "step": 35255
    },
    {
      "epoch": 618.6017699115044,
      "grad_norm": 1.9095401171398407e-07,
      "learning_rate": 1.7131917221930333e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20028280,
      "step": 35260
    },
    {
      "epoch": 618.6902654867257,
      "grad_norm": 1.0835940145170753e-07,
      "learning_rate": 1.7096218001993513e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20031176,
      "step": 35265
    },
    {
      "epoch": 618.7787610619469,
      "grad_norm": 3.0154492947076506e-07,
      "learning_rate": 1.706055469874676e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20033912,
      "step": 35270
    },
    {
      "epoch": 618.8672566371681,
      "grad_norm": 3.667099690574105e-07,
      "learning_rate": 1.702492731768976e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20036936,
      "step": 35275
    },
    {
      "epoch": 618.9557522123894,
      "grad_norm": 6.731564212714147e-07,
      "learning_rate": 1.6989335864316724e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20039928,
      "step": 35280
    },
    {
      "epoch": 619.0353982300885,
      "grad_norm": 5.046276783104986e-07,
      "learning_rate": 1.6953780344116265e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20041984,
      "step": 35285
    },
    {
      "epoch": 619.1238938053098,
      "grad_norm": 1.1791774312541747e-07,
      "learning_rate": 1.6918260762571497e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20044912,
      "step": 35290
    },
    {
      "epoch": 619.212389380531,
      "grad_norm": 2.450530303121923e-07,
      "learning_rate": 1.6882777125160093e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20047888,
      "step": 35295
    },
    {
      "epoch": 619.3008849557522,
      "grad_norm": 1.8541078361522523e-07,
      "learning_rate": 1.6847329437353899e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20050784,
      "step": 35300
    },
    {
      "epoch": 619.3893805309734,
      "grad_norm": 1.1851634553750046e-06,
      "learning_rate": 1.6811917704619511e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20053376,
      "step": 35305
    },
    {
      "epoch": 619.4778761061947,
      "grad_norm": 1.1822418599649609e-07,
      "learning_rate": 1.67765419324179e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20056288,
      "step": 35310
    },
    {
      "epoch": 619.566371681416,
      "grad_norm": 1.935439968292485e-07,
      "learning_rate": 1.6741202126204364e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20059344,
      "step": 35315
    },
    {
      "epoch": 619.6548672566372,
      "grad_norm": 5.006507421967399e-07,
      "learning_rate": 1.6705898291428767e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20062768,
      "step": 35320
    },
    {
      "epoch": 619.7433628318585,
      "grad_norm": 1.193184743897291e-06,
      "learning_rate": 1.6670630433535395e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20065344,
      "step": 35325
    },
    {
      "epoch": 619.8318584070796,
      "grad_norm": 2.3844739871492493e-07,
      "learning_rate": 1.6635398557962979e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20068240,
      "step": 35330
    },
    {
      "epoch": 619.9203539823009,
      "grad_norm": 7.454747219526325e-08,
      "learning_rate": 1.660020267014481e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20070784,
      "step": 35335
    },
    {
      "epoch": 620.0,
      "grad_norm": 7.785780553604127e-08,
      "learning_rate": 1.6565042775508438e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20073496,
      "step": 35340
    },
    {
      "epoch": 620.0884955752213,
      "grad_norm": 7.439016513899332e-08,
      "learning_rate": 1.6529918879475997e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20077080,
      "step": 35345
    },
    {
      "epoch": 620.1769911504425,
      "grad_norm": 1.2999318244055758e-07,
      "learning_rate": 1.6494830987464043e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20079960,
      "step": 35350
    },
    {
      "epoch": 620.2654867256637,
      "grad_norm": 1.8054537065381737e-07,
      "learning_rate": 1.6459779104883555e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20082744,
      "step": 35355
    },
    {
      "epoch": 620.3539823008849,
      "grad_norm": 1.3305827906151535e-07,
      "learning_rate": 1.6424763237140013e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20085416,
      "step": 35360
    },
    {
      "epoch": 620.4424778761062,
      "grad_norm": 1.0558359235801618e-06,
      "learning_rate": 1.6389783389633207e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20087928,
      "step": 35365
    },
    {
      "epoch": 620.5309734513274,
      "grad_norm": 9.698461411744574e-08,
      "learning_rate": 1.6354839567757546e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20090600,
      "step": 35370
    },
    {
      "epoch": 620.6194690265487,
      "grad_norm": 2.0715617665700847e-07,
      "learning_rate": 1.6319931776901831e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20093560,
      "step": 35375
    },
    {
      "epoch": 620.70796460177,
      "grad_norm": 3.683331613046903e-07,
      "learning_rate": 1.6285060022449229e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20096488,
      "step": 35380
    },
    {
      "epoch": 620.7964601769911,
      "grad_norm": 1.7447064237785526e-07,
      "learning_rate": 1.6250224309777434e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20099512,
      "step": 35385
    },
    {
      "epoch": 620.8849557522124,
      "grad_norm": 2.056789298876538e-07,
      "learning_rate": 1.6215424644258515e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20102568,
      "step": 35390
    },
    {
      "epoch": 620.9734513274336,
      "grad_norm": 1.245185558218509e-06,
      "learning_rate": 1.6180661031259036e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20105112,
      "step": 35395
    },
    {
      "epoch": 621.0530973451328,
      "grad_norm": 5.22409163750126e-06,
      "learning_rate": 1.614593347613999e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20107560,
      "step": 35400
    },
    {
      "epoch": 621.0530973451328,
      "eval_loss": 0.911011278629303,
      "eval_runtime": 1.0643,
      "eval_samples_per_second": 23.49,
      "eval_steps_per_second": 12.215,
      "num_input_tokens_seen": 20107560,
      "step": 35400
    },
    {
      "epoch": 621.141592920354,
      "grad_norm": 1.2306152541441406e-07,
      "learning_rate": 1.6111241984256758e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20110056,
      "step": 35405
    },
    {
      "epoch": 621.2300884955753,
      "grad_norm": 1.0984193465901626e-07,
      "learning_rate": 1.6076586560959257e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20113336,
      "step": 35410
    },
    {
      "epoch": 621.3185840707964,
      "grad_norm": 5.777714022769942e-07,
      "learning_rate": 1.604196721159182e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20115784,
      "step": 35415
    },
    {
      "epoch": 621.4070796460177,
      "grad_norm": 1.2427605042830692e-06,
      "learning_rate": 1.6007383941493092e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20118760,
      "step": 35420
    },
    {
      "epoch": 621.4955752212389,
      "grad_norm": 1.1388793552669085e-07,
      "learning_rate": 1.5972836755996285e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20121896,
      "step": 35425
    },
    {
      "epoch": 621.5840707964602,
      "grad_norm": 4.2427558355484507e-07,
      "learning_rate": 1.5938325660429076e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20124536,
      "step": 35430
    },
    {
      "epoch": 621.6725663716815,
      "grad_norm": 6.897625581814282e-08,
      "learning_rate": 1.5903850660113378e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20127448,
      "step": 35435
    },
    {
      "epoch": 621.7610619469026,
      "grad_norm": 1.4417103955111088e-07,
      "learning_rate": 1.5869411760365826e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20130504,
      "step": 35440
    },
    {
      "epoch": 621.8495575221239,
      "grad_norm": 1.0854457599407397e-07,
      "learning_rate": 1.58350089664972e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20133064,
      "step": 35445
    },
    {
      "epoch": 621.9380530973451,
      "grad_norm": 9.801653533259014e-08,
      "learning_rate": 1.5800642283812865e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20136632,
      "step": 35450
    },
    {
      "epoch": 622.0176991150443,
      "grad_norm": 1.301766161532214e-07,
      "learning_rate": 1.5766311717612698e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20139096,
      "step": 35455
    },
    {
      "epoch": 622.1061946902655,
      "grad_norm": 1.6367293653729575e-07,
      "learning_rate": 1.5732017273190818e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20141880,
      "step": 35460
    },
    {
      "epoch": 622.1946902654868,
      "grad_norm": 1.8500605847293627e-07,
      "learning_rate": 1.5697758955835806e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20144792,
      "step": 35465
    },
    {
      "epoch": 622.2831858407079,
      "grad_norm": 5.875947977074247e-07,
      "learning_rate": 1.566353677083085e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20147560,
      "step": 35470
    },
    {
      "epoch": 622.3716814159292,
      "grad_norm": 4.190453182673082e-06,
      "learning_rate": 1.562935072345334e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20150424,
      "step": 35475
    },
    {
      "epoch": 622.4601769911504,
      "grad_norm": 1.57860952754163e-07,
      "learning_rate": 1.5595200818975281e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20152920,
      "step": 35480
    },
    {
      "epoch": 622.5486725663717,
      "grad_norm": 9.271153089684958e-07,
      "learning_rate": 1.5561087062662905e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20156088,
      "step": 35485
    },
    {
      "epoch": 622.637168141593,
      "grad_norm": 1.673184488026891e-05,
      "learning_rate": 1.5527009459777087e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20158776,
      "step": 35490
    },
    {
      "epoch": 622.7256637168142,
      "grad_norm": 1.9655720961964107e-07,
      "learning_rate": 1.5492968015572984e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20161576,
      "step": 35495
    },
    {
      "epoch": 622.8141592920354,
      "grad_norm": 1.8498856491078186e-07,
      "learning_rate": 1.5458962735300203e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20164376,
      "step": 35500
    },
    {
      "epoch": 622.9026548672566,
      "grad_norm": 7.506919530442246e-08,
      "learning_rate": 1.54249936242028e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20167736,
      "step": 35505
    },
    {
      "epoch": 622.9911504424779,
      "grad_norm": 1.4954235894037993e-07,
      "learning_rate": 1.5391060687519222e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20170728,
      "step": 35510
    },
    {
      "epoch": 623.070796460177,
      "grad_norm": 1.2861140419317962e-07,
      "learning_rate": 1.5357163930482367e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20173000,
      "step": 35515
    },
    {
      "epoch": 623.1592920353983,
      "grad_norm": 3.4615945878613275e-06,
      "learning_rate": 1.532330335831955e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20175752,
      "step": 35520
    },
    {
      "epoch": 623.2477876106195,
      "grad_norm": 1.937919842021074e-05,
      "learning_rate": 1.5289478976252491e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20178216,
      "step": 35525
    },
    {
      "epoch": 623.3362831858407,
      "grad_norm": 1.2971558760455082e-07,
      "learning_rate": 1.5255690789497345e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20180904,
      "step": 35530
    },
    {
      "epoch": 623.4247787610619,
      "grad_norm": 6.470694557947354e-08,
      "learning_rate": 1.5221938803264641e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20183448,
      "step": 35535
    },
    {
      "epoch": 623.5132743362832,
      "grad_norm": 6.78117714869586e-07,
      "learning_rate": 1.518822302275938e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20186696,
      "step": 35540
    },
    {
      "epoch": 623.6017699115044,
      "grad_norm": 1.0607619316260752e-07,
      "learning_rate": 1.5154543453180958e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20189512,
      "step": 35545
    },
    {
      "epoch": 623.6902654867257,
      "grad_norm": 1.0919695370148474e-07,
      "learning_rate": 1.5120900099723167e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20192360,
      "step": 35550
    },
    {
      "epoch": 623.7787610619469,
      "grad_norm": 2.0587953031281359e-07,
      "learning_rate": 1.5087292967574273e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20195640,
      "step": 35555
    },
    {
      "epoch": 623.8672566371681,
      "grad_norm": 7.108793909083033e-08,
      "learning_rate": 1.5053722061916908e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20198856,
      "step": 35560
    },
    {
      "epoch": 623.9557522123894,
      "grad_norm": 1.295653305533051e-06,
      "learning_rate": 1.5020187387928124e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20201464,
      "step": 35565
    },
    {
      "epoch": 624.0353982300885,
      "grad_norm": 2.0162183034244663e-07,
      "learning_rate": 1.4986688950779343e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20204120,
      "step": 35570
    },
    {
      "epoch": 624.1238938053098,
      "grad_norm": 2.8074414331058506e-06,
      "learning_rate": 1.495322675563654e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20206728,
      "step": 35575
    },
    {
      "epoch": 624.212389380531,
      "grad_norm": 6.465608493044783e-08,
      "learning_rate": 1.4919800807659922e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20209912,
      "step": 35580
    },
    {
      "epoch": 624.3008849557522,
      "grad_norm": 2.938511727279547e-07,
      "learning_rate": 1.4886411112004255e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20212520,
      "step": 35585
    },
    {
      "epoch": 624.3893805309734,
      "grad_norm": 1.1073005623529752e-07,
      "learning_rate": 1.4853057673818588e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20215464,
      "step": 35590
    },
    {
      "epoch": 624.4778761061947,
      "grad_norm": 1.755406486836364e-07,
      "learning_rate": 1.481974049824647e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20218184,
      "step": 35595
    },
    {
      "epoch": 624.566371681416,
      "grad_norm": 5.084381768938329e-07,
      "learning_rate": 1.4786459590425849e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20220888,
      "step": 35600
    },
    {
      "epoch": 624.566371681416,
      "eval_loss": 0.8988681435585022,
      "eval_runtime": 1.0634,
      "eval_samples_per_second": 23.509,
      "eval_steps_per_second": 12.225,
      "num_input_tokens_seen": 20220888,
      "step": 35600
    },
    {
      "epoch": 624.6548672566372,
      "grad_norm": 6.65734773974691e-08,
      "learning_rate": 1.4753214955489036e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20224344,
      "step": 35605
    },
    {
      "epoch": 624.7433628318585,
      "grad_norm": 1.3098374438413884e-06,
      "learning_rate": 1.4720006598562737e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20226984,
      "step": 35610
    },
    {
      "epoch": 624.8318584070796,
      "grad_norm": 1.4239448375974462e-07,
      "learning_rate": 1.4686834524768185e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20229896,
      "step": 35615
    },
    {
      "epoch": 624.9203539823009,
      "grad_norm": 1.362673032190287e-07,
      "learning_rate": 1.4653698739220844e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20232712,
      "step": 35620
    },
    {
      "epoch": 625.0,
      "grad_norm": 1.3942302530267625e-06,
      "learning_rate": 1.4620599247030715e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20235192,
      "step": 35625
    },
    {
      "epoch": 625.0884955752213,
      "grad_norm": 2.096998059641919e-07,
      "learning_rate": 1.4587536053302125e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20237512,
      "step": 35630
    },
    {
      "epoch": 625.1769911504425,
      "grad_norm": 9.634050002205186e-07,
      "learning_rate": 1.4554509163133862e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20240280,
      "step": 35635
    },
    {
      "epoch": 625.2654867256637,
      "grad_norm": 3.45880664553988e-07,
      "learning_rate": 1.4521518581619098e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20243192,
      "step": 35640
    },
    {
      "epoch": 625.3539823008849,
      "grad_norm": 3.9328912748715084e-07,
      "learning_rate": 1.4488564313845348e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20246152,
      "step": 35645
    },
    {
      "epoch": 625.4424778761062,
      "grad_norm": 9.073281859173221e-08,
      "learning_rate": 1.4455646364894603e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20248936,
      "step": 35650
    },
    {
      "epoch": 625.5309734513274,
      "grad_norm": 2.8604497401829576e-06,
      "learning_rate": 1.4422764739843247e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20251816,
      "step": 35655
    },
    {
      "epoch": 625.6194690265487,
      "grad_norm": 7.280198843773178e-08,
      "learning_rate": 1.4389919443762e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20254776,
      "step": 35660
    },
    {
      "epoch": 625.70796460177,
      "grad_norm": 2.7068554686593416e-07,
      "learning_rate": 1.4357110481716063e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20257576,
      "step": 35665
    },
    {
      "epoch": 625.7964601769911,
      "grad_norm": 1.6946161451869557e-07,
      "learning_rate": 1.4324337858764941e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20260120,
      "step": 35670
    },
    {
      "epoch": 625.8849557522124,
      "grad_norm": 1.5537065678472572e-07,
      "learning_rate": 1.4291601579962622e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20263528,
      "step": 35675
    },
    {
      "epoch": 625.9734513274336,
      "grad_norm": 4.540121949503373e-07,
      "learning_rate": 1.42589016503574e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20266504,
      "step": 35680
    },
    {
      "epoch": 626.0530973451328,
      "grad_norm": 9.73199712461792e-06,
      "learning_rate": 1.4226238074992099e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20269536,
      "step": 35685
    },
    {
      "epoch": 626.141592920354,
      "grad_norm": 2.5729661956575e-06,
      "learning_rate": 1.4193610858903778e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20272560,
      "step": 35690
    },
    {
      "epoch": 626.2300884955753,
      "grad_norm": 5.128974294166255e-07,
      "learning_rate": 1.416102000712402e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20275440,
      "step": 35695
    },
    {
      "epoch": 626.3185840707964,
      "grad_norm": 7.487206943324054e-08,
      "learning_rate": 1.4128465524678668e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20278272,
      "step": 35700
    },
    {
      "epoch": 626.4070796460177,
      "grad_norm": 1.477495601420742e-07,
      "learning_rate": 1.4095947416588124e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20280976,
      "step": 35705
    },
    {
      "epoch": 626.4955752212389,
      "grad_norm": 1.266939989363891e-06,
      "learning_rate": 1.4063465687866983e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20283744,
      "step": 35710
    },
    {
      "epoch": 626.5840707964602,
      "grad_norm": 2.570439505689137e-07,
      "learning_rate": 1.4031020343524438e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20286256,
      "step": 35715
    },
    {
      "epoch": 626.6725663716815,
      "grad_norm": 1.1663503585168655e-07,
      "learning_rate": 1.3998611388563926e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20288976,
      "step": 35720
    },
    {
      "epoch": 626.7610619469026,
      "grad_norm": 2.682309059309773e-07,
      "learning_rate": 1.3966238827983314e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20292032,
      "step": 35725
    },
    {
      "epoch": 626.8495575221239,
      "grad_norm": 1.929053468074926e-07,
      "learning_rate": 1.393390266677483e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20294672,
      "step": 35730
    },
    {
      "epoch": 626.9380530973451,
      "grad_norm": 2.2736051619176578e-07,
      "learning_rate": 1.3901602909925204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20297520,
      "step": 35735
    },
    {
      "epoch": 627.0176991150443,
      "grad_norm": 3.9383888861266314e-07,
      "learning_rate": 1.3869339562415373e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20300392,
      "step": 35740
    },
    {
      "epoch": 627.1061946902655,
      "grad_norm": 1.1018202172863312e-07,
      "learning_rate": 1.38371126292208e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20303432,
      "step": 35745
    },
    {
      "epoch": 627.1946902654868,
      "grad_norm": 1.940656204624247e-07,
      "learning_rate": 1.3804922115311286e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20306616,
      "step": 35750
    },
    {
      "epoch": 627.2831858407079,
      "grad_norm": 1.4042433349459316e-07,
      "learning_rate": 1.3772768025650945e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20309384,
      "step": 35755
    },
    {
      "epoch": 627.3716814159292,
      "grad_norm": 2.275910588878105e-07,
      "learning_rate": 1.3740650365198448e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20312280,
      "step": 35760
    },
    {
      "epoch": 627.4601769911504,
      "grad_norm": 7.05663467215345e-07,
      "learning_rate": 1.3708569138906612e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20314664,
      "step": 35765
    },
    {
      "epoch": 627.5486725663717,
      "grad_norm": 6.66715823172126e-06,
      "learning_rate": 1.367652435172287e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20317512,
      "step": 35770
    },
    {
      "epoch": 627.637168141593,
      "grad_norm": 1.2242341540513735e-07,
      "learning_rate": 1.364451600858893e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20321016,
      "step": 35775
    },
    {
      "epoch": 627.7256637168142,
      "grad_norm": 6.978481081887367e-08,
      "learning_rate": 1.3612544114440823e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20323928,
      "step": 35780
    },
    {
      "epoch": 627.8141592920354,
      "grad_norm": 5.975024919280258e-07,
      "learning_rate": 1.3580608674209072e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20326360,
      "step": 35785
    },
    {
      "epoch": 627.9026548672566,
      "grad_norm": 1.1448114491940942e-06,
      "learning_rate": 1.3548709692818434e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20329016,
      "step": 35790
    },
    {
      "epoch": 627.9911504424779,
      "grad_norm": 3.9496487147516746e-07,
      "learning_rate": 1.3516847175188223e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20331704,
      "step": 35795
    },
    {
      "epoch": 628.070796460177,
      "grad_norm": 1.4258209546369471e-07,
      "learning_rate": 1.348502112623204e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20333904,
      "step": 35800
    },
    {
      "epoch": 628.070796460177,
      "eval_loss": 0.9270323514938354,
      "eval_runtime": 1.0693,
      "eval_samples_per_second": 23.381,
      "eval_steps_per_second": 12.158,
      "num_input_tokens_seen": 20333904,
      "step": 35800
    },
    {
      "epoch": 628.1592920353983,
      "grad_norm": 1.1422493884083451e-07,
      "learning_rate": 1.3453231550857787e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20337120,
      "step": 35805
    },
    {
      "epoch": 628.2477876106195,
      "grad_norm": 1.0417165441367615e-07,
      "learning_rate": 1.3421478453967878e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20339840,
      "step": 35810
    },
    {
      "epoch": 628.3362831858407,
      "grad_norm": 1.6725572038467362e-07,
      "learning_rate": 1.3389761840459065e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20342368,
      "step": 35815
    },
    {
      "epoch": 628.4247787610619,
      "grad_norm": 9.993227934046445e-08,
      "learning_rate": 1.3358081715222376e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20344928,
      "step": 35820
    },
    {
      "epoch": 628.5132743362832,
      "grad_norm": 1.885563563064352e-07,
      "learning_rate": 1.3326438083143295e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20347616,
      "step": 35825
    },
    {
      "epoch": 628.6017699115044,
      "grad_norm": 1.198295791482451e-07,
      "learning_rate": 1.3294830949101723e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20350736,
      "step": 35830
    },
    {
      "epoch": 628.6902654867257,
      "grad_norm": 1.5703989220128278e-07,
      "learning_rate": 1.3263260317971815e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20354064,
      "step": 35835
    },
    {
      "epoch": 628.7787610619469,
      "grad_norm": 1.5680973319831537e-07,
      "learning_rate": 1.3231726194622208e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20356624,
      "step": 35840
    },
    {
      "epoch": 628.8672566371681,
      "grad_norm": 7.630395089108788e-07,
      "learning_rate": 1.3200228583915814e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20359232,
      "step": 35845
    },
    {
      "epoch": 628.9557522123894,
      "grad_norm": 1.7549169228914252e-07,
      "learning_rate": 1.3168767490709971e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20362432,
      "step": 35850
    },
    {
      "epoch": 629.0353982300885,
      "grad_norm": 4.2803898736565316e-07,
      "learning_rate": 1.3137342919856437e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20364664,
      "step": 35855
    },
    {
      "epoch": 629.1238938053098,
      "grad_norm": 2.8058798307029065e-07,
      "learning_rate": 1.310595487620117e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20367576,
      "step": 35860
    },
    {
      "epoch": 629.212389380531,
      "grad_norm": 2.937374858902331e-07,
      "learning_rate": 1.3074603364584715e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20370120,
      "step": 35865
    },
    {
      "epoch": 629.3008849557522,
      "grad_norm": 2.0148387136487145e-07,
      "learning_rate": 1.3043288389841758e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20373448,
      "step": 35870
    },
    {
      "epoch": 629.3893805309734,
      "grad_norm": 5.33071670361096e-06,
      "learning_rate": 1.3012009956801546e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20376504,
      "step": 35875
    },
    {
      "epoch": 629.4778761061947,
      "grad_norm": 1.3728862313655554e-07,
      "learning_rate": 1.2980768070287586e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20379592,
      "step": 35880
    },
    {
      "epoch": 629.566371681416,
      "grad_norm": 1.0030237262981245e-07,
      "learning_rate": 1.2949562735117716e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20382168,
      "step": 35885
    },
    {
      "epoch": 629.6548672566372,
      "grad_norm": 1.0167486834689043e-06,
      "learning_rate": 1.291839395610428e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20385176,
      "step": 35890
    },
    {
      "epoch": 629.7433628318585,
      "grad_norm": 1.943443663776634e-07,
      "learning_rate": 1.2887261738053852e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20388248,
      "step": 35895
    },
    {
      "epoch": 629.8318584070796,
      "grad_norm": 5.590414389189391e-07,
      "learning_rate": 1.2856166085767396e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20390744,
      "step": 35900
    },
    {
      "epoch": 629.9203539823009,
      "grad_norm": 2.8279484354243323e-07,
      "learning_rate": 1.2825107004040272e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20393832,
      "step": 35905
    },
    {
      "epoch": 630.0,
      "grad_norm": 7.31610634829849e-06,
      "learning_rate": 1.2794084497662146e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20395800,
      "step": 35910
    },
    {
      "epoch": 630.0884955752213,
      "grad_norm": 3.254181990541838e-07,
      "learning_rate": 1.276309857141711e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20398424,
      "step": 35915
    },
    {
      "epoch": 630.1769911504425,
      "grad_norm": 1.1149781897756839e-07,
      "learning_rate": 1.273214923008359e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20401128,
      "step": 35920
    },
    {
      "epoch": 630.2654867256637,
      "grad_norm": 7.741824958884536e-08,
      "learning_rate": 1.2701236478434352e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20403816,
      "step": 35925
    },
    {
      "epoch": 630.3539823008849,
      "grad_norm": 3.509671557822003e-07,
      "learning_rate": 1.2670360321236502e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20406408,
      "step": 35930
    },
    {
      "epoch": 630.4424778761062,
      "grad_norm": 1.2222120915339474e-07,
      "learning_rate": 1.2639520763251617e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20409496,
      "step": 35935
    },
    {
      "epoch": 630.5309734513274,
      "grad_norm": 2.76709243962614e-07,
      "learning_rate": 1.2608717809235448e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20412616,
      "step": 35940
    },
    {
      "epoch": 630.6194690265487,
      "grad_norm": 8.031835108113228e-08,
      "learning_rate": 1.2577951463938282e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20415896,
      "step": 35945
    },
    {
      "epoch": 630.70796460177,
      "grad_norm": 8.369602255697828e-08,
      "learning_rate": 1.2547221732104569e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20418376,
      "step": 35950
    },
    {
      "epoch": 630.7964601769911,
      "grad_norm": 2.679882129541511e-07,
      "learning_rate": 1.25165286184733e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20421608,
      "step": 35955
    },
    {
      "epoch": 630.8849557522124,
      "grad_norm": 1.0015980933530955e-07,
      "learning_rate": 1.248587212777777e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20424360,
      "step": 35960
    },
    {
      "epoch": 630.9734513274336,
      "grad_norm": 7.199154339332381e-08,
      "learning_rate": 1.2455252264745532e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20427240,
      "step": 35965
    },
    {
      "epoch": 631.0530973451328,
      "grad_norm": 7.095907506027288e-08,
      "learning_rate": 1.2424669034098528e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20429808,
      "step": 35970
    },
    {
      "epoch": 631.141592920354,
      "grad_norm": 1.7773325566849962e-07,
      "learning_rate": 1.2394122440553185e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20433088,
      "step": 35975
    },
    {
      "epoch": 631.2300884955753,
      "grad_norm": 1.590272944440585e-07,
      "learning_rate": 1.2363612488820037e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20435744,
      "step": 35980
    },
    {
      "epoch": 631.3185840707964,
      "grad_norm": 9.234480558006908e-08,
      "learning_rate": 1.2333139183604208e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20438528,
      "step": 35985
    },
    {
      "epoch": 631.4070796460177,
      "grad_norm": 1.3776010746369138e-05,
      "learning_rate": 1.2302702529604998e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20441504,
      "step": 35990
    },
    {
      "epoch": 631.4955752212389,
      "grad_norm": 1.2247585345903644e-06,
      "learning_rate": 1.227230253151615e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20443952,
      "step": 35995
    },
    {
      "epoch": 631.5840707964602,
      "grad_norm": 1.3281646715768147e-07,
      "learning_rate": 1.2241939194025748e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20446736,
      "step": 36000
    },
    {
      "epoch": 631.5840707964602,
      "eval_loss": 0.8908863067626953,
      "eval_runtime": 1.063,
      "eval_samples_per_second": 23.519,
      "eval_steps_per_second": 12.23,
      "num_input_tokens_seen": 20446736,
      "step": 36000
    },
    {
      "epoch": 631.6725663716815,
      "grad_norm": 3.6431168837225414e-07,
      "learning_rate": 1.2211612521816156e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20449360,
      "step": 36005
    },
    {
      "epoch": 631.7610619469026,
      "grad_norm": 1.9975824727680447e-07,
      "learning_rate": 1.2181322519564137e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20452112,
      "step": 36010
    },
    {
      "epoch": 631.8495575221239,
      "grad_norm": 6.749207415168712e-08,
      "learning_rate": 1.2151069191940839e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20455136,
      "step": 36015
    },
    {
      "epoch": 631.9380530973451,
      "grad_norm": 7.300949533828316e-08,
      "learning_rate": 1.2120852543611644e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20458336,
      "step": 36020
    },
    {
      "epoch": 632.0176991150443,
      "grad_norm": 1.1500557661747735e-07,
      "learning_rate": 1.2090672579236379e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20460712,
      "step": 36025
    },
    {
      "epoch": 632.1061946902655,
      "grad_norm": 1.3368045870265632e-07,
      "learning_rate": 1.2060529303469126e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20463080,
      "step": 36030
    },
    {
      "epoch": 632.1946902654868,
      "grad_norm": 1.922227426121026e-07,
      "learning_rate": 1.2030422720958445e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20465736,
      "step": 36035
    },
    {
      "epoch": 632.2831858407079,
      "grad_norm": 1.237333009385111e-07,
      "learning_rate": 1.200035283634704e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20468808,
      "step": 36040
    },
    {
      "epoch": 632.3716814159292,
      "grad_norm": 2.13633910561839e-07,
      "learning_rate": 1.1970319654272144e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20471240,
      "step": 36045
    },
    {
      "epoch": 632.4601769911504,
      "grad_norm": 9.788177379732588e-08,
      "learning_rate": 1.1940323179365192e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20474536,
      "step": 36050
    },
    {
      "epoch": 632.5486725663717,
      "grad_norm": 2.9396721856755903e-07,
      "learning_rate": 1.1910363416252095e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20477512,
      "step": 36055
    },
    {
      "epoch": 632.637168141593,
      "grad_norm": 1.3069194437775877e-07,
      "learning_rate": 1.1880440369552964e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20480408,
      "step": 36060
    },
    {
      "epoch": 632.7256637168142,
      "grad_norm": 1.0852293996777007e-07,
      "learning_rate": 1.1850554043882328e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20483864,
      "step": 36065
    },
    {
      "epoch": 632.8141592920354,
      "grad_norm": 9.953600255130368e-08,
      "learning_rate": 1.1820704443849028e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20486632,
      "step": 36070
    },
    {
      "epoch": 632.9026548672566,
      "grad_norm": 1.8021850110017112e-06,
      "learning_rate": 1.1790891574056219e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20489480,
      "step": 36075
    },
    {
      "epoch": 632.9911504424779,
      "grad_norm": 2.2126494059193647e-07,
      "learning_rate": 1.1761115439101523e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20492392,
      "step": 36080
    },
    {
      "epoch": 633.070796460177,
      "grad_norm": 1.0196863087230668e-07,
      "learning_rate": 1.1731376043576659e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20495352,
      "step": 36085
    },
    {
      "epoch": 633.1592920353983,
      "grad_norm": 9.300946857138115e-08,
      "learning_rate": 1.1701673392067875e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20498184,
      "step": 36090
    },
    {
      "epoch": 633.2477876106195,
      "grad_norm": 4.582036581268767e-07,
      "learning_rate": 1.1672007489155757e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20500760,
      "step": 36095
    },
    {
      "epoch": 633.3362831858407,
      "grad_norm": 4.149257165408926e-06,
      "learning_rate": 1.164237833941506e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20503464,
      "step": 36100
    },
    {
      "epoch": 633.4247787610619,
      "grad_norm": 7.892336384429655e-08,
      "learning_rate": 1.1612785947415022e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20506168,
      "step": 36105
    },
    {
      "epoch": 633.5132743362832,
      "grad_norm": 6.024624354949992e-08,
      "learning_rate": 1.1583230317719185e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20509784,
      "step": 36110
    },
    {
      "epoch": 633.6017699115044,
      "grad_norm": 6.533602459057875e-08,
      "learning_rate": 1.1553711454885318e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20512824,
      "step": 36115
    },
    {
      "epoch": 633.6902654867257,
      "grad_norm": 3.207636609658948e-07,
      "learning_rate": 1.152422936346567e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20515640,
      "step": 36120
    },
    {
      "epoch": 633.7787610619469,
      "grad_norm": 2.011240241017731e-07,
      "learning_rate": 1.1494784048006718e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20518248,
      "step": 36125
    },
    {
      "epoch": 633.8672566371681,
      "grad_norm": 1.6725577722809248e-07,
      "learning_rate": 1.1465375513049326e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20520824,
      "step": 36130
    },
    {
      "epoch": 633.9557522123894,
      "grad_norm": 4.7156549953797366e-07,
      "learning_rate": 1.1436003763128616e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20523560,
      "step": 36135
    },
    {
      "epoch": 634.0353982300885,
      "grad_norm": 1.049640871997326e-07,
      "learning_rate": 1.1406668802774106e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20526128,
      "step": 36140
    },
    {
      "epoch": 634.1238938053098,
      "grad_norm": 4.570978830997774e-07,
      "learning_rate": 1.137737063650965e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20528576,
      "step": 36145
    },
    {
      "epoch": 634.212389380531,
      "grad_norm": 1.4917998214514228e-07,
      "learning_rate": 1.1348109268853323e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20531360,
      "step": 36150
    },
    {
      "epoch": 634.3008849557522,
      "grad_norm": 1.7490569348410645e-07,
      "learning_rate": 1.1318884704317634e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20534064,
      "step": 36155
    },
    {
      "epoch": 634.3893805309734,
      "grad_norm": 1.1035397307068706e-07,
      "learning_rate": 1.1289696947409417e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20537040,
      "step": 36160
    },
    {
      "epoch": 634.4778761061947,
      "grad_norm": 1.9903929171505297e-07,
      "learning_rate": 1.126054600262974e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20540064,
      "step": 36165
    },
    {
      "epoch": 634.566371681416,
      "grad_norm": 2.8538931928778766e-07,
      "learning_rate": 1.1231431874474064e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20542720,
      "step": 36170
    },
    {
      "epoch": 634.6548672566372,
      "grad_norm": 1.2604266430571442e-07,
      "learning_rate": 1.12023545674321e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20545760,
      "step": 36175
    },
    {
      "epoch": 634.7433628318585,
      "grad_norm": 5.279202468955191e-06,
      "learning_rate": 1.117331408598804e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20548464,
      "step": 36180
    },
    {
      "epoch": 634.8318584070796,
      "grad_norm": 1.6063101782037847e-07,
      "learning_rate": 1.1144310434620191e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20551200,
      "step": 36185
    },
    {
      "epoch": 634.9203539823009,
      "grad_norm": 1.1437633418154292e-07,
      "learning_rate": 1.1115343617801365e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20554720,
      "step": 36190
    },
    {
      "epoch": 635.0,
      "grad_norm": 1.7788816819575004e-07,
      "learning_rate": 1.1086413639998515e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20557256,
      "step": 36195
    },
    {
      "epoch": 635.0884955752213,
      "grad_norm": 1.1645445852082048e-07,
      "learning_rate": 1.1057520505673103e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20560472,
      "step": 36200
    },
    {
      "epoch": 635.0884955752213,
      "eval_loss": 0.9129029512405396,
      "eval_runtime": 1.0643,
      "eval_samples_per_second": 23.489,
      "eval_steps_per_second": 12.214,
      "num_input_tokens_seen": 20560472,
      "step": 36200
    },
    {
      "epoch": 635.1769911504425,
      "grad_norm": 1.447889559358373e-07,
      "learning_rate": 1.1028664219280727e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20563496,
      "step": 36205
    },
    {
      "epoch": 635.2654867256637,
      "grad_norm": 9.50328455928684e-08,
      "learning_rate": 1.0999844785271468e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20566216,
      "step": 36210
    },
    {
      "epoch": 635.3539823008849,
      "grad_norm": 1.10823791032999e-07,
      "learning_rate": 1.097106220808955e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20569064,
      "step": 36215
    },
    {
      "epoch": 635.4424778761062,
      "grad_norm": 9.492004693356648e-08,
      "learning_rate": 1.0942316492173698e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20571896,
      "step": 36220
    },
    {
      "epoch": 635.5309734513274,
      "grad_norm": 7.858771056135083e-08,
      "learning_rate": 1.0913607641956841e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20574440,
      "step": 36225
    },
    {
      "epoch": 635.6194690265487,
      "grad_norm": 2.8809498076043383e-07,
      "learning_rate": 1.0884935661866213e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20577096,
      "step": 36230
    },
    {
      "epoch": 635.70796460177,
      "grad_norm": 1.6629094545805856e-07,
      "learning_rate": 1.0856300556323418e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20579912,
      "step": 36235
    },
    {
      "epoch": 635.7964601769911,
      "grad_norm": 1.2526129467005376e-07,
      "learning_rate": 1.0827702329744365e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20582664,
      "step": 36240
    },
    {
      "epoch": 635.8849557522124,
      "grad_norm": 1.1772918924179976e-06,
      "learning_rate": 1.0799140986539197e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20585496,
      "step": 36245
    },
    {
      "epoch": 635.9734513274336,
      "grad_norm": 1.660331747643795e-07,
      "learning_rate": 1.0770616531112526e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20588536,
      "step": 36250
    },
    {
      "epoch": 636.0530973451328,
      "grad_norm": 1.0133147299029588e-07,
      "learning_rate": 1.0742128967863085e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20591024,
      "step": 36255
    },
    {
      "epoch": 636.141592920354,
      "grad_norm": 1.182604165705925e-07,
      "learning_rate": 1.071367830118411e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20593424,
      "step": 36260
    },
    {
      "epoch": 636.2300884955753,
      "grad_norm": 2.563701286817377e-07,
      "learning_rate": 1.068526453546298e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20596960,
      "step": 36265
    },
    {
      "epoch": 636.3185840707964,
      "grad_norm": 1.328321985738512e-07,
      "learning_rate": 1.0656887675081467e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20599664,
      "step": 36270
    },
    {
      "epoch": 636.4070796460177,
      "grad_norm": 4.5968140511831734e-06,
      "learning_rate": 1.0628547724415628e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20602688,
      "step": 36275
    },
    {
      "epoch": 636.4955752212389,
      "grad_norm": 1.4178230856032314e-07,
      "learning_rate": 1.0600244687835881e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20605920,
      "step": 36280
    },
    {
      "epoch": 636.5840707964602,
      "grad_norm": 9.625431829363151e-08,
      "learning_rate": 1.0571978569706876e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20608800,
      "step": 36285
    },
    {
      "epoch": 636.6725663716815,
      "grad_norm": 1.0941100470063247e-07,
      "learning_rate": 1.0543749374387652e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20611552,
      "step": 36290
    },
    {
      "epoch": 636.7610619469026,
      "grad_norm": 1.5399331232401892e-06,
      "learning_rate": 1.051555710623142e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20614256,
      "step": 36295
    },
    {
      "epoch": 636.8495575221239,
      "grad_norm": 2.023972314191269e-07,
      "learning_rate": 1.0487401769585847e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20617344,
      "step": 36300
    },
    {
      "epoch": 636.9380530973451,
      "grad_norm": 1.1761292739720375e-07,
      "learning_rate": 1.0459283368792845e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20619888,
      "step": 36305
    },
    {
      "epoch": 637.0176991150443,
      "grad_norm": 1.2664003179452266e-07,
      "learning_rate": 1.043120190818858e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20622528,
      "step": 36310
    },
    {
      "epoch": 637.1061946902655,
      "grad_norm": 1.1837994406960206e-07,
      "learning_rate": 1.0403157392103596e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20624848,
      "step": 36315
    },
    {
      "epoch": 637.1946902654868,
      "grad_norm": 9.74056106883836e-08,
      "learning_rate": 1.0375149824862735e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20628256,
      "step": 36320
    },
    {
      "epoch": 637.2831858407079,
      "grad_norm": 2.74576052561315e-07,
      "learning_rate": 1.034717921078507e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20631040,
      "step": 36325
    },
    {
      "epoch": 637.3716814159292,
      "grad_norm": 1.0374248660127705e-07,
      "learning_rate": 1.0319245554184009e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20634352,
      "step": 36330
    },
    {
      "epoch": 637.4601769911504,
      "grad_norm": 2.013626669850055e-07,
      "learning_rate": 1.0291348859367361e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20637104,
      "step": 36335
    },
    {
      "epoch": 637.5486725663717,
      "grad_norm": 9.16928470928724e-08,
      "learning_rate": 1.0263489130637016e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20640240,
      "step": 36340
    },
    {
      "epoch": 637.637168141593,
      "grad_norm": 1.0220566082352889e-07,
      "learning_rate": 1.0235666372289427e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20643136,
      "step": 36345
    },
    {
      "epoch": 637.7256637168142,
      "grad_norm": 1.3271098850964336e-06,
      "learning_rate": 1.0207880588615076e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20646000,
      "step": 36350
    },
    {
      "epoch": 637.8141592920354,
      "grad_norm": 1.1106939155069995e-06,
      "learning_rate": 1.0180131783898984e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20648592,
      "step": 36355
    },
    {
      "epoch": 637.9026548672566,
      "grad_norm": 4.414640898176003e-06,
      "learning_rate": 1.0152419962420362e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20651536,
      "step": 36360
    },
    {
      "epoch": 637.9911504424779,
      "grad_norm": 6.586541445585681e-08,
      "learning_rate": 1.0124745128452685e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20654448,
      "step": 36365
    },
    {
      "epoch": 638.070796460177,
      "grad_norm": 1.8443600424689066e-07,
      "learning_rate": 1.0097107286263758e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20657120,
      "step": 36370
    },
    {
      "epoch": 638.1592920353983,
      "grad_norm": 1.1371957953087986e-06,
      "learning_rate": 1.00695064401157e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20659856,
      "step": 36375
    },
    {
      "epoch": 638.2477876106195,
      "grad_norm": 1.1717087033957796e-07,
      "learning_rate": 1.0041942594264886e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20662240,
      "step": 36380
    },
    {
      "epoch": 638.3362831858407,
      "grad_norm": 4.647542937163962e-07,
      "learning_rate": 1.001441575296208e-06,
      "loss": 0.0,
      "num_input_tokens_seen": 20664880,
      "step": 36385
    },
    {
      "epoch": 638.4247787610619,
      "grad_norm": 3.531849017690547e-07,
      "learning_rate": 9.986925920452139e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20668016,
      "step": 36390
    },
    {
      "epoch": 638.5132743362832,
      "grad_norm": 1.1886809261341114e-05,
      "learning_rate": 9.959473100974475e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20670928,
      "step": 36395
    },
    {
      "epoch": 638.6017699115044,
      "grad_norm": 2.996459329551726e-07,
      "learning_rate": 9.932057298762564e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20673984,
      "step": 36400
    },
    {
      "epoch": 638.6017699115044,
      "eval_loss": 0.8988261818885803,
      "eval_runtime": 1.0641,
      "eval_samples_per_second": 23.493,
      "eval_steps_per_second": 12.216,
      "num_input_tokens_seen": 20673984,
      "step": 36400
    },
    {
      "epoch": 638.6902654867257,
      "grad_norm": 1.4855085339604557e-07,
      "learning_rate": 9.90467851804433e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20677056,
      "step": 36405
    },
    {
      "epoch": 638.7787610619469,
      "grad_norm": 7.34993221840341e-08,
      "learning_rate": 9.877336763041895e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20680048,
      "step": 36410
    },
    {
      "epoch": 638.8672566371681,
      "grad_norm": 1.1892890938725031e-07,
      "learning_rate": 9.850032037971662e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20682864,
      "step": 36415
    },
    {
      "epoch": 638.9557522123894,
      "grad_norm": 4.162282891684299e-07,
      "learning_rate": 9.822764347044406e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20685248,
      "step": 36420
    },
    {
      "epoch": 639.0353982300885,
      "grad_norm": 4.580408301535499e-07,
      "learning_rate": 9.795533694465175e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20687624,
      "step": 36425
    },
    {
      "epoch": 639.1238938053098,
      "grad_norm": 1.2081758882231952e-07,
      "learning_rate": 9.768340084433197e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20689992,
      "step": 36430
    },
    {
      "epoch": 639.212389380531,
      "grad_norm": 1.1155751167279959e-07,
      "learning_rate": 9.741183521142143e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20692520,
      "step": 36435
    },
    {
      "epoch": 639.3008849557522,
      "grad_norm": 4.3725566456487286e-07,
      "learning_rate": 9.714064008779889e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20695176,
      "step": 36440
    },
    {
      "epoch": 639.3893805309734,
      "grad_norm": 1.525485657793979e-07,
      "learning_rate": 9.686981551528584e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20698584,
      "step": 36445
    },
    {
      "epoch": 639.4778761061947,
      "grad_norm": 9.662478532845853e-07,
      "learning_rate": 9.65993615356467e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20701416,
      "step": 36450
    },
    {
      "epoch": 639.566371681416,
      "grad_norm": 2.7384078293835046e-07,
      "learning_rate": 9.632927819058917e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20704680,
      "step": 36455
    },
    {
      "epoch": 639.6548672566372,
      "grad_norm": 1.0356074398032433e-07,
      "learning_rate": 9.605956552176305e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20707304,
      "step": 36460
    },
    {
      "epoch": 639.7433628318585,
      "grad_norm": 3.1183083137875656e-07,
      "learning_rate": 9.579022357076223e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20710280,
      "step": 36465
    },
    {
      "epoch": 639.8318584070796,
      "grad_norm": 1.9497886682984245e-07,
      "learning_rate": 9.552125237912158e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20713256,
      "step": 36470
    },
    {
      "epoch": 639.9203539823009,
      "grad_norm": 1.3846299395936512e-07,
      "learning_rate": 9.525265198832096e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20716728,
      "step": 36475
    },
    {
      "epoch": 640.0,
      "grad_norm": 1.048256308422424e-06,
      "learning_rate": 9.498442243978112e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20718920,
      "step": 36480
    },
    {
      "epoch": 640.0884955752213,
      "grad_norm": 7.092597797964117e-08,
      "learning_rate": 9.471656377486649e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20721464,
      "step": 36485
    },
    {
      "epoch": 640.1769911504425,
      "grad_norm": 1.9609073831361457e-07,
      "learning_rate": 9.444907603488456e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20724536,
      "step": 36490
    },
    {
      "epoch": 640.2654867256637,
      "grad_norm": 9.662474553806533e-08,
      "learning_rate": 9.418195926108514e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20727496,
      "step": 36495
    },
    {
      "epoch": 640.3539823008849,
      "grad_norm": 1.7976566368815838e-07,
      "learning_rate": 9.391521349466053e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20730328,
      "step": 36500
    },
    {
      "epoch": 640.4424778761062,
      "grad_norm": 1.3437878010336135e-07,
      "learning_rate": 9.364883877674758e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20732840,
      "step": 36505
    },
    {
      "epoch": 640.5309734513274,
      "grad_norm": 6.379013228752228e-08,
      "learning_rate": 9.33828351484231e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20735528,
      "step": 36510
    },
    {
      "epoch": 640.6194690265487,
      "grad_norm": 6.518506978636651e-08,
      "learning_rate": 9.311720265070906e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20738520,
      "step": 36515
    },
    {
      "epoch": 640.70796460177,
      "grad_norm": 1.0883642431736007e-07,
      "learning_rate": 9.285194132456931e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20741624,
      "step": 36520
    },
    {
      "epoch": 640.7964601769911,
      "grad_norm": 2.7180067263543606e-07,
      "learning_rate": 9.258705121091032e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20745176,
      "step": 36525
    },
    {
      "epoch": 640.8849557522124,
      "grad_norm": 1.0233533203063416e-06,
      "learning_rate": 9.232253235058136e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20748024,
      "step": 36530
    },
    {
      "epoch": 640.9734513274336,
      "grad_norm": 1.5631755445610906e-07,
      "learning_rate": 9.205838478437478e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20750584,
      "step": 36535
    },
    {
      "epoch": 641.0530973451328,
      "grad_norm": 5.457477072923211e-06,
      "learning_rate": 9.179460855302524e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20752816,
      "step": 36540
    },
    {
      "epoch": 641.141592920354,
      "grad_norm": 1.3834990397754154e-07,
      "learning_rate": 9.153120369721046e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20755440,
      "step": 36545
    },
    {
      "epoch": 641.2300884955753,
      "grad_norm": 1.1273773026232448e-07,
      "learning_rate": 9.126817025755103e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20757968,
      "step": 36550
    },
    {
      "epoch": 641.3185840707964,
      "grad_norm": 5.817334880475755e-08,
      "learning_rate": 9.100550827460947e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20761344,
      "step": 36555
    },
    {
      "epoch": 641.4070796460177,
      "grad_norm": 2.0908258591134654e-07,
      "learning_rate": 9.0743217788892e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20763728,
      "step": 36560
    },
    {
      "epoch": 641.4955752212389,
      "grad_norm": 7.140460184018593e-06,
      "learning_rate": 9.048129884084683e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20766512,
      "step": 36565
    },
    {
      "epoch": 641.5840707964602,
      "grad_norm": 1.0665574734503025e-07,
      "learning_rate": 9.021975147086553e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20769168,
      "step": 36570
    },
    {
      "epoch": 641.6725663716815,
      "grad_norm": 1.045041173597383e-07,
      "learning_rate": 8.995857571928141e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20771456,
      "step": 36575
    },
    {
      "epoch": 641.7610619469026,
      "grad_norm": 3.001303639393882e-06,
      "learning_rate": 8.969777162637139e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20775248,
      "step": 36580
    },
    {
      "epoch": 641.8495575221239,
      "grad_norm": 1.456541269817535e-07,
      "learning_rate": 8.943733923235525e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20778432,
      "step": 36585
    },
    {
      "epoch": 641.9380530973451,
      "grad_norm": 1.4465197750723746e-07,
      "learning_rate": 8.917727857739394e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20781216,
      "step": 36590
    },
    {
      "epoch": 642.0176991150443,
      "grad_norm": 4.793644166056765e-07,
      "learning_rate": 8.891758970159258e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20783856,
      "step": 36595
    },
    {
      "epoch": 642.1061946902655,
      "grad_norm": 9.282291557610733e-08,
      "learning_rate": 8.86582726449986e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20786240,
      "step": 36600
    },
    {
      "epoch": 642.1061946902655,
      "eval_loss": 0.8976834416389465,
      "eval_runtime": 1.0672,
      "eval_samples_per_second": 23.426,
      "eval_steps_per_second": 12.182,
      "num_input_tokens_seen": 20786240,
      "step": 36600
    },
    {
      "epoch": 642.1946902654868,
      "grad_norm": 9.182621170111815e-07,
      "learning_rate": 8.839932744760165e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20789136,
      "step": 36605
    },
    {
      "epoch": 642.2831858407079,
      "grad_norm": 7.534876544923463e-08,
      "learning_rate": 8.814075414933482e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20792432,
      "step": 36610
    },
    {
      "epoch": 642.3716814159292,
      "grad_norm": 8.320699862451875e-08,
      "learning_rate": 8.788255279007257e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20795424,
      "step": 36615
    },
    {
      "epoch": 642.4601769911504,
      "grad_norm": 9.540327283730221e-08,
      "learning_rate": 8.762472340963362e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20798112,
      "step": 36620
    },
    {
      "epoch": 642.5486725663717,
      "grad_norm": 1.0889054635754292e-07,
      "learning_rate": 8.736726604777811e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20801360,
      "step": 36625
    },
    {
      "epoch": 642.637168141593,
      "grad_norm": 2.377039891143795e-06,
      "learning_rate": 8.711018074420901e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20804464,
      "step": 36630
    },
    {
      "epoch": 642.7256637168142,
      "grad_norm": 8.667367978887341e-08,
      "learning_rate": 8.685346753857209e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20807040,
      "step": 36635
    },
    {
      "epoch": 642.8141592920354,
      "grad_norm": 1.378676017793623e-07,
      "learning_rate": 8.659712647045654e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20809696,
      "step": 36640
    },
    {
      "epoch": 642.9026548672566,
      "grad_norm": 8.17403986275167e-07,
      "learning_rate": 8.634115757939209e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20812320,
      "step": 36645
    },
    {
      "epoch": 642.9911504424779,
      "grad_norm": 3.814965907622536e-07,
      "learning_rate": 8.608556090485387e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20814832,
      "step": 36650
    },
    {
      "epoch": 643.070796460177,
      "grad_norm": 7.597661522140697e-08,
      "learning_rate": 8.583033648625671e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20817328,
      "step": 36655
    },
    {
      "epoch": 643.1592920353983,
      "grad_norm": 9.94432411971502e-08,
      "learning_rate": 8.557548436295998e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20820272,
      "step": 36660
    },
    {
      "epoch": 643.2477876106195,
      "grad_norm": 1.9364040326763643e-07,
      "learning_rate": 8.532100457426556e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20822992,
      "step": 36665
    },
    {
      "epoch": 643.3362831858407,
      "grad_norm": 1.432369174381165e-07,
      "learning_rate": 8.506689715941679e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20825984,
      "step": 36670
    },
    {
      "epoch": 643.4247787610619,
      "grad_norm": 1.0822818552469471e-07,
      "learning_rate": 8.481316215760011e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20828704,
      "step": 36675
    },
    {
      "epoch": 643.5132743362832,
      "grad_norm": 1.160236280384197e-07,
      "learning_rate": 8.455979960794558e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20831696,
      "step": 36680
    },
    {
      "epoch": 643.6017699115044,
      "grad_norm": 1.098495374662889e-07,
      "learning_rate": 8.430680954952364e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20834800,
      "step": 36685
    },
    {
      "epoch": 643.6902654867257,
      "grad_norm": 1.4732300712694268e-07,
      "learning_rate": 8.405419202134974e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20837744,
      "step": 36690
    },
    {
      "epoch": 643.7787610619469,
      "grad_norm": 1.5511844253524032e-07,
      "learning_rate": 8.380194706237993e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20840480,
      "step": 36695
    },
    {
      "epoch": 643.8672566371681,
      "grad_norm": 1.2826646411667753e-07,
      "learning_rate": 8.355007471151366e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20843088,
      "step": 36700
    },
    {
      "epoch": 643.9557522123894,
      "grad_norm": 1.2768525436968048e-07,
      "learning_rate": 8.329857500759292e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20846304,
      "step": 36705
    },
    {
      "epoch": 644.0353982300885,
      "grad_norm": 8.138692919601453e-07,
      "learning_rate": 8.304744798940194e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20848320,
      "step": 36710
    },
    {
      "epoch": 644.1238938053098,
      "grad_norm": 1.9558818564746616e-07,
      "learning_rate": 8.279669369566756e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20851008,
      "step": 36715
    },
    {
      "epoch": 644.212389380531,
      "grad_norm": 9.476056561652513e-08,
      "learning_rate": 8.254631216505993e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20853856,
      "step": 36720
    },
    {
      "epoch": 644.3008849557522,
      "grad_norm": 4.297559996757627e-07,
      "learning_rate": 8.229630343619038e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20857056,
      "step": 36725
    },
    {
      "epoch": 644.3893805309734,
      "grad_norm": 2.1175989672883588e-07,
      "learning_rate": 8.204666754761392e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20860064,
      "step": 36730
    },
    {
      "epoch": 644.4778761061947,
      "grad_norm": 6.144140911601426e-07,
      "learning_rate": 8.179740453782669e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20863216,
      "step": 36735
    },
    {
      "epoch": 644.566371681416,
      "grad_norm": 1.5462883595773746e-07,
      "learning_rate": 8.154851444526907e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20866288,
      "step": 36740
    },
    {
      "epoch": 644.6548672566372,
      "grad_norm": 6.104161798248242e-08,
      "learning_rate": 8.129999730832283e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20868816,
      "step": 36745
    },
    {
      "epoch": 644.7433628318585,
      "grad_norm": 2.1846376796474942e-07,
      "learning_rate": 8.105185316531178e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20871056,
      "step": 36750
    },
    {
      "epoch": 644.8318584070796,
      "grad_norm": 4.917579190077959e-06,
      "learning_rate": 8.08040820545039e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20874096,
      "step": 36755
    },
    {
      "epoch": 644.9203539823009,
      "grad_norm": 1.645639287062295e-07,
      "learning_rate": 8.055668401410782e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20877168,
      "step": 36760
    },
    {
      "epoch": 645.0,
      "grad_norm": 3.516971673889202e-07,
      "learning_rate": 8.030965908227578e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20879464,
      "step": 36765
    },
    {
      "epoch": 645.0884955752213,
      "grad_norm": 1.5704220857060136e-07,
      "learning_rate": 8.006300729710203e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20882296,
      "step": 36770
    },
    {
      "epoch": 645.1769911504425,
      "grad_norm": 2.4971504899440333e-06,
      "learning_rate": 7.981672869662337e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20885384,
      "step": 36775
    },
    {
      "epoch": 645.2654867256637,
      "grad_norm": 1.3883571625683544e-07,
      "learning_rate": 7.957082331881888e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20887736,
      "step": 36780
    },
    {
      "epoch": 645.3539823008849,
      "grad_norm": 1.1221342788303446e-07,
      "learning_rate": 7.932529120161069e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20890680,
      "step": 36785
    },
    {
      "epoch": 645.4424778761062,
      "grad_norm": 5.038025392423151e-06,
      "learning_rate": 7.908013238286243e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20893480,
      "step": 36790
    },
    {
      "epoch": 645.5309734513274,
      "grad_norm": 8.913816884614789e-08,
      "learning_rate": 7.883534690038136e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20896312,
      "step": 36795
    },
    {
      "epoch": 645.6194690265487,
      "grad_norm": 5.771398150500318e-07,
      "learning_rate": 7.859093479191559e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20899128,
      "step": 36800
    },
    {
      "epoch": 645.6194690265487,
      "eval_loss": 0.8956204652786255,
      "eval_runtime": 1.0717,
      "eval_samples_per_second": 23.327,
      "eval_steps_per_second": 12.13,
      "num_input_tokens_seen": 20899128,
      "step": 36800
    },
    {
      "epoch": 645.70796460177,
      "grad_norm": 3.7800455743308703e-07,
      "learning_rate": 7.834689609515722e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20901880,
      "step": 36805
    },
    {
      "epoch": 645.7964601769911,
      "grad_norm": 1.5143402265493933e-07,
      "learning_rate": 7.810323084774002e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20904840,
      "step": 36810
    },
    {
      "epoch": 645.8849557522124,
      "grad_norm": 4.352115752226382e-07,
      "learning_rate": 7.785993908723976e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20907688,
      "step": 36815
    },
    {
      "epoch": 645.9734513274336,
      "grad_norm": 6.73343976131946e-08,
      "learning_rate": 7.761702085117534e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20911096,
      "step": 36820
    },
    {
      "epoch": 646.0530973451328,
      "grad_norm": 2.134213303861543e-07,
      "learning_rate": 7.737447617700844e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20913912,
      "step": 36825
    },
    {
      "epoch": 646.141592920354,
      "grad_norm": 1.1360877039123807e-07,
      "learning_rate": 7.713230510214136e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20916776,
      "step": 36830
    },
    {
      "epoch": 646.2300884955753,
      "grad_norm": 2.642790093432268e-07,
      "learning_rate": 7.689050766392092e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20919736,
      "step": 36835
    },
    {
      "epoch": 646.3185840707964,
      "grad_norm": 1.620509948452309e-07,
      "learning_rate": 7.664908389963477e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20922872,
      "step": 36840
    },
    {
      "epoch": 646.4070796460177,
      "grad_norm": 2.6822300469575566e-07,
      "learning_rate": 7.64080338465134e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20925352,
      "step": 36845
    },
    {
      "epoch": 646.4955752212389,
      "grad_norm": 1.3366051462071482e-05,
      "learning_rate": 7.616735754173043e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20928024,
      "step": 36850
    },
    {
      "epoch": 646.5840707964602,
      "grad_norm": 9.073983164853416e-08,
      "learning_rate": 7.592705502240005e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20930792,
      "step": 36855
    },
    {
      "epoch": 646.6725663716815,
      "grad_norm": 9.270191014820739e-08,
      "learning_rate": 7.568712632558095e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20933800,
      "step": 36860
    },
    {
      "epoch": 646.7610619469026,
      "grad_norm": 1.0445514675438972e-07,
      "learning_rate": 7.544757148827297e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20936744,
      "step": 36865
    },
    {
      "epoch": 646.8495575221239,
      "grad_norm": 1.4015911631304334e-07,
      "learning_rate": 7.520839054741797e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20939656,
      "step": 36870
    },
    {
      "epoch": 646.9380530973451,
      "grad_norm": 1.3865277992408664e-07,
      "learning_rate": 7.496958353990113e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20942680,
      "step": 36875
    },
    {
      "epoch": 647.0176991150443,
      "grad_norm": 6.707866617716718e-08,
      "learning_rate": 7.473115050254941e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20944744,
      "step": 36880
    },
    {
      "epoch": 647.1061946902655,
      "grad_norm": 1.305019168285071e-07,
      "learning_rate": 7.449309147213173e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20947528,
      "step": 36885
    },
    {
      "epoch": 647.1946902654868,
      "grad_norm": 2.182054572585912e-07,
      "learning_rate": 7.425540648536067e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20950200,
      "step": 36890
    },
    {
      "epoch": 647.2831858407079,
      "grad_norm": 9.025704343912366e-07,
      "learning_rate": 7.40180955788894e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20952680,
      "step": 36895
    },
    {
      "epoch": 647.3716814159292,
      "grad_norm": 2.0173207815332717e-07,
      "learning_rate": 7.378115878931474e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20955272,
      "step": 36900
    },
    {
      "epoch": 647.4601769911504,
      "grad_norm": 1.8514785438128456e-07,
      "learning_rate": 7.354459615317527e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20958232,
      "step": 36905
    },
    {
      "epoch": 647.5486725663717,
      "grad_norm": 4.120880930713611e-06,
      "learning_rate": 7.33084077069518e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20961640,
      "step": 36910
    },
    {
      "epoch": 647.637168141593,
      "grad_norm": 1.288628368456557e-07,
      "learning_rate": 7.307259348706768e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20964552,
      "step": 36915
    },
    {
      "epoch": 647.7256637168142,
      "grad_norm": 1.3345335503345268e-07,
      "learning_rate": 7.283715352988801e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20967304,
      "step": 36920
    },
    {
      "epoch": 647.8141592920354,
      "grad_norm": 5.104792535348679e-07,
      "learning_rate": 7.260208787172068e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20970504,
      "step": 36925
    },
    {
      "epoch": 647.9026548672566,
      "grad_norm": 1.4358850819462532e-07,
      "learning_rate": 7.23673965488167e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20973240,
      "step": 36930
    },
    {
      "epoch": 647.9911504424779,
      "grad_norm": 8.83858461975251e-08,
      "learning_rate": 7.213307959736709e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20976248,
      "step": 36935
    },
    {
      "epoch": 648.070796460177,
      "grad_norm": 1.836457670378877e-07,
      "learning_rate": 7.189913705350715e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20978880,
      "step": 36940
    },
    {
      "epoch": 648.1592920353983,
      "grad_norm": 2.995254249071877e-07,
      "learning_rate": 7.166556895331411e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20981456,
      "step": 36945
    },
    {
      "epoch": 648.2477876106195,
      "grad_norm": 1.8799029533056455e-07,
      "learning_rate": 7.143237533280639e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20984208,
      "step": 36950
    },
    {
      "epoch": 648.3362831858407,
      "grad_norm": 7.888250763699034e-08,
      "learning_rate": 7.119955622794578e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20986896,
      "step": 36955
    },
    {
      "epoch": 648.4247787610619,
      "grad_norm": 5.959467230809423e-08,
      "learning_rate": 7.096711167463577e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20989504,
      "step": 36960
    },
    {
      "epoch": 648.5132743362832,
      "grad_norm": 8.227468839550056e-08,
      "learning_rate": 7.073504170872213e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20992400,
      "step": 36965
    },
    {
      "epoch": 648.6017699115044,
      "grad_norm": 9.345773150926107e-07,
      "learning_rate": 7.05033463659932e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20995376,
      "step": 36970
    },
    {
      "epoch": 648.6902654867257,
      "grad_norm": 1.309173967456445e-06,
      "learning_rate": 7.027202568217928e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 20998352,
      "step": 36975
    },
    {
      "epoch": 648.7787610619469,
      "grad_norm": 2.329526012090355e-07,
      "learning_rate": 7.004107969295293e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21001280,
      "step": 36980
    },
    {
      "epoch": 648.8672566371681,
      "grad_norm": 9.233166053945752e-08,
      "learning_rate": 6.9810508433929e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21004672,
      "step": 36985
    },
    {
      "epoch": 648.9557522123894,
      "grad_norm": 1.2968737905794114e-07,
      "learning_rate": 6.958031194066406e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21007280,
      "step": 36990
    },
    {
      "epoch": 649.0353982300885,
      "grad_norm": 4.340785153544857e-07,
      "learning_rate": 6.935049024865776e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21009448,
      "step": 36995
    },
    {
      "epoch": 649.1238938053098,
      "grad_norm": 1.7429712784178264e-07,
      "learning_rate": 6.912104339335118e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21011928,
      "step": 37000
    },
    {
      "epoch": 649.1238938053098,
      "eval_loss": 0.9296573996543884,
      "eval_runtime": 1.0772,
      "eval_samples_per_second": 23.207,
      "eval_steps_per_second": 12.068,
      "num_input_tokens_seen": 21011928,
      "step": 37000
    },
    {
      "epoch": 649.212389380531,
      "grad_norm": 3.3565058288331784e-07,
      "learning_rate": 6.889197141012799e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21014648,
      "step": 37005
    },
    {
      "epoch": 649.3008849557522,
      "grad_norm": 1.2422441386661376e-06,
      "learning_rate": 6.866327433431435e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21017384,
      "step": 37010
    },
    {
      "epoch": 649.3893805309734,
      "grad_norm": 1.5486986058022012e-06,
      "learning_rate": 6.843495220117735e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21020280,
      "step": 37015
    },
    {
      "epoch": 649.4778761061947,
      "grad_norm": 9.769284048388727e-08,
      "learning_rate": 6.820700504592798e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21023752,
      "step": 37020
    },
    {
      "epoch": 649.566371681416,
      "grad_norm": 1.4057403632250498e-06,
      "learning_rate": 6.797943290371839e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21026904,
      "step": 37025
    },
    {
      "epoch": 649.6548672566372,
      "grad_norm": 1.5496385685764835e-07,
      "learning_rate": 6.775223580964274e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21029224,
      "step": 37030
    },
    {
      "epoch": 649.7433628318585,
      "grad_norm": 1.4619465105170093e-07,
      "learning_rate": 6.7525413798738e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21032296,
      "step": 37035
    },
    {
      "epoch": 649.8318584070796,
      "grad_norm": 9.035488233166689e-07,
      "learning_rate": 6.729896690598259e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21035096,
      "step": 37040
    },
    {
      "epoch": 649.9203539823009,
      "grad_norm": 2.5747922904884035e-07,
      "learning_rate": 6.707289516629772e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21037944,
      "step": 37045
    },
    {
      "epoch": 650.0,
      "grad_norm": 1.087835954649563e-07,
      "learning_rate": 6.684719861454692e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21040632,
      "step": 37050
    },
    {
      "epoch": 650.0884955752213,
      "grad_norm": 1.4057476960260828e-07,
      "learning_rate": 6.662187728553481e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21043736,
      "step": 37055
    },
    {
      "epoch": 650.1769911504425,
      "grad_norm": 1.1551824741218297e-07,
      "learning_rate": 6.639693121400892e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21046376,
      "step": 37060
    },
    {
      "epoch": 650.2654867256637,
      "grad_norm": 9.530851485806124e-08,
      "learning_rate": 6.617236043465868e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21049368,
      "step": 37065
    },
    {
      "epoch": 650.3539823008849,
      "grad_norm": 7.150638339226134e-07,
      "learning_rate": 6.594816498211587e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21051832,
      "step": 37070
    },
    {
      "epoch": 650.4424778761062,
      "grad_norm": 1.3391554603003897e-05,
      "learning_rate": 6.572434489095447e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21055064,
      "step": 37075
    },
    {
      "epoch": 650.5309734513274,
      "grad_norm": 1.2399939919305325e-07,
      "learning_rate": 6.550090019568994e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21057800,
      "step": 37080
    },
    {
      "epoch": 650.6194690265487,
      "grad_norm": 7.62111795893361e-08,
      "learning_rate": 6.527783093078027e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21061336,
      "step": 37085
    },
    {
      "epoch": 650.70796460177,
      "grad_norm": 9.772765707793951e-08,
      "learning_rate": 6.5055137130626e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21064280,
      "step": 37090
    },
    {
      "epoch": 650.7964601769911,
      "grad_norm": 2.7658143153530546e-07,
      "learning_rate": 6.483281882956854e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21066888,
      "step": 37095
    },
    {
      "epoch": 650.8849557522124,
      "grad_norm": 2.675533323781565e-06,
      "learning_rate": 6.461087606189298e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21069544,
      "step": 37100
    },
    {
      "epoch": 650.9734513274336,
      "grad_norm": 1.0696747665406292e-07,
      "learning_rate": 6.438930886182554e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21072312,
      "step": 37105
    },
    {
      "epoch": 651.0530973451328,
      "grad_norm": 4.6025363076296344e-07,
      "learning_rate": 6.416811726353417e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21074672,
      "step": 37110
    },
    {
      "epoch": 651.141592920354,
      "grad_norm": 1.9921638738651382e-07,
      "learning_rate": 6.394730130112991e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21077696,
      "step": 37115
    },
    {
      "epoch": 651.2300884955753,
      "grad_norm": 5.497233814821811e-07,
      "learning_rate": 6.372686100866471e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21080592,
      "step": 37120
    },
    {
      "epoch": 651.3185840707964,
      "grad_norm": 1.324052760764971e-07,
      "learning_rate": 6.350679642013413e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21083920,
      "step": 37125
    },
    {
      "epoch": 651.4070796460177,
      "grad_norm": 1.8897100062531536e-07,
      "learning_rate": 6.328710756947437e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21086736,
      "step": 37130
    },
    {
      "epoch": 651.4955752212389,
      "grad_norm": 7.108451427484397e-08,
      "learning_rate": 6.306779449056416e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21088992,
      "step": 37135
    },
    {
      "epoch": 651.5840707964602,
      "grad_norm": 5.2523436266938006e-08,
      "learning_rate": 6.284885721722422e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21092416,
      "step": 37140
    },
    {
      "epoch": 651.6725663716815,
      "grad_norm": 3.5470921488922613e-07,
      "learning_rate": 6.26302957832181e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21095232,
      "step": 37145
    },
    {
      "epoch": 651.7610619469026,
      "grad_norm": 1.281394048646689e-07,
      "learning_rate": 6.241211022224997e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21098064,
      "step": 37150
    },
    {
      "epoch": 651.8495575221239,
      "grad_norm": 1.1456305060164595e-07,
      "learning_rate": 6.219430056796732e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21101136,
      "step": 37155
    },
    {
      "epoch": 651.9380530973451,
      "grad_norm": 8.356431635547779e-08,
      "learning_rate": 6.19768668539586e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21104112,
      "step": 37160
    },
    {
      "epoch": 652.0176991150443,
      "grad_norm": 7.097838050640348e-08,
      "learning_rate": 6.175980911375528e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21106448,
      "step": 37165
    },
    {
      "epoch": 652.1061946902655,
      "grad_norm": 1.8694406378472195e-07,
      "learning_rate": 6.154312738083034e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21109408,
      "step": 37170
    },
    {
      "epoch": 652.1946902654868,
      "grad_norm": 8.171787158062216e-08,
      "learning_rate": 6.132682168859843e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21112464,
      "step": 37175
    },
    {
      "epoch": 652.2831858407079,
      "grad_norm": 1.6101381561384187e-07,
      "learning_rate": 6.111089207041704e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21115408,
      "step": 37180
    },
    {
      "epoch": 652.3716814159292,
      "grad_norm": 1.8608714924539527e-07,
      "learning_rate": 6.089533855958507e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21118480,
      "step": 37185
    },
    {
      "epoch": 652.4601769911504,
      "grad_norm": 1.3067814279565937e-06,
      "learning_rate": 6.068016118934372e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21121104,
      "step": 37190
    },
    {
      "epoch": 652.5486725663717,
      "grad_norm": 4.103460469195852e-07,
      "learning_rate": 6.04653599928759e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21124112,
      "step": 37195
    },
    {
      "epoch": 652.637168141593,
      "grad_norm": 1.0476142620063911e-07,
      "learning_rate": 6.025093500330675e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21126880,
      "step": 37200
    },
    {
      "epoch": 652.637168141593,
      "eval_loss": 0.8969581127166748,
      "eval_runtime": 1.071,
      "eval_samples_per_second": 23.343,
      "eval_steps_per_second": 12.138,
      "num_input_tokens_seen": 21126880,
      "step": 37200
    },
    {
      "epoch": 652.7256637168142,
      "grad_norm": 7.572336926386924e-07,
      "learning_rate": 6.003688625370291e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21130272,
      "step": 37205
    },
    {
      "epoch": 652.8141592920354,
      "grad_norm": 1.2591917197823932e-07,
      "learning_rate": 5.982321377707406e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21132608,
      "step": 37210
    },
    {
      "epoch": 652.9026548672566,
      "grad_norm": 8.08986087008634e-08,
      "learning_rate": 5.96099176063708e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21135184,
      "step": 37215
    },
    {
      "epoch": 652.9911504424779,
      "grad_norm": 4.104085746803321e-07,
      "learning_rate": 5.93969977744857e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21138096,
      "step": 37220
    },
    {
      "epoch": 653.070796460177,
      "grad_norm": 2.1789719539810903e-07,
      "learning_rate": 5.918445431425445e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21140472,
      "step": 37225
    },
    {
      "epoch": 653.1592920353983,
      "grad_norm": 7.176028304911597e-08,
      "learning_rate": 5.897228725845333e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21142856,
      "step": 37230
    },
    {
      "epoch": 653.2477876106195,
      "grad_norm": 1.1881775918709536e-07,
      "learning_rate": 5.876049663980171e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21145688,
      "step": 37235
    },
    {
      "epoch": 653.3362831858407,
      "grad_norm": 1.365208390780026e-07,
      "learning_rate": 5.854908249095959e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21148728,
      "step": 37240
    },
    {
      "epoch": 653.4247787610619,
      "grad_norm": 9.661810906891333e-08,
      "learning_rate": 5.833804484453031e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21151800,
      "step": 37245
    },
    {
      "epoch": 653.5132743362832,
      "grad_norm": 1.2566822249482357e-07,
      "learning_rate": 5.81273837330587e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21154456,
      "step": 37250
    },
    {
      "epoch": 653.6017699115044,
      "grad_norm": 2.198181903168006e-07,
      "learning_rate": 5.791709918903071e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21157192,
      "step": 37255
    },
    {
      "epoch": 653.6902654867257,
      "grad_norm": 1.4024421091107797e-07,
      "learning_rate": 5.770719124487483e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21160504,
      "step": 37260
    },
    {
      "epoch": 653.7787610619469,
      "grad_norm": 7.570465072603838e-07,
      "learning_rate": 5.749765993296241e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21163080,
      "step": 37265
    },
    {
      "epoch": 653.8672566371681,
      "grad_norm": 8.526429695621118e-08,
      "learning_rate": 5.728850528560509e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21166392,
      "step": 37270
    },
    {
      "epoch": 653.9557522123894,
      "grad_norm": 2.624081787416799e-07,
      "learning_rate": 5.707972733505707e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21169320,
      "step": 37275
    },
    {
      "epoch": 654.0353982300885,
      "grad_norm": 6.883222170017689e-08,
      "learning_rate": 5.687132611351509e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21171848,
      "step": 37280
    },
    {
      "epoch": 654.1238938053098,
      "grad_norm": 9.150156188297842e-08,
      "learning_rate": 5.666330165311651e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21174520,
      "step": 37285
    },
    {
      "epoch": 654.212389380531,
      "grad_norm": 3.564062183158967e-07,
      "learning_rate": 5.645565398594204e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21177896,
      "step": 37290
    },
    {
      "epoch": 654.3008849557522,
      "grad_norm": 1.5167790934356162e-07,
      "learning_rate": 5.624838314401304e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21181176,
      "step": 37295
    },
    {
      "epoch": 654.3893805309734,
      "grad_norm": 2.448263387577754e-07,
      "learning_rate": 5.604148915929336e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21184296,
      "step": 37300
    },
    {
      "epoch": 654.4778761061947,
      "grad_norm": 5.410754965851083e-06,
      "learning_rate": 5.583497206368887e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21186808,
      "step": 37305
    },
    {
      "epoch": 654.566371681416,
      "grad_norm": 2.8565460752361105e-07,
      "learning_rate": 5.562883188904688e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21189352,
      "step": 37310
    },
    {
      "epoch": 654.6548672566372,
      "grad_norm": 1.470006054660189e-07,
      "learning_rate": 5.542306866715724e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21192088,
      "step": 37315
    },
    {
      "epoch": 654.7433628318585,
      "grad_norm": 9.365550823758895e-08,
      "learning_rate": 5.52176824297504e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21195112,
      "step": 37320
    },
    {
      "epoch": 654.8318584070796,
      "grad_norm": 1.1090992302342784e-07,
      "learning_rate": 5.501267320850018e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21197720,
      "step": 37325
    },
    {
      "epoch": 654.9203539823009,
      "grad_norm": 1.1803257393694366e-07,
      "learning_rate": 5.480804103502157e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21200888,
      "step": 37330
    },
    {
      "epoch": 655.0,
      "grad_norm": 3.84463483271702e-08,
      "learning_rate": 5.460378594087101e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21202968,
      "step": 37335
    },
    {
      "epoch": 655.0884955752213,
      "grad_norm": 1.4655044822120544e-07,
      "learning_rate": 5.439990795754773e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21206328,
      "step": 37340
    },
    {
      "epoch": 655.1769911504425,
      "grad_norm": 1.33038986405154e-06,
      "learning_rate": 5.419640711649188e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21209128,
      "step": 37345
    },
    {
      "epoch": 655.2654867256637,
      "grad_norm": 6.267862886488729e-07,
      "learning_rate": 5.399328344908583e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21211752,
      "step": 37350
    },
    {
      "epoch": 655.3539823008849,
      "grad_norm": 7.182939754102335e-08,
      "learning_rate": 5.379053698665399e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21214664,
      "step": 37355
    },
    {
      "epoch": 655.4424778761062,
      "grad_norm": 1.1651415121605169e-07,
      "learning_rate": 5.358816776046216e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21217400,
      "step": 37360
    },
    {
      "epoch": 655.5309734513274,
      "grad_norm": 2.1048312248694856e-07,
      "learning_rate": 5.338617580171817e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21219896,
      "step": 37365
    },
    {
      "epoch": 655.6194690265487,
      "grad_norm": 6.693529712720192e-07,
      "learning_rate": 5.318456114157239e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21223032,
      "step": 37370
    },
    {
      "epoch": 655.70796460177,
      "grad_norm": 1.345803610774965e-07,
      "learning_rate": 5.298332381111576e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21225592,
      "step": 37375
    },
    {
      "epoch": 655.7964601769911,
      "grad_norm": 1.055247764725209e-07,
      "learning_rate": 5.27824638413818e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21228680,
      "step": 37380
    },
    {
      "epoch": 655.8849557522124,
      "grad_norm": 3.5296463352096907e-07,
      "learning_rate": 5.258198126334546e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21231832,
      "step": 37385
    },
    {
      "epoch": 655.9734513274336,
      "grad_norm": 1.0323652759325341e-06,
      "learning_rate": 5.238187610792367e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21234840,
      "step": 37390
    },
    {
      "epoch": 656.0530973451328,
      "grad_norm": 1.1013983680641104e-07,
      "learning_rate": 5.218214840597563e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21237344,
      "step": 37395
    },
    {
      "epoch": 656.141592920354,
      "grad_norm": 6.72507880494777e-08,
      "learning_rate": 5.198279818830115e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21239760,
      "step": 37400
    },
    {
      "epoch": 656.141592920354,
      "eval_loss": 0.9158565998077393,
      "eval_runtime": 1.0721,
      "eval_samples_per_second": 23.32,
      "eval_steps_per_second": 12.126,
      "num_input_tokens_seen": 21239760,
      "step": 37400
    },
    {
      "epoch": 656.2300884955753,
      "grad_norm": 1.0879937661911754e-07,
      "learning_rate": 5.178382548564287e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21242448,
      "step": 37405
    },
    {
      "epoch": 656.3185840707964,
      "grad_norm": 5.295193616916549e-08,
      "learning_rate": 5.15852303286854e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21245472,
      "step": 37410
    },
    {
      "epoch": 656.4070796460177,
      "grad_norm": 2.6277695042153937e-07,
      "learning_rate": 5.138701274805396e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21248432,
      "step": 37415
    },
    {
      "epoch": 656.4955752212389,
      "grad_norm": 1.1932548886761651e-06,
      "learning_rate": 5.118917277431606e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21251152,
      "step": 37420
    },
    {
      "epoch": 656.5840707964602,
      "grad_norm": 2.60568782550763e-07,
      "learning_rate": 5.099171043798145e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21254528,
      "step": 37425
    },
    {
      "epoch": 656.6725663716815,
      "grad_norm": 1.1206907402083743e-07,
      "learning_rate": 5.079462576950133e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21257216,
      "step": 37430
    },
    {
      "epoch": 656.7610619469026,
      "grad_norm": 1.5215568964777049e-05,
      "learning_rate": 5.059791879926862e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21259872,
      "step": 37435
    },
    {
      "epoch": 656.8495575221239,
      "grad_norm": 1.3141797694515844e-07,
      "learning_rate": 5.040158955761793e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21262784,
      "step": 37440
    },
    {
      "epoch": 656.9380530973451,
      "grad_norm": 8.364229415747104e-07,
      "learning_rate": 5.020563807482559e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21265920,
      "step": 37445
    },
    {
      "epoch": 657.0176991150443,
      "grad_norm": 1.7759542458861688e-07,
      "learning_rate": 5.001006438110995e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21268192,
      "step": 37450
    },
    {
      "epoch": 657.1061946902655,
      "grad_norm": 5.0672181117761284e-08,
      "learning_rate": 4.981486850663075e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21270656,
      "step": 37455
    },
    {
      "epoch": 657.1946902654868,
      "grad_norm": 1.5889553139913914e-07,
      "learning_rate": 4.962005048149005e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21274016,
      "step": 37460
    },
    {
      "epoch": 657.2831858407079,
      "grad_norm": 1.7348405378925236e-07,
      "learning_rate": 4.942561033573073e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21277312,
      "step": 37465
    },
    {
      "epoch": 657.3716814159292,
      "grad_norm": 3.072880190302385e-07,
      "learning_rate": 4.923154809933827e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21280400,
      "step": 37470
    },
    {
      "epoch": 657.4601769911504,
      "grad_norm": 8.888780484994641e-07,
      "learning_rate": 4.903786380223957e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21283216,
      "step": 37475
    },
    {
      "epoch": 657.5486725663717,
      "grad_norm": 4.714308374786924e-07,
      "learning_rate": 4.884455747430266e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21285856,
      "step": 37480
    },
    {
      "epoch": 657.637168141593,
      "grad_norm": 1.1147559320079381e-07,
      "learning_rate": 4.865162914533816e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21288784,
      "step": 37485
    },
    {
      "epoch": 657.7256637168142,
      "grad_norm": 1.1221442264286452e-06,
      "learning_rate": 4.845907884509809e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21291312,
      "step": 37490
    },
    {
      "epoch": 657.8141592920354,
      "grad_norm": 2.7123724066768773e-07,
      "learning_rate": 4.82669066032762e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21294640,
      "step": 37495
    },
    {
      "epoch": 657.9026548672566,
      "grad_norm": 1.2259381776402734e-07,
      "learning_rate": 4.807511244950768e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21297344,
      "step": 37500
    },
    {
      "epoch": 657.9911504424779,
      "grad_norm": 1.3517868637791253e-06,
      "learning_rate": 4.788369641336943e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21299936,
      "step": 37505
    },
    {
      "epoch": 658.070796460177,
      "grad_norm": 9.131223066560779e-08,
      "learning_rate": 4.769265852438032e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21302256,
      "step": 37510
    },
    {
      "epoch": 658.1592920353983,
      "grad_norm": 8.197183376523753e-08,
      "learning_rate": 4.750199881200124e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21305648,
      "step": 37515
    },
    {
      "epoch": 658.2477876106195,
      "grad_norm": 2.0218084273437853e-07,
      "learning_rate": 4.7311717305633664e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21308576,
      "step": 37520
    },
    {
      "epoch": 658.3362831858407,
      "grad_norm": 1.2873383639089298e-06,
      "learning_rate": 4.7121814034621623e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21311584,
      "step": 37525
    },
    {
      "epoch": 658.4247787610619,
      "grad_norm": 7.951269509476333e-08,
      "learning_rate": 4.693228902825114e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21314480,
      "step": 37530
    },
    {
      "epoch": 658.5132743362832,
      "grad_norm": 6.236153353711416e-07,
      "learning_rate": 4.6743142315748277e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21317104,
      "step": 37535
    },
    {
      "epoch": 658.6017699115044,
      "grad_norm": 8.728752476372392e-08,
      "learning_rate": 4.655437392628276e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21319792,
      "step": 37540
    },
    {
      "epoch": 658.6902654867257,
      "grad_norm": 2.803111271987291e-07,
      "learning_rate": 4.636598388896463e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21322816,
      "step": 37545
    },
    {
      "epoch": 658.7787610619469,
      "grad_norm": 2.7567662641558854e-07,
      "learning_rate": 4.6177972232845925e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21325536,
      "step": 37550
    },
    {
      "epoch": 658.8672566371681,
      "grad_norm": 9.095411002135734e-08,
      "learning_rate": 4.5990338986920953e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21328272,
      "step": 37555
    },
    {
      "epoch": 658.9557522123894,
      "grad_norm": 1.034222236739879e-07,
      "learning_rate": 4.5803084180124633e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21331344,
      "step": 37560
    },
    {
      "epoch": 659.0353982300885,
      "grad_norm": 3.4691007044784783e-07,
      "learning_rate": 4.561620784133386e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21333712,
      "step": 37565
    },
    {
      "epoch": 659.1238938053098,
      "grad_norm": 2.673277776921168e-07,
      "learning_rate": 4.5429709999367796e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21336400,
      "step": 37570
    },
    {
      "epoch": 659.212389380531,
      "grad_norm": 6.857485601585722e-08,
      "learning_rate": 4.5243590682986223e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21339440,
      "step": 37575
    },
    {
      "epoch": 659.3008849557522,
      "grad_norm": 1.5474003589588392e-07,
      "learning_rate": 4.5057849920891735e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21342160,
      "step": 37580
    },
    {
      "epoch": 659.3893805309734,
      "grad_norm": 1.1288038876955397e-06,
      "learning_rate": 4.487248774172698e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21345168,
      "step": 37585
    },
    {
      "epoch": 659.4778761061947,
      "grad_norm": 5.344629130377143e-07,
      "learning_rate": 4.4687504174077965e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21347808,
      "step": 37590
    },
    {
      "epoch": 659.566371681416,
      "grad_norm": 3.082022317357769e-07,
      "learning_rate": 4.450289924647133e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21350704,
      "step": 37595
    },
    {
      "epoch": 659.6548672566372,
      "grad_norm": 1.9140553320085019e-07,
      "learning_rate": 4.431867298737513e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21353776,
      "step": 37600
    },
    {
      "epoch": 659.6548672566372,
      "eval_loss": 0.9120059609413147,
      "eval_runtime": 1.066,
      "eval_samples_per_second": 23.452,
      "eval_steps_per_second": 12.195,
      "num_input_tokens_seen": 21353776,
      "step": 37600
    },
    {
      "epoch": 659.7433628318585,
      "grad_norm": 1.1849672887365159e-07,
      "learning_rate": 4.41348254251997e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21356960,
      "step": 37605
    },
    {
      "epoch": 659.8318584070796,
      "grad_norm": 1.6304289829349727e-07,
      "learning_rate": 4.395135658829652e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21359680,
      "step": 37610
    },
    {
      "epoch": 659.9203539823009,
      "grad_norm": 1.6267188129859278e-07,
      "learning_rate": 4.376826650495852e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21362736,
      "step": 37615
    },
    {
      "epoch": 660.0,
      "grad_norm": 1.6995835494526546e-07,
      "learning_rate": 4.358555520342117e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21365112,
      "step": 37620
    },
    {
      "epoch": 660.0884955752213,
      "grad_norm": 1.9487070801460504e-07,
      "learning_rate": 4.3403222711860257e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21368568,
      "step": 37625
    },
    {
      "epoch": 660.1769911504425,
      "grad_norm": 7.189194661805232e-08,
      "learning_rate": 4.3221269058394133e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21371352,
      "step": 37630
    },
    {
      "epoch": 660.2654867256637,
      "grad_norm": 2.9653878641511255e-07,
      "learning_rate": 4.303969427108173e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21373912,
      "step": 37635
    },
    {
      "epoch": 660.3539823008849,
      "grad_norm": 1.1873272853790695e-07,
      "learning_rate": 4.2858498377924825e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21376952,
      "step": 37640
    },
    {
      "epoch": 660.4424778761062,
      "grad_norm": 1.1340429750816838e-07,
      "learning_rate": 4.267768140686579e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21379656,
      "step": 37645
    },
    {
      "epoch": 660.5309734513274,
      "grad_norm": 1.3266055702843005e-06,
      "learning_rate": 4.2497243385788975e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21381944,
      "step": 37650
    },
    {
      "epoch": 660.6194690265487,
      "grad_norm": 9.787361676671935e-08,
      "learning_rate": 4.231718434251991e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21384792,
      "step": 37655
    },
    {
      "epoch": 660.70796460177,
      "grad_norm": 8.371969215659192e-07,
      "learning_rate": 4.213750430482666e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21387288,
      "step": 37660
    },
    {
      "epoch": 660.7964601769911,
      "grad_norm": 9.563697744852107e-08,
      "learning_rate": 4.1958203300417054e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21390072,
      "step": 37665
    },
    {
      "epoch": 660.8849557522124,
      "grad_norm": 1.4354348820688756e-07,
      "learning_rate": 4.177928135694259e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21393304,
      "step": 37670
    },
    {
      "epoch": 660.9734513274336,
      "grad_norm": 2.8655614414674346e-07,
      "learning_rate": 4.1600738501994807e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21396600,
      "step": 37675
    },
    {
      "epoch": 661.0530973451328,
      "grad_norm": 3.020504948381131e-07,
      "learning_rate": 4.1422574763107237e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21398984,
      "step": 37680
    },
    {
      "epoch": 661.141592920354,
      "grad_norm": 8.517937288843314e-08,
      "learning_rate": 4.124479016775512e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21401896,
      "step": 37685
    },
    {
      "epoch": 661.2300884955753,
      "grad_norm": 1.1223167462048877e-07,
      "learning_rate": 4.106738474335514e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21404488,
      "step": 37690
    },
    {
      "epoch": 661.3185840707964,
      "grad_norm": 7.730580620091132e-08,
      "learning_rate": 4.089035851726486e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21407368,
      "step": 37695
    },
    {
      "epoch": 661.4070796460177,
      "grad_norm": 1.7942930696790427e-07,
      "learning_rate": 4.0713711516784937e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21409768,
      "step": 37700
    },
    {
      "epoch": 661.4955752212389,
      "grad_norm": 1.0852149756601648e-07,
      "learning_rate": 4.05374437691558e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21412952,
      "step": 37705
    },
    {
      "epoch": 661.5840707964602,
      "grad_norm": 7.883680552822625e-08,
      "learning_rate": 4.036155530156044e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21415720,
      "step": 37710
    },
    {
      "epoch": 661.6725663716815,
      "grad_norm": 2.2945660305140336e-07,
      "learning_rate": 4.018604614112298e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21418968,
      "step": 37715
    },
    {
      "epoch": 661.7610619469026,
      "grad_norm": 1.422926771965649e-07,
      "learning_rate": 4.0010916314908996e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21421832,
      "step": 37720
    },
    {
      "epoch": 661.8495575221239,
      "grad_norm": 4.24805102738901e-06,
      "learning_rate": 3.983616584992578e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21424504,
      "step": 37725
    },
    {
      "epoch": 661.9380530973451,
      "grad_norm": 5.960288618211962e-08,
      "learning_rate": 3.9661794773122595e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21427624,
      "step": 37730
    },
    {
      "epoch": 662.0176991150443,
      "grad_norm": 9.938471379200564e-08,
      "learning_rate": 3.9487803111388777e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21430248,
      "step": 37735
    },
    {
      "epoch": 662.1061946902655,
      "grad_norm": 8.29429254167735e-08,
      "learning_rate": 3.9314190891556747e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21433352,
      "step": 37740
    },
    {
      "epoch": 662.1946902654868,
      "grad_norm": 2.5878964038383856e-07,
      "learning_rate": 3.914095814039925e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21436120,
      "step": 37745
    },
    {
      "epoch": 662.2831858407079,
      "grad_norm": 2.6125078989025496e-07,
      "learning_rate": 3.896810488463104e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21438856,
      "step": 37750
    },
    {
      "epoch": 662.3716814159292,
      "grad_norm": 8.849025334711769e-08,
      "learning_rate": 3.8795631150908565e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21441576,
      "step": 37755
    },
    {
      "epoch": 662.4601769911504,
      "grad_norm": 1.1219662354733373e-07,
      "learning_rate": 3.862353696582888e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21444552,
      "step": 37760
    },
    {
      "epoch": 662.5486725663717,
      "grad_norm": 3.6191002550367557e-07,
      "learning_rate": 3.8451822355931313e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21447208,
      "step": 37765
    },
    {
      "epoch": 662.637168141593,
      "grad_norm": 8.920666516587517e-08,
      "learning_rate": 3.82804873476969e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21450504,
      "step": 37770
    },
    {
      "epoch": 662.7256637168142,
      "grad_norm": 1.5109931439383217e-07,
      "learning_rate": 3.810953196754702e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21453704,
      "step": 37775
    },
    {
      "epoch": 662.8141592920354,
      "grad_norm": 1.1289326806718236e-07,
      "learning_rate": 3.793895624184529e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21456856,
      "step": 37780
    },
    {
      "epoch": 662.9026548672566,
      "grad_norm": 5.671982421517896e-07,
      "learning_rate": 3.776876019689679e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21459224,
      "step": 37785
    },
    {
      "epoch": 662.9911504424779,
      "grad_norm": 5.4027694318392605e-08,
      "learning_rate": 3.7598943858947743e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21462200,
      "step": 37790
    },
    {
      "epoch": 663.070796460177,
      "grad_norm": 1.1871462390899978e-07,
      "learning_rate": 3.742950725418637e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21464776,
      "step": 37795
    },
    {
      "epoch": 663.1592920353983,
      "grad_norm": 1.0228346525309462e-07,
      "learning_rate": 3.726045040874093e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21467368,
      "step": 37800
    },
    {
      "epoch": 663.1592920353983,
      "eval_loss": 0.8969388604164124,
      "eval_runtime": 1.0604,
      "eval_samples_per_second": 23.577,
      "eval_steps_per_second": 12.26,
      "num_input_tokens_seen": 21467368,
      "step": 37800
    },
    {
      "epoch": 663.2477876106195,
      "grad_norm": 1.2106866051908582e-06,
      "learning_rate": 3.709177334868308e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21469960,
      "step": 37805
    },
    {
      "epoch": 663.3362831858407,
      "grad_norm": 1.6544018421882356e-07,
      "learning_rate": 3.692347610002478e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21473000,
      "step": 37810
    },
    {
      "epoch": 663.4247787610619,
      "grad_norm": 4.652283962514048e-07,
      "learning_rate": 3.675555868871916e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21475912,
      "step": 37815
    },
    {
      "epoch": 663.5132743362832,
      "grad_norm": 1.48313816339396e-07,
      "learning_rate": 3.658802114066162e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21479224,
      "step": 37820
    },
    {
      "epoch": 663.6017699115044,
      "grad_norm": 2.8390522288646025e-07,
      "learning_rate": 3.6420863481688437e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21482200,
      "step": 37825
    },
    {
      "epoch": 663.6902654867257,
      "grad_norm": 1.1784785414192811e-07,
      "learning_rate": 3.625408573757705e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21485160,
      "step": 37830
    },
    {
      "epoch": 663.7787610619469,
      "grad_norm": 8.975916188092015e-08,
      "learning_rate": 3.608768793404743e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21488104,
      "step": 37835
    },
    {
      "epoch": 663.8672566371681,
      "grad_norm": 6.871000834962615e-08,
      "learning_rate": 3.592167009675934e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21490952,
      "step": 37840
    },
    {
      "epoch": 663.9557522123894,
      "grad_norm": 1.2122002601699933e-07,
      "learning_rate": 3.575603225131563e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21493880,
      "step": 37845
    },
    {
      "epoch": 664.0353982300885,
      "grad_norm": 3.3617138228692056e-07,
      "learning_rate": 3.55907744232592e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21496280,
      "step": 37850
    },
    {
      "epoch": 664.1238938053098,
      "grad_norm": 8.677742613372175e-08,
      "learning_rate": 3.5425896638075217e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21499048,
      "step": 37855
    },
    {
      "epoch": 664.212389380531,
      "grad_norm": 5.129855367158598e-07,
      "learning_rate": 3.5261398921189736e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21502184,
      "step": 37860
    },
    {
      "epoch": 664.3008849557522,
      "grad_norm": 1.0495531910237332e-07,
      "learning_rate": 3.509728129797024e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21504856,
      "step": 37865
    },
    {
      "epoch": 664.3893805309734,
      "grad_norm": 1.5591967894579284e-07,
      "learning_rate": 3.4933543793725656e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21508200,
      "step": 37870
    },
    {
      "epoch": 664.4778761061947,
      "grad_norm": 1.2488943923472107e-07,
      "learning_rate": 3.4770186433707163e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21511096,
      "step": 37875
    },
    {
      "epoch": 664.566371681416,
      "grad_norm": 4.6214501026042853e-07,
      "learning_rate": 3.4607209243105453e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21513864,
      "step": 37880
    },
    {
      "epoch": 664.6548672566372,
      "grad_norm": 1.7388565254350397e-07,
      "learning_rate": 3.444461224705431e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21516584,
      "step": 37885
    },
    {
      "epoch": 664.7433628318585,
      "grad_norm": 1.0072047018638841e-07,
      "learning_rate": 3.4282395470628116e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21519432,
      "step": 37890
    },
    {
      "epoch": 664.8318584070796,
      "grad_norm": 1.6040364414493524e-07,
      "learning_rate": 3.4120558938842417e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21522168,
      "step": 37895
    },
    {
      "epoch": 664.9203539823009,
      "grad_norm": 1.2486251534937765e-06,
      "learning_rate": 3.395910267665503e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21525368,
      "step": 37900
    },
    {
      "epoch": 665.0,
      "grad_norm": 2.1842419073436758e-07,
      "learning_rate": 3.3798026708964094e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21527784,
      "step": 37905
    },
    {
      "epoch": 665.0884955752213,
      "grad_norm": 6.651059720752528e-07,
      "learning_rate": 3.3637331060609456e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21530072,
      "step": 37910
    },
    {
      "epoch": 665.1769911504425,
      "grad_norm": 4.851546009376762e-07,
      "learning_rate": 3.3477015756372966e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21532872,
      "step": 37915
    },
    {
      "epoch": 665.2654867256637,
      "grad_norm": 4.577606773636944e-07,
      "learning_rate": 3.3317080820976785e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21535672,
      "step": 37920
    },
    {
      "epoch": 665.3539823008849,
      "grad_norm": 1.5747005477351195e-07,
      "learning_rate": 3.315752627908508e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21538840,
      "step": 37925
    },
    {
      "epoch": 665.4424778761062,
      "grad_norm": 7.710057303711437e-08,
      "learning_rate": 3.299835215530317e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21541464,
      "step": 37930
    },
    {
      "epoch": 665.5309734513274,
      "grad_norm": 9.187035487911999e-08,
      "learning_rate": 3.2839558474177245e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21544216,
      "step": 37935
    },
    {
      "epoch": 665.6194690265487,
      "grad_norm": 1.1859777515610403e-07,
      "learning_rate": 3.2681145260196056e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21547576,
      "step": 37940
    },
    {
      "epoch": 665.70796460177,
      "grad_norm": 1.3058441084012884e-07,
      "learning_rate": 3.252311253778839e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21550392,
      "step": 37945
    },
    {
      "epoch": 665.7964601769911,
      "grad_norm": 1.1908990700248978e-06,
      "learning_rate": 3.2365460331325034e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21553416,
      "step": 37950
    },
    {
      "epoch": 665.8849557522124,
      "grad_norm": 4.076769982930273e-06,
      "learning_rate": 3.2208188665117934e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21556792,
      "step": 37955
    },
    {
      "epoch": 665.9734513274336,
      "grad_norm": 1.240992872908464e-07,
      "learning_rate": 3.205129756342018e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21559416,
      "step": 37960
    },
    {
      "epoch": 666.0530973451328,
      "grad_norm": 2.1486151524641173e-07,
      "learning_rate": 3.189478705042659e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21561768,
      "step": 37965
    },
    {
      "epoch": 666.141592920354,
      "grad_norm": 1.098661778087262e-05,
      "learning_rate": 3.173865715027341e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21564568,
      "step": 37970
    },
    {
      "epoch": 666.2300884955753,
      "grad_norm": 1.3298627266067342e-07,
      "learning_rate": 3.158290788703694e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21567640,
      "step": 37975
    },
    {
      "epoch": 666.3185840707964,
      "grad_norm": 8.543735674493291e-08,
      "learning_rate": 3.1427539284736297e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21570568,
      "step": 37980
    },
    {
      "epoch": 666.4070796460177,
      "grad_norm": 5.22850712059153e-07,
      "learning_rate": 3.127255136733093e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21573176,
      "step": 37985
    },
    {
      "epoch": 666.4955752212389,
      "grad_norm": 7.239696486749381e-08,
      "learning_rate": 3.1117944158722544e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21576184,
      "step": 37990
    },
    {
      "epoch": 666.5840707964602,
      "grad_norm": 1.4230869282982894e-07,
      "learning_rate": 3.0963717682752635e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21578840,
      "step": 37995
    },
    {
      "epoch": 666.6725663716815,
      "grad_norm": 8.68803482489966e-08,
      "learning_rate": 3.080987196320578e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21581512,
      "step": 38000
    },
    {
      "epoch": 666.6725663716815,
      "eval_loss": 0.8925443887710571,
      "eval_runtime": 1.0639,
      "eval_samples_per_second": 23.498,
      "eval_steps_per_second": 12.219,
      "num_input_tokens_seen": 21581512,
      "step": 38000
    },
    {
      "epoch": 666.7610619469026,
      "grad_norm": 1.0416067652840866e-07,
      "learning_rate": 3.065640702380607e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21584920,
      "step": 38005
    },
    {
      "epoch": 666.8495575221239,
      "grad_norm": 3.160547805691749e-07,
      "learning_rate": 3.050332288822011e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21587512,
      "step": 38010
    },
    {
      "epoch": 666.9380530973451,
      "grad_norm": 1.2711925592157058e-06,
      "learning_rate": 3.035061958005542e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21590312,
      "step": 38015
    },
    {
      "epoch": 667.0176991150443,
      "grad_norm": 1.6703123151273758e-07,
      "learning_rate": 3.019829712286093e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21592400,
      "step": 38020
    },
    {
      "epoch": 667.1061946902655,
      "grad_norm": 6.347207914814135e-08,
      "learning_rate": 3.004635554012647e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21595792,
      "step": 38025
    },
    {
      "epoch": 667.1946902654868,
      "grad_norm": 1.319427610724233e-06,
      "learning_rate": 2.9894794855283017e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21598464,
      "step": 38030
    },
    {
      "epoch": 667.2831858407079,
      "grad_norm": 8.216862568133365e-08,
      "learning_rate": 2.9743615091703816e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21601408,
      "step": 38035
    },
    {
      "epoch": 667.3716814159292,
      "grad_norm": 5.813297775603132e-06,
      "learning_rate": 2.959281627270216e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21604192,
      "step": 38040
    },
    {
      "epoch": 667.4601769911504,
      "grad_norm": 2.0659189203797723e-07,
      "learning_rate": 2.944239842153362e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21607360,
      "step": 38045
    },
    {
      "epoch": 667.5486725663717,
      "grad_norm": 1.7319097196377697e-06,
      "learning_rate": 2.929236156139381e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21610000,
      "step": 38050
    },
    {
      "epoch": 667.637168141593,
      "grad_norm": 1.5856284107940155e-07,
      "learning_rate": 2.9142705715420883e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21613296,
      "step": 38055
    },
    {
      "epoch": 667.7256637168142,
      "grad_norm": 7.305733618068189e-08,
      "learning_rate": 2.8993430906693595e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21615648,
      "step": 38060
    },
    {
      "epoch": 667.8141592920354,
      "grad_norm": 5.623919108188602e-08,
      "learning_rate": 2.88445371582316e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21618352,
      "step": 38065
    },
    {
      "epoch": 667.9026548672566,
      "grad_norm": 8.061122969138523e-08,
      "learning_rate": 2.8696024492996796e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21621520,
      "step": 38070
    },
    {
      "epoch": 667.9911504424779,
      "grad_norm": 1.3047011293565447e-07,
      "learning_rate": 2.854789293389115e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21624368,
      "step": 38075
    },
    {
      "epoch": 668.070796460177,
      "grad_norm": 2.3154456130214385e-07,
      "learning_rate": 2.8400142503758606e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21626816,
      "step": 38080
    },
    {
      "epoch": 668.1592920353983,
      "grad_norm": 6.557369403026314e-08,
      "learning_rate": 2.8252773225384276e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21629504,
      "step": 38085
    },
    {
      "epoch": 668.2477876106195,
      "grad_norm": 1.9102968451534252e-07,
      "learning_rate": 2.8105785121494143e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21632416,
      "step": 38090
    },
    {
      "epoch": 668.3362831858407,
      "grad_norm": 1.5679241016641754e-07,
      "learning_rate": 2.795917821475563e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21635264,
      "step": 38095
    },
    {
      "epoch": 668.4247787610619,
      "grad_norm": 7.82787537900731e-07,
      "learning_rate": 2.78129525277776e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21638192,
      "step": 38100
    },
    {
      "epoch": 668.5132743362832,
      "grad_norm": 5.606830200122204e-07,
      "learning_rate": 2.766710808310952e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21641504,
      "step": 38105
    },
    {
      "epoch": 668.6017699115044,
      "grad_norm": 2.3307377432502108e-06,
      "learning_rate": 2.7521644903242827e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21644672,
      "step": 38110
    },
    {
      "epoch": 668.6902654867257,
      "grad_norm": 1.3309776250025607e-06,
      "learning_rate": 2.7376563010609593e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21647200,
      "step": 38115
    },
    {
      "epoch": 668.7787610619469,
      "grad_norm": 1.3277980315251625e-07,
      "learning_rate": 2.72318624275833e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21649808,
      "step": 38120
    },
    {
      "epoch": 668.8672566371681,
      "grad_norm": 5.476487103805994e-07,
      "learning_rate": 2.7087543176478324e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21652784,
      "step": 38125
    },
    {
      "epoch": 668.9557522123894,
      "grad_norm": 7.399432178090137e-08,
      "learning_rate": 2.694360527955103e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21655456,
      "step": 38130
    },
    {
      "epoch": 669.0353982300885,
      "grad_norm": 1.5817097676062986e-07,
      "learning_rate": 2.680004875899811e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21657464,
      "step": 38135
    },
    {
      "epoch": 669.1238938053098,
      "grad_norm": 1.2344310107437195e-07,
      "learning_rate": 2.665687363695768e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21660184,
      "step": 38140
    },
    {
      "epoch": 669.212389380531,
      "grad_norm": 1.5057085533953796e-07,
      "learning_rate": 2.6514079935509584e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21662776,
      "step": 38145
    },
    {
      "epoch": 669.3008849557522,
      "grad_norm": 6.317085166074321e-08,
      "learning_rate": 2.6371667676673983e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21665384,
      "step": 38150
    },
    {
      "epoch": 669.3893805309734,
      "grad_norm": 1.6308248405039194e-06,
      "learning_rate": 2.6229636882412755e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21668760,
      "step": 38155
    },
    {
      "epoch": 669.4778761061947,
      "grad_norm": 9.66129860557885e-08,
      "learning_rate": 2.6087987574628935e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21671592,
      "step": 38160
    },
    {
      "epoch": 669.566371681416,
      "grad_norm": 2.0458621463603777e-07,
      "learning_rate": 2.5946719775166437e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21674664,
      "step": 38165
    },
    {
      "epoch": 669.6548672566372,
      "grad_norm": 1.2635324253551516e-07,
      "learning_rate": 2.5805833505810616e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21677672,
      "step": 38170
    },
    {
      "epoch": 669.7433628318585,
      "grad_norm": 4.3980154629252866e-08,
      "learning_rate": 2.566532878828798e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21680728,
      "step": 38175
    },
    {
      "epoch": 669.8318584070796,
      "grad_norm": 8.452695254845821e-08,
      "learning_rate": 2.552520564426619e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21683464,
      "step": 38180
    },
    {
      "epoch": 669.9203539823009,
      "grad_norm": 1.0420604894534335e-06,
      "learning_rate": 2.5385464095353803e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21686568,
      "step": 38185
    },
    {
      "epoch": 670.0,
      "grad_norm": 1.6351410181414394e-07,
      "learning_rate": 2.5246104163100804e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21688728,
      "step": 38190
    },
    {
      "epoch": 670.0884955752213,
      "grad_norm": 1.007833270705305e-05,
      "learning_rate": 2.510712586899833e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21691496,
      "step": 38195
    },
    {
      "epoch": 670.1769911504425,
      "grad_norm": 3.1634405672775756e-07,
      "learning_rate": 2.4968529234478124e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21694376,
      "step": 38200
    },
    {
      "epoch": 670.1769911504425,
      "eval_loss": 0.899570882320404,
      "eval_runtime": 1.0645,
      "eval_samples_per_second": 23.485,
      "eval_steps_per_second": 12.212,
      "num_input_tokens_seen": 21694376,
      "step": 38200
    },
    {
      "epoch": 670.2654867256637,
      "grad_norm": 1.0563072549985009e-07,
      "learning_rate": 2.483031428091448e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21697336,
      "step": 38205
    },
    {
      "epoch": 670.3539823008849,
      "grad_norm": 9.84484742616587e-08,
      "learning_rate": 2.469248102962091e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21700440,
      "step": 38210
    },
    {
      "epoch": 670.4424778761062,
      "grad_norm": 9.671783374187726e-08,
      "learning_rate": 2.4555029501853455e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21703448,
      "step": 38215
    },
    {
      "epoch": 670.5309734513274,
      "grad_norm": 2.0666408317993046e-07,
      "learning_rate": 2.441795971880906e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21706408,
      "step": 38220
    },
    {
      "epoch": 670.6194690265487,
      "grad_norm": 5.5910408747195106e-08,
      "learning_rate": 2.4281271701625255e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21709368,
      "step": 38225
    },
    {
      "epoch": 670.70796460177,
      "grad_norm": 1.753866882836519e-07,
      "learning_rate": 2.4144965471381007e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21712104,
      "step": 38230
    },
    {
      "epoch": 670.7964601769911,
      "grad_norm": 1.1843531666499985e-07,
      "learning_rate": 2.400904104909674e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21715304,
      "step": 38235
    },
    {
      "epoch": 670.8849557522124,
      "grad_norm": 1.1579671991057694e-05,
      "learning_rate": 2.3873498455733725e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21718184,
      "step": 38240
    },
    {
      "epoch": 670.9734513274336,
      "grad_norm": 1.338761649094522e-06,
      "learning_rate": 2.3738337712194137e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21720712,
      "step": 38245
    },
    {
      "epoch": 671.0530973451328,
      "grad_norm": 1.6859418394687964e-07,
      "learning_rate": 2.3603558839321305e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21723472,
      "step": 38250
    },
    {
      "epoch": 671.141592920354,
      "grad_norm": 7.653881084479508e-08,
      "learning_rate": 2.3469161857900267e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21726368,
      "step": 38255
    },
    {
      "epoch": 671.2300884955753,
      "grad_norm": 8.020335684477686e-08,
      "learning_rate": 2.3335146788656393e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21729168,
      "step": 38260
    },
    {
      "epoch": 671.3185840707964,
      "grad_norm": 4.731558931325708e-07,
      "learning_rate": 2.3201513652256757e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21731824,
      "step": 38265
    },
    {
      "epoch": 671.4070796460177,
      "grad_norm": 1.3784546126771602e-07,
      "learning_rate": 2.3068262469308766e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21734896,
      "step": 38270
    },
    {
      "epoch": 671.4955752212389,
      "grad_norm": 7.138264663808513e-07,
      "learning_rate": 2.2935393260362093e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21737392,
      "step": 38275
    },
    {
      "epoch": 671.5840707964602,
      "grad_norm": 8.359352676734488e-08,
      "learning_rate": 2.2802906045906458e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21740080,
      "step": 38280
    },
    {
      "epoch": 671.6725663716815,
      "grad_norm": 8.687775476801107e-08,
      "learning_rate": 2.2670800846373018e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21743056,
      "step": 38285
    },
    {
      "epoch": 671.7610619469026,
      "grad_norm": 2.176704356315895e-06,
      "learning_rate": 2.2539077682134367e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21745856,
      "step": 38290
    },
    {
      "epoch": 671.8495575221239,
      "grad_norm": 1.422492914571194e-07,
      "learning_rate": 2.2407736573503423e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21748784,
      "step": 38295
    },
    {
      "epoch": 671.9380530973451,
      "grad_norm": 4.950376251144917e-07,
      "learning_rate": 2.2276777540735093e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21751360,
      "step": 38300
    },
    {
      "epoch": 672.0176991150443,
      "grad_norm": 7.601749274499525e-08,
      "learning_rate": 2.2146200604024613e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21754216,
      "step": 38305
    },
    {
      "epoch": 672.1061946902655,
      "grad_norm": 1.5850180545839976e-07,
      "learning_rate": 2.2016005783508375e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21757096,
      "step": 38310
    },
    {
      "epoch": 672.1946902654868,
      "grad_norm": 2.648615122780029e-07,
      "learning_rate": 2.1886193099264763e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21759656,
      "step": 38315
    },
    {
      "epoch": 672.2831858407079,
      "grad_norm": 3.0623985480815463e-07,
      "learning_rate": 2.175676257131165e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21762728,
      "step": 38320
    },
    {
      "epoch": 672.3716814159292,
      "grad_norm": 1.1577217264857609e-06,
      "learning_rate": 2.162771421960974e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21765384,
      "step": 38325
    },
    {
      "epoch": 672.4601769911504,
      "grad_norm": 1.3159933587303385e-07,
      "learning_rate": 2.1499048064059224e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21768072,
      "step": 38330
    },
    {
      "epoch": 672.5486725663717,
      "grad_norm": 5.8066806474244004e-08,
      "learning_rate": 2.1370764124502285e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21770776,
      "step": 38335
    },
    {
      "epoch": 672.637168141593,
      "grad_norm": 1.3315009539383027e-07,
      "learning_rate": 2.1242862420721988e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21773688,
      "step": 38340
    },
    {
      "epoch": 672.7256637168142,
      "grad_norm": 1.2632432344616973e-07,
      "learning_rate": 2.1115342972442276e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21776520,
      "step": 38345
    },
    {
      "epoch": 672.8141592920354,
      "grad_norm": 8.552749932277948e-06,
      "learning_rate": 2.0988205799328252e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21779912,
      "step": 38350
    },
    {
      "epoch": 672.9026548672566,
      "grad_norm": 8.972701692755436e-08,
      "learning_rate": 2.0861450920986182e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21783064,
      "step": 38355
    },
    {
      "epoch": 672.9911504424779,
      "grad_norm": 1.412537216083365e-07,
      "learning_rate": 2.07350783569632e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21785752,
      "step": 38360
    },
    {
      "epoch": 673.070796460177,
      "grad_norm": 4.150101631239522e-06,
      "learning_rate": 2.060908812674761e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21788280,
      "step": 38365
    },
    {
      "epoch": 673.1592920353983,
      "grad_norm": 5.687054027703198e-08,
      "learning_rate": 2.0483480249768317e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21790888,
      "step": 38370
    },
    {
      "epoch": 673.2477876106195,
      "grad_norm": 7.697698833908362e-07,
      "learning_rate": 2.035825474539621e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21793880,
      "step": 38375
    },
    {
      "epoch": 673.3362831858407,
      "grad_norm": 8.31483760066476e-07,
      "learning_rate": 2.0233411632942235e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21796712,
      "step": 38380
    },
    {
      "epoch": 673.4247787610619,
      "grad_norm": 1.407428698030344e-07,
      "learning_rate": 2.0108950931658764e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21799816,
      "step": 38385
    },
    {
      "epoch": 673.5132743362832,
      "grad_norm": 1.7413049135939218e-07,
      "learning_rate": 1.998487266073934e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21802776,
      "step": 38390
    },
    {
      "epoch": 673.6017699115044,
      "grad_norm": 1.0242731462994925e-07,
      "learning_rate": 1.986117683931865e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21805768,
      "step": 38395
    },
    {
      "epoch": 673.6902654867257,
      "grad_norm": 1.2364124302166601e-07,
      "learning_rate": 1.9737863486471442e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21808568,
      "step": 38400
    },
    {
      "epoch": 673.6902654867257,
      "eval_loss": 0.8810822367668152,
      "eval_runtime": 1.0637,
      "eval_samples_per_second": 23.504,
      "eval_steps_per_second": 12.222,
      "num_input_tokens_seen": 21808568,
      "step": 38400
    },
    {
      "epoch": 673.7787610619469,
      "grad_norm": 5.967287108887831e-08,
      "learning_rate": 1.9614932621215e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21811496,
      "step": 38405
    },
    {
      "epoch": 673.8672566371681,
      "grad_norm": 1.0681631010811543e-06,
      "learning_rate": 1.9492384262506102e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21813784,
      "step": 38410
    },
    {
      "epoch": 673.9557522123894,
      "grad_norm": 1.2450196607005637e-07,
      "learning_rate": 1.9370218429243524e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21817224,
      "step": 38415
    },
    {
      "epoch": 674.0353982300885,
      "grad_norm": 9.727935434966639e-08,
      "learning_rate": 1.9248435140267197e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21819688,
      "step": 38420
    },
    {
      "epoch": 674.1238938053098,
      "grad_norm": 1.3884532563679386e-06,
      "learning_rate": 1.9127034414356814e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21822312,
      "step": 38425
    },
    {
      "epoch": 674.212389380531,
      "grad_norm": 1.1011876921429575e-07,
      "learning_rate": 1.9006016270234627e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21825144,
      "step": 38430
    },
    {
      "epoch": 674.3008849557522,
      "grad_norm": 1.2095580359527958e-07,
      "learning_rate": 1.888538072656293e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21828008,
      "step": 38435
    },
    {
      "epoch": 674.3893805309734,
      "grad_norm": 1.0326710935260053e-06,
      "learning_rate": 1.8765127801944893e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21831432,
      "step": 38440
    },
    {
      "epoch": 674.4778761061947,
      "grad_norm": 3.028915500635776e-07,
      "learning_rate": 1.8645257514925406e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21834472,
      "step": 38445
    },
    {
      "epoch": 674.566371681416,
      "grad_norm": 6.952063245080353e-07,
      "learning_rate": 1.8525769883989685e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21836840,
      "step": 38450
    },
    {
      "epoch": 674.6548672566372,
      "grad_norm": 2.856160676856234e-07,
      "learning_rate": 1.8406664927564654e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21839992,
      "step": 38455
    },
    {
      "epoch": 674.7433628318585,
      "grad_norm": 8.2924039190857e-08,
      "learning_rate": 1.8287942664017566e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21842968,
      "step": 38460
    },
    {
      "epoch": 674.8318584070796,
      "grad_norm": 2.1541691808124597e-07,
      "learning_rate": 1.8169603111656552e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21845400,
      "step": 38465
    },
    {
      "epoch": 674.9203539823009,
      "grad_norm": 3.1095203212316846e-07,
      "learning_rate": 1.805164628873146e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21848168,
      "step": 38470
    },
    {
      "epoch": 675.0,
      "grad_norm": 3.8118375300655316e-07,
      "learning_rate": 1.793407221343274e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21850760,
      "step": 38475
    },
    {
      "epoch": 675.0884955752213,
      "grad_norm": 2.8483694336500776e-07,
      "learning_rate": 1.781688090389172e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21853336,
      "step": 38480
    },
    {
      "epoch": 675.1769911504425,
      "grad_norm": 9.308639050686907e-07,
      "learning_rate": 1.770007237818061e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21856296,
      "step": 38485
    },
    {
      "epoch": 675.2654867256637,
      "grad_norm": 8.255042871496698e-08,
      "learning_rate": 1.7583646654313059e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21858936,
      "step": 38490
    },
    {
      "epoch": 675.3539823008849,
      "grad_norm": 3.928227272353979e-07,
      "learning_rate": 1.7467603750242757e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21862184,
      "step": 38495
    },
    {
      "epoch": 675.4424778761062,
      "grad_norm": 1.0517859294623122e-07,
      "learning_rate": 1.7351943683865944e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21864792,
      "step": 38500
    },
    {
      "epoch": 675.5309734513274,
      "grad_norm": 5.64287120141671e-07,
      "learning_rate": 1.723666647301808e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21867800,
      "step": 38505
    },
    {
      "epoch": 675.6194690265487,
      "grad_norm": 1.4831604744358629e-07,
      "learning_rate": 1.712177213547661e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21870968,
      "step": 38510
    },
    {
      "epoch": 675.70796460177,
      "grad_norm": 8.214981761511808e-08,
      "learning_rate": 1.7007260688959581e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21873736,
      "step": 38515
    },
    {
      "epoch": 675.7964601769911,
      "grad_norm": 9.811458312469767e-08,
      "learning_rate": 1.68931321511262e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21876808,
      "step": 38520
    },
    {
      "epoch": 675.8849557522124,
      "grad_norm": 1.0799627148116997e-07,
      "learning_rate": 1.6779386539576835e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21879944,
      "step": 38525
    },
    {
      "epoch": 675.9734513274336,
      "grad_norm": 1.059330472230613e-07,
      "learning_rate": 1.666602387185162e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21882920,
      "step": 38530
    },
    {
      "epoch": 676.0530973451328,
      "grad_norm": 1.3126553710662847e-07,
      "learning_rate": 1.655304416543352e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21885056,
      "step": 38535
    },
    {
      "epoch": 676.141592920354,
      "grad_norm": 7.584033312468819e-08,
      "learning_rate": 1.6440447437744698e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21887664,
      "step": 38540
    },
    {
      "epoch": 676.2300884955753,
      "grad_norm": 1.0693254353100201e-06,
      "learning_rate": 1.6328233706149332e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21890864,
      "step": 38545
    },
    {
      "epoch": 676.3185840707964,
      "grad_norm": 4.513748308454524e-07,
      "learning_rate": 1.6216402987951906e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21893232,
      "step": 38550
    },
    {
      "epoch": 676.4070796460177,
      "grad_norm": 2.800830429805501e-07,
      "learning_rate": 1.6104955300398627e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21896128,
      "step": 38555
    },
    {
      "epoch": 676.4955752212389,
      "grad_norm": 2.985109176734113e-06,
      "learning_rate": 1.5993890660675748e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21898768,
      "step": 38560
    },
    {
      "epoch": 676.5840707964602,
      "grad_norm": 1.0227712721189164e-07,
      "learning_rate": 1.5883209085910678e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21901824,
      "step": 38565
    },
    {
      "epoch": 676.6725663716815,
      "grad_norm": 5.919451240288254e-08,
      "learning_rate": 1.5772910593172264e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21904816,
      "step": 38570
    },
    {
      "epoch": 676.7610619469026,
      "grad_norm": 1.1512184272532977e-07,
      "learning_rate": 1.5662995199469954e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21908208,
      "step": 38575
    },
    {
      "epoch": 676.8495575221239,
      "grad_norm": 9.555382973758242e-08,
      "learning_rate": 1.5553462921753802e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21911136,
      "step": 38580
    },
    {
      "epoch": 676.9380530973451,
      "grad_norm": 1.864512739757629e-07,
      "learning_rate": 1.544431377691502e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21913952,
      "step": 38585
    },
    {
      "epoch": 677.0176991150443,
      "grad_norm": 6.695874787965295e-08,
      "learning_rate": 1.5335547781785975e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21916616,
      "step": 38590
    },
    {
      "epoch": 677.1061946902655,
      "grad_norm": 1.417812285353648e-07,
      "learning_rate": 1.5227164953139917e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21919464,
      "step": 38595
    },
    {
      "epoch": 677.1946902654868,
      "grad_norm": 4.378241555968998e-06,
      "learning_rate": 1.511916530769042e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21922424,
      "step": 38600
    },
    {
      "epoch": 677.1946902654868,
      "eval_loss": 0.9197937846183777,
      "eval_runtime": 1.0631,
      "eval_samples_per_second": 23.517,
      "eval_steps_per_second": 12.229,
      "num_input_tokens_seen": 21922424,
      "step": 38600
    },
    {
      "epoch": 677.2831858407079,
      "grad_norm": 1.4535605430410214e-07,
      "learning_rate": 1.5011548862092773e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21925352,
      "step": 38605
    },
    {
      "epoch": 677.3716814159292,
      "grad_norm": 6.403809749144784e-08,
      "learning_rate": 1.490431563294231e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21928232,
      "step": 38610
    },
    {
      "epoch": 677.4601769911504,
      "grad_norm": 9.61439496904859e-08,
      "learning_rate": 1.4797465636776365e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21930936,
      "step": 38615
    },
    {
      "epoch": 677.5486725663717,
      "grad_norm": 1.7063142649931251e-06,
      "learning_rate": 1.4690998890072027e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21933832,
      "step": 38620
    },
    {
      "epoch": 677.637168141593,
      "grad_norm": 9.13828159809782e-08,
      "learning_rate": 1.4584915409248112e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21936584,
      "step": 38625
    },
    {
      "epoch": 677.7256637168142,
      "grad_norm": 1.1227109553146875e-06,
      "learning_rate": 1.4479215210663754e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21938936,
      "step": 38630
    },
    {
      "epoch": 677.8141592920354,
      "grad_norm": 2.3951420757839514e-07,
      "learning_rate": 1.4373898310619528e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21941928,
      "step": 38635
    },
    {
      "epoch": 677.9026548672566,
      "grad_norm": 9.248936549965947e-08,
      "learning_rate": 1.4268964725356604e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21944952,
      "step": 38640
    },
    {
      "epoch": 677.9911504424779,
      "grad_norm": 7.881197916503879e-08,
      "learning_rate": 1.4164414471056764e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21947768,
      "step": 38645
    },
    {
      "epoch": 678.070796460177,
      "grad_norm": 1.8912372752311057e-06,
      "learning_rate": 1.4060247563843497e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21950296,
      "step": 38650
    },
    {
      "epoch": 678.1592920353983,
      "grad_norm": 8.539487339476182e-07,
      "learning_rate": 1.3956464019780068e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21953080,
      "step": 38655
    },
    {
      "epoch": 678.2477876106195,
      "grad_norm": 1.6729630658574024e-07,
      "learning_rate": 1.385306385487145e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21955736,
      "step": 38660
    },
    {
      "epoch": 678.3362831858407,
      "grad_norm": 1.6282827175473358e-07,
      "learning_rate": 1.3750047085063222e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21958952,
      "step": 38665
    },
    {
      "epoch": 678.4247787610619,
      "grad_norm": 1.1852255710209647e-07,
      "learning_rate": 1.3647413726242119e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21962056,
      "step": 38670
    },
    {
      "epoch": 678.5132743362832,
      "grad_norm": 9.38662196858786e-08,
      "learning_rate": 1.3545163794235205e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21964808,
      "step": 38675
    },
    {
      "epoch": 678.6017699115044,
      "grad_norm": 5.552512334361381e-07,
      "learning_rate": 1.3443297304810698e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21967928,
      "step": 38680
    },
    {
      "epoch": 678.6902654867257,
      "grad_norm": 1.566229030913746e-07,
      "learning_rate": 1.3341814273677977e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21970376,
      "step": 38685
    },
    {
      "epoch": 678.7787610619469,
      "grad_norm": 1.230764326010103e-07,
      "learning_rate": 1.324071471648647e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21973208,
      "step": 38690
    },
    {
      "epoch": 678.8672566371681,
      "grad_norm": 1.324239775613023e-07,
      "learning_rate": 1.3139998648827312e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21976712,
      "step": 38695
    },
    {
      "epoch": 678.9557522123894,
      "grad_norm": 2.1264455085656664e-07,
      "learning_rate": 1.3039666086232526e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21979784,
      "step": 38700
    },
    {
      "epoch": 679.0353982300885,
      "grad_norm": 3.348434631789132e-07,
      "learning_rate": 1.2939717044174183e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21982024,
      "step": 38705
    },
    {
      "epoch": 679.1238938053098,
      "grad_norm": 7.767685161752524e-08,
      "learning_rate": 1.284015153806578e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21984936,
      "step": 38710
    },
    {
      "epoch": 679.212389380531,
      "grad_norm": 2.685048059447581e-07,
      "learning_rate": 1.274096958326171e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21988008,
      "step": 38715
    },
    {
      "epoch": 679.3008849557522,
      "grad_norm": 5.354377776711772e-07,
      "learning_rate": 1.2642171195056952e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21990760,
      "step": 38720
    },
    {
      "epoch": 679.3893805309734,
      "grad_norm": 4.80674913205803e-08,
      "learning_rate": 1.2543756388687377e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21993464,
      "step": 38725
    },
    {
      "epoch": 679.4778761061947,
      "grad_norm": 1.0872255273852716e-07,
      "learning_rate": 1.2445725179330014e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21996376,
      "step": 38730
    },
    {
      "epoch": 679.566371681416,
      "grad_norm": 1.8468080043021473e-07,
      "learning_rate": 1.2348077582102212e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 21998856,
      "step": 38735
    },
    {
      "epoch": 679.6548672566372,
      "grad_norm": 8.836550335900029e-08,
      "learning_rate": 1.2250813612062762e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22001784,
      "step": 38740
    },
    {
      "epoch": 679.7433628318585,
      "grad_norm": 8.050383257796057e-07,
      "learning_rate": 1.215393328421105e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22004616,
      "step": 38745
    },
    {
      "epoch": 679.8318584070796,
      "grad_norm": 1.1062338245437786e-07,
      "learning_rate": 1.2057436613486796e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22007400,
      "step": 38750
    },
    {
      "epoch": 679.9203539823009,
      "grad_norm": 1.0639898562203598e-07,
      "learning_rate": 1.1961323614771424e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22010792,
      "step": 38755
    },
    {
      "epoch": 680.0,
      "grad_norm": 7.361920495441154e-08,
      "learning_rate": 1.1865594302886418e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22013496,
      "step": 38760
    },
    {
      "epoch": 680.0884955752213,
      "grad_norm": 1.2339091881585773e-07,
      "learning_rate": 1.1770248692594687e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22016200,
      "step": 38765
    },
    {
      "epoch": 680.1769911504425,
      "grad_norm": 6.198736173246289e-07,
      "learning_rate": 1.167528679859975e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22019496,
      "step": 38770
    },
    {
      "epoch": 680.2654867256637,
      "grad_norm": 2.354480102439993e-06,
      "learning_rate": 1.1580708635545446e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22022168,
      "step": 38775
    },
    {
      "epoch": 680.3539823008849,
      "grad_norm": 5.808639969018259e-08,
      "learning_rate": 1.1486514218017885e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22025416,
      "step": 38780
    },
    {
      "epoch": 680.4424778761062,
      "grad_norm": 1.5547684029115771e-07,
      "learning_rate": 1.1392703560542117e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22028696,
      "step": 38785
    },
    {
      "epoch": 680.5309734513274,
      "grad_norm": 1.2673993410317053e-07,
      "learning_rate": 1.129927667758518e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22031016,
      "step": 38790
    },
    {
      "epoch": 680.6194690265487,
      "grad_norm": 2.415633559849084e-07,
      "learning_rate": 1.1206233583554992e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22033752,
      "step": 38795
    },
    {
      "epoch": 680.70796460177,
      "grad_norm": 2.3957349526426697e-07,
      "learning_rate": 1.1113574292799523e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22036600,
      "step": 38800
    },
    {
      "epoch": 680.70796460177,
      "eval_loss": 0.9036830067634583,
      "eval_runtime": 1.063,
      "eval_samples_per_second": 23.518,
      "eval_steps_per_second": 12.23,
      "num_input_tokens_seen": 22036600,
      "step": 38800
    },
    {
      "epoch": 680.7964601769911,
      "grad_norm": 7.525461853674642e-08,
      "learning_rate": 1.1021298819608449e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22039480,
      "step": 38805
    },
    {
      "epoch": 680.8849557522124,
      "grad_norm": 2.0744587914123258e-07,
      "learning_rate": 1.0929407178211226e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22042696,
      "step": 38810
    },
    {
      "epoch": 680.9734513274336,
      "grad_norm": 1.0351059387403438e-07,
      "learning_rate": 1.0837899382779293e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22045272,
      "step": 38815
    },
    {
      "epoch": 681.0530973451328,
      "grad_norm": 1.1790528020583224e-07,
      "learning_rate": 1.0746775447423862e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22047824,
      "step": 38820
    },
    {
      "epoch": 681.141592920354,
      "grad_norm": 1.1664890564588859e-07,
      "learning_rate": 1.0656035386197583e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22051280,
      "step": 38825
    },
    {
      "epoch": 681.2300884955753,
      "grad_norm": 1.2811385374789097e-07,
      "learning_rate": 1.0565679213093982e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22054256,
      "step": 38830
    },
    {
      "epoch": 681.3185840707964,
      "grad_norm": 5.484103127173512e-08,
      "learning_rate": 1.0475706942046638e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22056848,
      "step": 38835
    },
    {
      "epoch": 681.4070796460177,
      "grad_norm": 1.9130071393647086e-07,
      "learning_rate": 1.0386118586930282e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22059824,
      "step": 38840
    },
    {
      "epoch": 681.4955752212389,
      "grad_norm": 1.503427995430684e-07,
      "learning_rate": 1.0296914161561367e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22062544,
      "step": 38845
    },
    {
      "epoch": 681.5840707964602,
      "grad_norm": 4.776652531290893e-07,
      "learning_rate": 1.0208093679695552e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22065504,
      "step": 38850
    },
    {
      "epoch": 681.6725663716815,
      "grad_norm": 1.6941711464824039e-06,
      "learning_rate": 1.0119657155030493e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22068304,
      "step": 38855
    },
    {
      "epoch": 681.7610619469026,
      "grad_norm": 9.603174078165466e-08,
      "learning_rate": 1.003160460120417e-07,
      "loss": 0.0,
      "num_input_tokens_seen": 22070832,
      "step": 38860
    },
    {
      "epoch": 681.8495575221239,
      "grad_norm": 1.4143701321245317e-07,
      "learning_rate": 9.943936031795165e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22073968,
      "step": 38865
    },
    {
      "epoch": 681.9380530973451,
      "grad_norm": 1.080550049437079e-07,
      "learning_rate": 9.856651460323219e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22076576,
      "step": 38870
    },
    {
      "epoch": 682.0176991150443,
      "grad_norm": 1.0230675684397283e-07,
      "learning_rate": 9.769750900248953e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22078808,
      "step": 38875
    },
    {
      "epoch": 682.1061946902655,
      "grad_norm": 1.3648265451138286e-07,
      "learning_rate": 9.683234364973038e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22082264,
      "step": 38880
    },
    {
      "epoch": 682.1946902654868,
      "grad_norm": 7.573891736001315e-08,
      "learning_rate": 9.597101867837854e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22085080,
      "step": 38885
    },
    {
      "epoch": 682.2831858407079,
      "grad_norm": 9.908131914926344e-08,
      "learning_rate": 9.511353422125835e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22088376,
      "step": 38890
    },
    {
      "epoch": 682.3716814159292,
      "grad_norm": 1.1362092209310504e-06,
      "learning_rate": 9.42598904106029e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22091160,
      "step": 38895
    },
    {
      "epoch": 682.4601769911504,
      "grad_norm": 4.436952622199897e-06,
      "learning_rate": 9.341008737806245e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22094216,
      "step": 38900
    },
    {
      "epoch": 682.5486725663717,
      "grad_norm": 1.9517460714268964e-07,
      "learning_rate": 9.256412525467661e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22097736,
      "step": 38905
    },
    {
      "epoch": 682.637168141593,
      "grad_norm": 1.2768795443207637e-07,
      "learning_rate": 9.172200417091326e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22100328,
      "step": 38910
    },
    {
      "epoch": 682.7256637168142,
      "grad_norm": 1.9133274520299892e-07,
      "learning_rate": 9.088372425663239e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22103096,
      "step": 38915
    },
    {
      "epoch": 682.8141592920354,
      "grad_norm": 8.915657900843144e-08,
      "learning_rate": 9.004928564110837e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22105624,
      "step": 38920
    },
    {
      "epoch": 682.9026548672566,
      "grad_norm": 1.6540391811759037e-07,
      "learning_rate": 8.92186884530244e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22108104,
      "step": 38925
    },
    {
      "epoch": 682.9911504424779,
      "grad_norm": 2.421506906102877e-07,
      "learning_rate": 8.83919328204641e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22110808,
      "step": 38930
    },
    {
      "epoch": 683.070796460177,
      "grad_norm": 1.0877198519665399e-07,
      "learning_rate": 8.756901887093105e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22112896,
      "step": 38935
    },
    {
      "epoch": 683.1592920353983,
      "grad_norm": 5.801953761874756e-07,
      "learning_rate": 8.674994673132098e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22116272,
      "step": 38940
    },
    {
      "epoch": 683.2477876106195,
      "grad_norm": 1.1799630073028311e-07,
      "learning_rate": 8.593471652794949e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22119456,
      "step": 38945
    },
    {
      "epoch": 683.3362831858407,
      "grad_norm": 1.7823427356233879e-07,
      "learning_rate": 8.512332838653548e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22122544,
      "step": 38950
    },
    {
      "epoch": 683.4247787610619,
      "grad_norm": 1.0033652841912044e-07,
      "learning_rate": 8.431578243220106e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22125728,
      "step": 38955
    },
    {
      "epoch": 683.5132743362832,
      "grad_norm": 2.0839888748014346e-07,
      "learning_rate": 8.351207878948552e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22128864,
      "step": 38960
    },
    {
      "epoch": 683.6017699115044,
      "grad_norm": 1.0830714103349237e-07,
      "learning_rate": 8.271221758232583e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22131408,
      "step": 38965
    },
    {
      "epoch": 683.6902654867257,
      "grad_norm": 9.799941125265832e-08,
      "learning_rate": 8.191619893407332e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22133920,
      "step": 38970
    },
    {
      "epoch": 683.7787610619469,
      "grad_norm": 1.0675667283521761e-07,
      "learning_rate": 8.112402296748534e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22136384,
      "step": 38975
    },
    {
      "epoch": 683.8672566371681,
      "grad_norm": 1.0000847083802e-07,
      "learning_rate": 8.033568980471973e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22139024,
      "step": 38980
    },
    {
      "epoch": 683.9557522123894,
      "grad_norm": 2.137901162768685e-07,
      "learning_rate": 7.955119956735146e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22142064,
      "step": 38985
    },
    {
      "epoch": 684.0353982300885,
      "grad_norm": 2.3803545445844065e-07,
      "learning_rate": 7.877055237636155e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22144496,
      "step": 38990
    },
    {
      "epoch": 684.1238938053098,
      "grad_norm": 1.4419946126054128e-07,
      "learning_rate": 7.79937483521287e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22147328,
      "step": 38995
    },
    {
      "epoch": 684.212389380531,
      "grad_norm": 6.41456097127957e-08,
      "learning_rate": 7.722078761444873e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22150992,
      "step": 39000
    },
    {
      "epoch": 684.212389380531,
      "eval_loss": 0.8996666669845581,
      "eval_runtime": 1.0661,
      "eval_samples_per_second": 23.45,
      "eval_steps_per_second": 12.194,
      "num_input_tokens_seen": 22150992,
      "step": 39000
    },
    {
      "epoch": 684.3008849557522,
      "grad_norm": 1.1317268899802002e-07,
      "learning_rate": 7.645167028252631e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22153408,
      "step": 39005
    },
    {
      "epoch": 684.3893805309734,
      "grad_norm": 1.0960989129671361e-06,
      "learning_rate": 7.568639647496379e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22156784,
      "step": 39010
    },
    {
      "epoch": 684.4778761061947,
      "grad_norm": 2.973548873796972e-07,
      "learning_rate": 7.492496630977508e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22159968,
      "step": 39015
    },
    {
      "epoch": 684.566371681416,
      "grad_norm": 1.6029869698286348e-07,
      "learning_rate": 7.416737990438571e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22162464,
      "step": 39020
    },
    {
      "epoch": 684.6548672566372,
      "grad_norm": 1.1331253517710138e-06,
      "learning_rate": 7.341363737562445e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22164880,
      "step": 39025
    },
    {
      "epoch": 684.7433628318585,
      "grad_norm": 1.2866507859143894e-07,
      "learning_rate": 7.266373883972887e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22167280,
      "step": 39030
    },
    {
      "epoch": 684.8318584070796,
      "grad_norm": 1.0603748279436331e-07,
      "learning_rate": 7.191768441233981e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22170144,
      "step": 39035
    },
    {
      "epoch": 684.9203539823009,
      "grad_norm": 2.1761437096756708e-07,
      "learning_rate": 7.11754742085069e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22172912,
      "step": 39040
    },
    {
      "epoch": 685.0,
      "grad_norm": 6.011522054905072e-05,
      "learning_rate": 7.043710834269413e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22175264,
      "step": 39045
    },
    {
      "epoch": 685.0884955752213,
      "grad_norm": 1.2606260213487985e-07,
      "learning_rate": 6.970258692876319e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22178128,
      "step": 39050
    },
    {
      "epoch": 685.1769911504425,
      "grad_norm": 1.8532134049564775e-07,
      "learning_rate": 6.897191007998738e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22180960,
      "step": 39055
    },
    {
      "epoch": 685.2654867256637,
      "grad_norm": 1.361675572297827e-07,
      "learning_rate": 6.824507790904599e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22183760,
      "step": 39060
    },
    {
      "epoch": 685.3539823008849,
      "grad_norm": 4.859885507357831e-07,
      "learning_rate": 6.752209052802439e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22186400,
      "step": 39065
    },
    {
      "epoch": 685.4424778761062,
      "grad_norm": 5.769787136955529e-08,
      "learning_rate": 6.680294804841946e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22189040,
      "step": 39070
    },
    {
      "epoch": 685.5309734513274,
      "grad_norm": 9.17491718155361e-08,
      "learning_rate": 6.608765058112865e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22191728,
      "step": 39075
    },
    {
      "epoch": 685.6194690265487,
      "grad_norm": 3.904245033936604e-07,
      "learning_rate": 6.537619823646368e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22194880,
      "step": 39080
    },
    {
      "epoch": 685.70796460177,
      "grad_norm": 8.593928413347385e-08,
      "learning_rate": 6.466859112413404e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22198032,
      "step": 39085
    },
    {
      "epoch": 685.7964601769911,
      "grad_norm": 2.348467944557342e-07,
      "learning_rate": 6.39648293532663e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22201184,
      "step": 39090
    },
    {
      "epoch": 685.8849557522124,
      "grad_norm": 7.378277899761088e-08,
      "learning_rate": 6.32649130323848e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22204144,
      "step": 39095
    },
    {
      "epoch": 685.9734513274336,
      "grad_norm": 1.2768343538027693e-07,
      "learning_rate": 6.256884226943094e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22206912,
      "step": 39100
    },
    {
      "epoch": 686.0530973451328,
      "grad_norm": 1.0888977186596094e-07,
      "learning_rate": 6.187661717174386e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22209152,
      "step": 39105
    },
    {
      "epoch": 686.141592920354,
      "grad_norm": 8.06042407930363e-07,
      "learning_rate": 6.118823784607708e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22211904,
      "step": 39110
    },
    {
      "epoch": 686.2300884955753,
      "grad_norm": 1.231816924018858e-07,
      "learning_rate": 6.050370439858178e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22214864,
      "step": 39115
    },
    {
      "epoch": 686.3185840707964,
      "grad_norm": 1.5082324011927994e-07,
      "learning_rate": 5.98230169348235e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22217424,
      "step": 39120
    },
    {
      "epoch": 686.4070796460177,
      "grad_norm": 5.568428491642408e-07,
      "learning_rate": 5.914617555977664e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22219936,
      "step": 39125
    },
    {
      "epoch": 686.4955752212389,
      "grad_norm": 7.504789323320438e-08,
      "learning_rate": 5.8473180377816017e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22222432,
      "step": 39130
    },
    {
      "epoch": 686.5840707964602,
      "grad_norm": 2.850779594609776e-07,
      "learning_rate": 5.780403149272251e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22225456,
      "step": 39135
    },
    {
      "epoch": 686.6725663716815,
      "grad_norm": 3.1329324201578856e-07,
      "learning_rate": 5.7138729007694126e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22228304,
      "step": 39140
    },
    {
      "epoch": 686.7610619469026,
      "grad_norm": 1.0581327813952157e-07,
      "learning_rate": 5.64772730253238e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22231824,
      "step": 39145
    },
    {
      "epoch": 686.8495575221239,
      "grad_norm": 9.322940286438097e-07,
      "learning_rate": 5.5819663647618814e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22234688,
      "step": 39150
    },
    {
      "epoch": 686.9380530973451,
      "grad_norm": 2.411570108051819e-07,
      "learning_rate": 5.5165900975989723e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22237632,
      "step": 39155
    },
    {
      "epoch": 687.0176991150443,
      "grad_norm": 2.3555165284960822e-07,
      "learning_rate": 5.451598511125311e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22240208,
      "step": 39160
    },
    {
      "epoch": 687.1061946902655,
      "grad_norm": 5.319129741110373e-07,
      "learning_rate": 5.3869916153637124e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22243472,
      "step": 39165
    },
    {
      "epoch": 687.1946902654868,
      "grad_norm": 8.44736121052847e-08,
      "learning_rate": 5.322769420277318e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22246384,
      "step": 39170
    },
    {
      "epoch": 687.2831858407079,
      "grad_norm": 1.593152205714432e-07,
      "learning_rate": 5.258931935769873e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22249520,
      "step": 39175
    },
    {
      "epoch": 687.3716814159292,
      "grad_norm": 7.375135169240821e-08,
      "learning_rate": 5.19547917168628e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22252336,
      "step": 39180
    },
    {
      "epoch": 687.4601769911504,
      "grad_norm": 4.3106794578307017e-07,
      "learning_rate": 5.13241113781121e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22254912,
      "step": 39185
    },
    {
      "epoch": 687.5486725663717,
      "grad_norm": 1.1975182587775635e-06,
      "learning_rate": 5.0697278438707755e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22258080,
      "step": 39190
    },
    {
      "epoch": 687.637168141593,
      "grad_norm": 2.0056872074292187e-07,
      "learning_rate": 5.0074292995316854e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22261072,
      "step": 39195
    },
    {
      "epoch": 687.7256637168142,
      "grad_norm": 1.246693273060373e-07,
      "learning_rate": 4.945515514400978e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22263616,
      "step": 39200
    },
    {
      "epoch": 687.7256637168142,
      "eval_loss": 0.9019477367401123,
      "eval_runtime": 1.0665,
      "eval_samples_per_second": 23.441,
      "eval_steps_per_second": 12.189,
      "num_input_tokens_seen": 22263616,
      "step": 39200
    },
    {
      "epoch": 687.8141592920354,
      "grad_norm": 1.4932295755443192e-07,
      "learning_rate": 4.883986498026571e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22266288,
      "step": 39205
    },
    {
      "epoch": 687.9026548672566,
      "grad_norm": 9.846420567782843e-08,
      "learning_rate": 4.822842259896987e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22269216,
      "step": 39210
    },
    {
      "epoch": 687.9911504424779,
      "grad_norm": 1.2500579771312914e-07,
      "learning_rate": 4.762082809441626e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22271984,
      "step": 39215
    },
    {
      "epoch": 688.070796460177,
      "grad_norm": 1.1091438523180841e-07,
      "learning_rate": 4.7017081560302156e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22274032,
      "step": 39220
    },
    {
      "epoch": 688.1592920353983,
      "grad_norm": 1.1632086227564287e-07,
      "learning_rate": 4.6417183089730866e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22276752,
      "step": 39225
    },
    {
      "epoch": 688.2477876106195,
      "grad_norm": 2.1197973865128006e-07,
      "learning_rate": 4.5821132775217265e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22279616,
      "step": 39230
    },
    {
      "epoch": 688.3362831858407,
      "grad_norm": 1.1796169729905159e-07,
      "learning_rate": 4.5228930708679504e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22282272,
      "step": 39235
    },
    {
      "epoch": 688.4247787610619,
      "grad_norm": 6.419491569431557e-07,
      "learning_rate": 4.464057698144175e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22285376,
      "step": 39240
    },
    {
      "epoch": 688.5132743362832,
      "grad_norm": 9.406633694197808e-08,
      "learning_rate": 4.4056071684236974e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22288400,
      "step": 39245
    },
    {
      "epoch": 688.6017699115044,
      "grad_norm": 8.879785440285559e-08,
      "learning_rate": 4.347541490719864e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22291408,
      "step": 39250
    },
    {
      "epoch": 688.6902654867257,
      "grad_norm": 1.795322646103159e-07,
      "learning_rate": 4.2898606739877336e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22294608,
      "step": 39255
    },
    {
      "epoch": 688.7787610619469,
      "grad_norm": 1.6916823142310022e-07,
      "learning_rate": 4.232564727122135e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22297488,
      "step": 39260
    },
    {
      "epoch": 688.8672566371681,
      "grad_norm": 9.269788847632299e-08,
      "learning_rate": 4.1756536589585004e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22300544,
      "step": 39265
    },
    {
      "epoch": 688.9557522123894,
      "grad_norm": 8.61892459624869e-08,
      "learning_rate": 4.119127478273976e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22303408,
      "step": 39270
    },
    {
      "epoch": 689.0353982300885,
      "grad_norm": 1.1142633837835092e-07,
      "learning_rate": 4.062986193784923e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22305656,
      "step": 39275
    },
    {
      "epoch": 689.1238938053098,
      "grad_norm": 4.5807564674760215e-06,
      "learning_rate": 4.007229814149416e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22308152,
      "step": 39280
    },
    {
      "epoch": 689.212389380531,
      "grad_norm": 1.397610276399064e-07,
      "learning_rate": 3.951858347965576e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22311112,
      "step": 39285
    },
    {
      "epoch": 689.3008849557522,
      "grad_norm": 3.267118700023275e-07,
      "learning_rate": 3.896871803772684e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22313784,
      "step": 39290
    },
    {
      "epoch": 689.3893805309734,
      "grad_norm": 5.142333634466922e-07,
      "learning_rate": 3.842270190050068e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22316552,
      "step": 39295
    },
    {
      "epoch": 689.4778761061947,
      "grad_norm": 7.349444786086679e-08,
      "learning_rate": 3.7880535152179376e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22319544,
      "step": 39300
    },
    {
      "epoch": 689.566371681416,
      "grad_norm": 8.856153499436914e-08,
      "learning_rate": 3.734221787637382e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22322152,
      "step": 39305
    },
    {
      "epoch": 689.6548672566372,
      "grad_norm": 7.311433023460268e-07,
      "learning_rate": 3.680775015609817e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22325224,
      "step": 39310
    },
    {
      "epoch": 689.7433628318585,
      "grad_norm": 1.3085375485388795e-06,
      "learning_rate": 3.627713207377537e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22328312,
      "step": 39315
    },
    {
      "epoch": 689.8318584070796,
      "grad_norm": 1.715200710350473e-07,
      "learning_rate": 3.575036371123164e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22330952,
      "step": 39320
    },
    {
      "epoch": 689.9203539823009,
      "grad_norm": 1.2726147247121844e-07,
      "learning_rate": 3.5227445149704776e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22334248,
      "step": 39325
    },
    {
      "epoch": 690.0,
      "grad_norm": 1.0522663984602332e-07,
      "learning_rate": 3.470837646983027e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22336792,
      "step": 39330
    },
    {
      "epoch": 690.0884955752213,
      "grad_norm": 3.2357323220821854e-07,
      "learning_rate": 3.419315775165799e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22339912,
      "step": 39335
    },
    {
      "epoch": 690.1769911504425,
      "grad_norm": 1.0191126875724876e-06,
      "learning_rate": 3.368178907464103e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22342344,
      "step": 39340
    },
    {
      "epoch": 690.2654867256637,
      "grad_norm": 8.385780461139802e-07,
      "learning_rate": 3.317427051763855e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22345512,
      "step": 39345
    },
    {
      "epoch": 690.3539823008849,
      "grad_norm": 8.640783732971613e-08,
      "learning_rate": 3.267060215891571e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22348264,
      "step": 39350
    },
    {
      "epoch": 690.4424778761062,
      "grad_norm": 7.068236840268582e-08,
      "learning_rate": 3.217078407614649e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22351208,
      "step": 39355
    },
    {
      "epoch": 690.5309734513274,
      "grad_norm": 9.589081884087136e-08,
      "learning_rate": 3.1674816346405345e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22354824,
      "step": 39360
    },
    {
      "epoch": 690.6194690265487,
      "grad_norm": 8.94615226343376e-08,
      "learning_rate": 3.11826990461811e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22357864,
      "step": 39365
    },
    {
      "epoch": 690.70796460177,
      "grad_norm": 1.3409723464974377e-07,
      "learning_rate": 3.069443225136304e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22360520,
      "step": 39370
    },
    {
      "epoch": 690.7964601769911,
      "grad_norm": 2.9334174200812413e-07,
      "learning_rate": 3.021001603724372e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22363416,
      "step": 39375
    },
    {
      "epoch": 690.8849557522124,
      "grad_norm": 2.877771976272925e-07,
      "learning_rate": 2.9729450478532818e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22366376,
      "step": 39380
    },
    {
      "epoch": 690.9734513274336,
      "grad_norm": 9.356631380796898e-07,
      "learning_rate": 2.9252735649337726e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22368984,
      "step": 39385
    },
    {
      "epoch": 691.0530973451328,
      "grad_norm": 1.0388191640231526e-07,
      "learning_rate": 2.8779871623171863e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22371184,
      "step": 39390
    },
    {
      "epoch": 691.141592920354,
      "grad_norm": 3.941781869798433e-07,
      "learning_rate": 2.8310858472957448e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22374272,
      "step": 39395
    },
    {
      "epoch": 691.2300884955753,
      "grad_norm": 1.8962880687922734e-07,
      "learning_rate": 2.784569627101996e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22377936,
      "step": 39400
    },
    {
      "epoch": 691.2300884955753,
      "eval_loss": 0.8945109844207764,
      "eval_runtime": 1.0626,
      "eval_samples_per_second": 23.527,
      "eval_steps_per_second": 12.234,
      "num_input_tokens_seen": 22377936,
      "step": 39400
    },
    {
      "epoch": 691.3185840707964,
      "grad_norm": 1.272444194455602e-07,
      "learning_rate": 2.738438508909924e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22380656,
      "step": 39405
    },
    {
      "epoch": 691.4070796460177,
      "grad_norm": 6.476789593534704e-08,
      "learning_rate": 2.692692499833005e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22383424,
      "step": 39410
    },
    {
      "epoch": 691.4955752212389,
      "grad_norm": 1.2120921155656106e-07,
      "learning_rate": 2.647331606926151e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22385840,
      "step": 39415
    },
    {
      "epoch": 691.5840707964602,
      "grad_norm": 1.2464441567772155e-07,
      "learning_rate": 2.6023558371843225e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22388720,
      "step": 39420
    },
    {
      "epoch": 691.6725663716815,
      "grad_norm": 1.0748126300086369e-07,
      "learning_rate": 2.557765197543638e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22391520,
      "step": 39425
    },
    {
      "epoch": 691.7610619469026,
      "grad_norm": 9.412915602524663e-08,
      "learning_rate": 2.513559694880263e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22394432,
      "step": 39430
    },
    {
      "epoch": 691.8495575221239,
      "grad_norm": 9.454951168663683e-07,
      "learning_rate": 2.469739336011523e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22397264,
      "step": 39435
    },
    {
      "epoch": 691.9380530973451,
      "grad_norm": 8.771067427915114e-08,
      "learning_rate": 2.4263041276947894e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22400176,
      "step": 39440
    },
    {
      "epoch": 692.0176991150443,
      "grad_norm": 8.567007370174906e-08,
      "learning_rate": 2.3832540766283164e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22402368,
      "step": 39445
    },
    {
      "epoch": 692.1061946902655,
      "grad_norm": 5.049377023169654e-07,
      "learning_rate": 2.3405891894512366e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22404960,
      "step": 39450
    },
    {
      "epoch": 692.1946902654868,
      "grad_norm": 1.9685238328293053e-07,
      "learning_rate": 2.29830947274301e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22407568,
      "step": 39455
    },
    {
      "epoch": 692.2831858407079,
      "grad_norm": 2.9676493795705028e-06,
      "learning_rate": 2.2564149330231432e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22410464,
      "step": 39460
    },
    {
      "epoch": 692.3716814159292,
      "grad_norm": 1.687055828369921e-07,
      "learning_rate": 2.2149055767528572e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22413088,
      "step": 39465
    },
    {
      "epoch": 692.4601769911504,
      "grad_norm": 1.2252878889285057e-07,
      "learning_rate": 2.1737814103334197e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22416144,
      "step": 39470
    },
    {
      "epoch": 692.5486725663717,
      "grad_norm": 1.3834261380907265e-07,
      "learning_rate": 2.1330424401064253e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22418752,
      "step": 39475
    },
    {
      "epoch": 692.637168141593,
      "grad_norm": 1.8944624002870114e-07,
      "learning_rate": 2.092688672354348e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22422064,
      "step": 39480
    },
    {
      "epoch": 692.7256637168142,
      "grad_norm": 2.1567110763953679e-07,
      "learning_rate": 2.0527201133005435e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22425712,
      "step": 39485
    },
    {
      "epoch": 692.8141592920354,
      "grad_norm": 1.123077481679502e-06,
      "learning_rate": 2.0131367691084148e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22428624,
      "step": 39490
    },
    {
      "epoch": 692.9026548672566,
      "grad_norm": 1.020048259192663e-07,
      "learning_rate": 1.9739386458819675e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22431344,
      "step": 39495
    },
    {
      "epoch": 692.9911504424779,
      "grad_norm": 1.0422967022805096e-07,
      "learning_rate": 1.9351257496666442e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22434000,
      "step": 39500
    },
    {
      "epoch": 693.070796460177,
      "grad_norm": 4.5823267669220513e-07,
      "learning_rate": 1.896698086447657e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22436352,
      "step": 39505
    },
    {
      "epoch": 693.1592920353983,
      "grad_norm": 1.330959094048012e-07,
      "learning_rate": 1.8586556621505436e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22439056,
      "step": 39510
    },
    {
      "epoch": 693.2477876106195,
      "grad_norm": 7.190090514086478e-07,
      "learning_rate": 1.820998482642833e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22442384,
      "step": 39515
    },
    {
      "epoch": 693.3362831858407,
      "grad_norm": 1.4173450608723215e-06,
      "learning_rate": 1.7837265537309912e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22445760,
      "step": 39520
    },
    {
      "epoch": 693.4247787610619,
      "grad_norm": 9.417903612529699e-08,
      "learning_rate": 1.7468398811629206e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22448672,
      "step": 39525
    },
    {
      "epoch": 693.5132743362832,
      "grad_norm": 2.852528666608123e-07,
      "learning_rate": 1.710338470627404e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22451120,
      "step": 39530
    },
    {
      "epoch": 693.6017699115044,
      "grad_norm": 1.3080401117804286e-07,
      "learning_rate": 1.6742223277529945e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22453728,
      "step": 39535
    },
    {
      "epoch": 693.6902654867257,
      "grad_norm": 1.6530263735603512e-07,
      "learning_rate": 1.6384914581094036e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22456832,
      "step": 39540
    },
    {
      "epoch": 693.7787610619469,
      "grad_norm": 3.739905025668122e-07,
      "learning_rate": 1.6031458672069455e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22459456,
      "step": 39545
    },
    {
      "epoch": 693.8672566371681,
      "grad_norm": 1.4868480491259106e-07,
      "learning_rate": 1.5681855604962602e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22462416,
      "step": 39550
    },
    {
      "epoch": 693.9557522123894,
      "grad_norm": 1.7736786617206235e-07,
      "learning_rate": 1.5336105433683135e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22465168,
      "step": 39555
    },
    {
      "epoch": 694.0353982300885,
      "grad_norm": 4.873733701060701e-07,
      "learning_rate": 1.499420821155506e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22467592,
      "step": 39560
    },
    {
      "epoch": 694.1238938053098,
      "grad_norm": 7.616653618924829e-08,
      "learning_rate": 1.4656163991302874e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22470280,
      "step": 39565
    },
    {
      "epoch": 694.212389380531,
      "grad_norm": 9.71288116602409e-08,
      "learning_rate": 1.4321972825051544e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22473576,
      "step": 39570
    },
    {
      "epoch": 694.3008849557522,
      "grad_norm": 8.60090736409802e-08,
      "learning_rate": 1.3991634764345951e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22475960,
      "step": 39575
    },
    {
      "epoch": 694.3893805309734,
      "grad_norm": 6.684510367449548e-08,
      "learning_rate": 1.3665149860120352e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22478872,
      "step": 39580
    },
    {
      "epoch": 694.4778761061947,
      "grad_norm": 1.0045884835108154e-07,
      "learning_rate": 1.3342518162728912e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22481368,
      "step": 39585
    },
    {
      "epoch": 694.566371681416,
      "grad_norm": 1.1020408408057847e-07,
      "learning_rate": 1.30237397219235e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22484472,
      "step": 39590
    },
    {
      "epoch": 694.6548672566372,
      "grad_norm": 8.544405716293113e-08,
      "learning_rate": 1.2708814586862016e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22487672,
      "step": 39595
    },
    {
      "epoch": 694.7433628318585,
      "grad_norm": 2.8059002943336964e-07,
      "learning_rate": 1.2397742806111168e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22490328,
      "step": 39600
    },
    {
      "epoch": 694.7433628318585,
      "eval_loss": 0.9180490970611572,
      "eval_runtime": 1.0698,
      "eval_samples_per_second": 23.368,
      "eval_steps_per_second": 12.151,
      "num_input_tokens_seen": 22490328,
      "step": 39600
    },
    {
      "epoch": 694.8318584070796,
      "grad_norm": 4.290368451620452e-07,
      "learning_rate": 1.209052442764369e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22493160,
      "step": 39605
    },
    {
      "epoch": 694.9203539823009,
      "grad_norm": 1.214715439346037e-07,
      "learning_rate": 1.17871594988328e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22496040,
      "step": 39610
    },
    {
      "epoch": 695.0,
      "grad_norm": 8.951104746302008e-08,
      "learning_rate": 1.1487648066466072e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22498816,
      "step": 39615
    },
    {
      "epoch": 695.0884955752213,
      "grad_norm": 2.221804322743992e-07,
      "learning_rate": 1.1191990176728784e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22501760,
      "step": 39620
    },
    {
      "epoch": 695.1769911504425,
      "grad_norm": 9.27560037666808e-08,
      "learning_rate": 1.0900185875215018e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22504800,
      "step": 39625
    },
    {
      "epoch": 695.2654867256637,
      "grad_norm": 1.2403732796428812e-07,
      "learning_rate": 1.0612235206924891e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22507888,
      "step": 39630
    },
    {
      "epoch": 695.3539823008849,
      "grad_norm": 1.0765684237412643e-05,
      "learning_rate": 1.0328138216264549e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22510160,
      "step": 39635
    },
    {
      "epoch": 695.4424778761062,
      "grad_norm": 1.1642392649946487e-07,
      "learning_rate": 1.004789494704339e-08,
      "loss": 0.0,
      "num_input_tokens_seen": 22512816,
      "step": 39640
    },
    {
      "epoch": 695.5309734513274,
      "grad_norm": 8.060453069447249e-07,
      "learning_rate": 9.771505442482397e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22515760,
      "step": 39645
    },
    {
      "epoch": 695.6194690265487,
      "grad_norm": 9.58391623839816e-08,
      "learning_rate": 9.498969745200259e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22518624,
      "step": 39650
    },
    {
      "epoch": 695.70796460177,
      "grad_norm": 1.0406987627220587e-07,
      "learning_rate": 9.230287897230017e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22521344,
      "step": 39655
    },
    {
      "epoch": 695.7964601769911,
      "grad_norm": 1.0073325285020474e-07,
      "learning_rate": 8.965459940002419e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22524096,
      "step": 39660
    },
    {
      "epoch": 695.8849557522124,
      "grad_norm": 1.2680008865117998e-07,
      "learning_rate": 8.704485914357019e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22527280,
      "step": 39665
    },
    {
      "epoch": 695.9734513274336,
      "grad_norm": 1.5456091659871163e-06,
      "learning_rate": 8.447365860539402e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22530672,
      "step": 39670
    },
    {
      "epoch": 696.0530973451328,
      "grad_norm": 1.2543935667963524e-07,
      "learning_rate": 8.194099818201184e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22532944,
      "step": 39675
    },
    {
      "epoch": 696.141592920354,
      "grad_norm": 1.1484625161983786e-07,
      "learning_rate": 7.944687826400011e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22536400,
      "step": 39680
    },
    {
      "epoch": 696.2300884955753,
      "grad_norm": 1.3248906327589793e-07,
      "learning_rate": 7.699129923599557e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22538688,
      "step": 39685
    },
    {
      "epoch": 696.3185840707964,
      "grad_norm": 1.2098585955300223e-07,
      "learning_rate": 7.457426147663982e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22542336,
      "step": 39690
    },
    {
      "epoch": 696.4070796460177,
      "grad_norm": 8.112408522720216e-08,
      "learning_rate": 7.219576535871797e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22545184,
      "step": 39695
    },
    {
      "epoch": 696.4955752212389,
      "grad_norm": 1.0132735894785583e-07,
      "learning_rate": 6.985581124896445e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22547920,
      "step": 39700
    },
    {
      "epoch": 696.5840707964602,
      "grad_norm": 3.676234427985037e-07,
      "learning_rate": 6.755439950828501e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22550592,
      "step": 39705
    },
    {
      "epoch": 696.6725663716815,
      "grad_norm": 6.179024580887926e-08,
      "learning_rate": 6.5291530491562444e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22553216,
      "step": 39710
    },
    {
      "epoch": 696.7610619469026,
      "grad_norm": 1.203746649025561e-07,
      "learning_rate": 6.3067204547739845e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22556144,
      "step": 39715
    },
    {
      "epoch": 696.8495575221239,
      "grad_norm": 1.1574952623050194e-06,
      "learning_rate": 6.088142201987612e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22559264,
      "step": 39720
    },
    {
      "epoch": 696.9380530973451,
      "grad_norm": 1.034034369240544e-07,
      "learning_rate": 5.873418324503499e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22562016,
      "step": 39725
    },
    {
      "epoch": 697.0176991150443,
      "grad_norm": 1.1427209756220691e-07,
      "learning_rate": 5.6625488554340465e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22564280,
      "step": 39730
    },
    {
      "epoch": 697.1061946902655,
      "grad_norm": 7.479910379970534e-08,
      "learning_rate": 5.455533827297688e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22567400,
      "step": 39735
    },
    {
      "epoch": 697.1946902654868,
      "grad_norm": 2.8030305543325085e-07,
      "learning_rate": 5.252373272018885e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22570520,
      "step": 39740
    },
    {
      "epoch": 697.2831858407079,
      "grad_norm": 1.9901148107237532e-07,
      "learning_rate": 5.053067220925356e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22573336,
      "step": 39745
    },
    {
      "epoch": 697.3716814159292,
      "grad_norm": 1.4250589401854086e-06,
      "learning_rate": 4.857615704759177e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22576152,
      "step": 39750
    },
    {
      "epoch": 697.4601769911504,
      "grad_norm": 5.421242349257227e-06,
      "learning_rate": 4.666018753654577e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22578920,
      "step": 39755
    },
    {
      "epoch": 697.5486725663717,
      "grad_norm": 1.979461785595049e-06,
      "learning_rate": 4.478276397162917e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22581352,
      "step": 39760
    },
    {
      "epoch": 697.637168141593,
      "grad_norm": 9.526127087156055e-08,
      "learning_rate": 4.294388664233262e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22584152,
      "step": 39765
    },
    {
      "epoch": 697.7256637168142,
      "grad_norm": 2.6545936293587147e-07,
      "learning_rate": 4.114355583223484e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22587384,
      "step": 39770
    },
    {
      "epoch": 697.8141592920354,
      "grad_norm": 8.685766061944378e-08,
      "learning_rate": 3.9381771818974845e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22590296,
      "step": 39775
    },
    {
      "epoch": 697.9026548672566,
      "grad_norm": 1.7519164430268575e-07,
      "learning_rate": 3.765853487427973e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22593032,
      "step": 39780
    },
    {
      "epoch": 697.9911504424779,
      "grad_norm": 8.282357555344788e-08,
      "learning_rate": 3.5973845263825857e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22596136,
      "step": 39785
    },
    {
      "epoch": 698.070796460177,
      "grad_norm": 8.379107896416826e-08,
      "learning_rate": 3.4327703247488684e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22598352,
      "step": 39790
    },
    {
      "epoch": 698.1592920353983,
      "grad_norm": 1.4689366025777417e-06,
      "learning_rate": 3.2720109079037443e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22601024,
      "step": 39795
    },
    {
      "epoch": 698.2477876106195,
      "grad_norm": 8.947438345785486e-08,
      "learning_rate": 3.1151063006468193e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22604096,
      "step": 39800
    },
    {
      "epoch": 698.2477876106195,
      "eval_loss": 0.909034013748169,
      "eval_runtime": 1.0651,
      "eval_samples_per_second": 23.471,
      "eval_steps_per_second": 12.205,
      "num_input_tokens_seen": 22604096,
      "step": 39800
    },
    {
      "epoch": 698.3362831858407,
      "grad_norm": 6.517389465443557e-06,
      "learning_rate": 2.962056527169854e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22607040,
      "step": 39805
    },
    {
      "epoch": 698.4247787610619,
      "grad_norm": 3.814282365510735e-07,
      "learning_rate": 2.8128616110761898e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22609568,
      "step": 39810
    },
    {
      "epoch": 698.5132743362832,
      "grad_norm": 1.51110000956578e-07,
      "learning_rate": 2.6675215753724223e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22612320,
      "step": 39815
    },
    {
      "epoch": 698.6017699115044,
      "grad_norm": 1.2700297702394892e-06,
      "learning_rate": 2.5260364424739557e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22615328,
      "step": 39820
    },
    {
      "epoch": 698.6902654867257,
      "grad_norm": 8.981282917375211e-08,
      "learning_rate": 2.3884062341994475e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22618784,
      "step": 39825
    },
    {
      "epoch": 698.7787610619469,
      "grad_norm": 1.1413629863454844e-07,
      "learning_rate": 2.25463097177081e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22621376,
      "step": 39830
    },
    {
      "epoch": 698.8672566371681,
      "grad_norm": 8.95367335829178e-08,
      "learning_rate": 2.1247106758215397e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22623936,
      "step": 39835
    },
    {
      "epoch": 698.9557522123894,
      "grad_norm": 1.557972808541308e-07,
      "learning_rate": 1.998645366382834e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22627120,
      "step": 39840
    },
    {
      "epoch": 699.0353982300885,
      "grad_norm": 1.0642645520420047e-07,
      "learning_rate": 1.876435062897475e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22629960,
      "step": 39845
    },
    {
      "epoch": 699.1238938053098,
      "grad_norm": 1.814490673268665e-07,
      "learning_rate": 1.758079784211497e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22632840,
      "step": 39850
    },
    {
      "epoch": 699.212389380531,
      "grad_norm": 9.2095967829664e-07,
      "learning_rate": 1.6435795485797434e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22635656,
      "step": 39855
    },
    {
      "epoch": 699.3008849557522,
      "grad_norm": 1.9230648717893928e-07,
      "learning_rate": 1.5329343736547596e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22638632,
      "step": 39860
    },
    {
      "epoch": 699.3893805309734,
      "grad_norm": 1.305937189499673e-07,
      "learning_rate": 1.4261442765006739e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22641400,
      "step": 39865
    },
    {
      "epoch": 699.4778761061947,
      "grad_norm": 1.0999904986874753e-07,
      "learning_rate": 1.3232092735876445e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22644264,
      "step": 39870
    },
    {
      "epoch": 699.566371681416,
      "grad_norm": 8.7850395402711e-07,
      "learning_rate": 1.2241293807918607e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22647128,
      "step": 39875
    },
    {
      "epoch": 699.6548672566372,
      "grad_norm": 7.32073885956197e-08,
      "learning_rate": 1.128904613387216e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22649928,
      "step": 39880
    },
    {
      "epoch": 699.7433628318585,
      "grad_norm": 1.479298958884101e-07,
      "learning_rate": 1.0375349860591853e-09,
      "loss": 0.0,
      "num_input_tokens_seen": 22652840,
      "step": 39885
    },
    {
      "epoch": 699.8318584070796,
      "grad_norm": 1.03695136033366e-07,
      "learning_rate": 9.5002051290205e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22655384,
      "step": 39890
    },
    {
      "epoch": 699.9203539823009,
      "grad_norm": 2.817580934788566e-06,
      "learning_rate": 8.663612074077954e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22658232,
      "step": 39895
    },
    {
      "epoch": 700.0,
      "grad_norm": 6.010787910781801e-05,
      "learning_rate": 7.865570824799884e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22660488,
      "step": 39900
    },
    {
      "epoch": 700.0884955752213,
      "grad_norm": 6.235112550712074e-07,
      "learning_rate": 7.106081504254514e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22663336,
      "step": 39905
    },
    {
      "epoch": 700.1769911504425,
      "grad_norm": 2.667025000846479e-07,
      "learning_rate": 6.385144229570372e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22666120,
      "step": 39910
    },
    {
      "epoch": 700.2654867256637,
      "grad_norm": 1.6694718851795187e-07,
      "learning_rate": 5.70275911190854e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22668648,
      "step": 39915
    },
    {
      "epoch": 700.3539823008849,
      "grad_norm": 6.91408104103175e-08,
      "learning_rate": 5.058926256490403e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22671464,
      "step": 39920
    },
    {
      "epoch": 700.4424778761062,
      "grad_norm": 7.93124357301167e-08,
      "learning_rate": 4.4536457626254134e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22674360,
      "step": 39925
    },
    {
      "epoch": 700.5309734513274,
      "grad_norm": 1.8227547116111964e-07,
      "learning_rate": 3.88691772365557e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22677432,
      "step": 39930
    },
    {
      "epoch": 700.6194690265487,
      "grad_norm": 1.1157212043144682e-07,
      "learning_rate": 3.358742226955425e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22680296,
      "step": 39935
    },
    {
      "epoch": 700.70796460177,
      "grad_norm": 6.206837355193784e-08,
      "learning_rate": 2.8691193539875925e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22683304,
      "step": 39940
    },
    {
      "epoch": 700.7964601769911,
      "grad_norm": 1.354144205834018e-05,
      "learning_rate": 2.418049180274995e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22686520,
      "step": 39945
    },
    {
      "epoch": 700.8849557522124,
      "grad_norm": 5.8301264260762764e-08,
      "learning_rate": 2.005531775373104e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22689240,
      "step": 39950
    },
    {
      "epoch": 700.9734513274336,
      "grad_norm": 1.2366233193006337e-07,
      "learning_rate": 1.6315672028699435e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22692696,
      "step": 39955
    },
    {
      "epoch": 701.0530973451328,
      "grad_norm": 1.5420874888150138e-07,
      "learning_rate": 1.2961555204693555e-10,
      "loss": 0.0,
      "num_input_tokens_seen": 22694888,
      "step": 39960
    },
    {
      "epoch": 701.141592920354,
      "grad_norm": 7.452680961250735e-08,
      "learning_rate": 9.992967798799768e-11,
      "loss": 0.0,
      "num_input_tokens_seen": 22698120,
      "step": 39965
    },
    {
      "epoch": 701.2300884955753,
      "grad_norm": 5.9450005807093476e-08,
      "learning_rate": 7.409910268707521e-11,
      "loss": 0.0,
      "num_input_tokens_seen": 22701000,
      "step": 39970
    },
    {
      "epoch": 701.3185840707964,
      "grad_norm": 3.6922219237567333e-07,
      "learning_rate": 5.212383012986877e-11,
      "loss": 0.0,
      "num_input_tokens_seen": 22704392,
      "step": 39975
    },
    {
      "epoch": 701.4070796460177,
      "grad_norm": 2.2077928463204444e-07,
      "learning_rate": 3.400386370533415e-11,
      "loss": 0.0,
      "num_input_tokens_seen": 22707096,
      "step": 39980
    },
    {
      "epoch": 701.4955752212389,
      "grad_norm": 2.34843810176244e-06,
      "learning_rate": 1.9739206205682258e-11,
      "loss": 0.0,
      "num_input_tokens_seen": 22710184,
      "step": 39985
    },
    {
      "epoch": 701.5840707964602,
      "grad_norm": 1.3546009824949579e-07,
      "learning_rate": 9.329859829154685e-12,
      "loss": 0.0,
      "num_input_tokens_seen": 22713240,
      "step": 39990
    },
    {
      "epoch": 701.6725663716815,
      "grad_norm": 1.1771578556363238e-07,
      "learning_rate": 2.7758261855748148e-12,
      "loss": 0.0,
      "num_input_tokens_seen": 22715928,
      "step": 39995
    },
    {
      "epoch": 701.7610619469026,
      "grad_norm": 9.129552722697554e-07,
      "learning_rate": 7.710628524559838e-14,
      "loss": 0.0,
      "num_input_tokens_seen": 22718312,
      "step": 40000
    },
    {
      "epoch": 701.7610619469026,
      "eval_loss": 0.9119619131088257,
      "eval_runtime": 1.0686,
      "eval_samples_per_second": 23.395,
      "eval_steps_per_second": 12.166,
      "num_input_tokens_seen": 22718312,
      "step": 40000
    },
    {
      "epoch": 701.7610619469026,
      "num_input_tokens_seen": 22718312,
      "step": 40000,
      "total_flos": 9.574707253860557e+16,
      "train_loss": 0.0016468607849636553,
      "train_runtime": 17323.8189,
      "train_samples_per_second": 9.236,
      "train_steps_per_second": 2.309
    }
  ],
  "logging_steps": 5,
  "max_steps": 40000,
  "num_input_tokens_seen": 22718312,
  "num_train_epochs": 715,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.574707253860557e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}