{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.06666666666666667,
  "eval_steps": 50,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.3988752365112305,
      "epoch": 0.0013333333333333333,
      "grad_norm": 0.01450820083840917,
      "importance_ratio": 0.9983458518981934,
      "learning_rate": 0.0,
      "loss": -0.0028,
      "mismatch_kl": 0.004329901188611984,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 1,
      "timing/generation_ms": 12196.653502061963,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 12196.653502061963,
      "tokens/completion": 562.04296875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 163.39810061454773
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.0297880172729492,
      "epoch": 0.0026666666666666666,
      "grad_norm": 0.006125098422428371,
      "importance_ratio": 0.9977808594703674,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0118,
      "mismatch_kl": 0.0036596579011529684,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 2,
      "timing/generation_ms": 10855.522208847106,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 10855.522208847106,
      "tokens/completion": 652.203125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 98.15957498550415
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.2343848943710327,
      "epoch": 0.004,
      "grad_norm": 0.0093110934895908,
      "importance_ratio": 0.9983258843421936,
      "learning_rate": 2.0000000000000003e-06,
      "loss": -0.0068,
      "mismatch_kl": 0.00391958886757493,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 3,
      "timing/generation_ms": 14581.869984045625,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 14581.869984045625,
      "tokens/completion": 722.37109375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 171.60404181480408
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.581649661064148,
      "epoch": 0.005333333333333333,
      "grad_norm": 0.007696628408420481,
      "importance_ratio": 0.9986447095870972,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "mismatch_kl": 0.0024762798566371202,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 4,
      "timing/generation_ms": 11191.347393207252,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 11191.347393207252,
      "tokens/completion": 595.73046875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 186.33580946922302
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8588891625404358,
      "epoch": 0.006666666666666667,
      "grad_norm": 0.0055080213738955075,
      "importance_ratio": 0.9988943934440613,
      "learning_rate": 4.000000000000001e-06,
      "loss": -0.0033,
      "mismatch_kl": 0.0031517043244093657,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 5,
      "timing/generation_ms": 10668.582463636994,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 10668.582463636994,
      "tokens/completion": 636.53125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 82.27488708496094
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.071407675743103,
      "epoch": 0.008,
      "grad_norm": 0.02271832942623967,
      "importance_ratio": 0.998067319393158,
      "learning_rate": 5e-06,
      "loss": 0.0019,
      "mismatch_kl": 0.003643231000751257,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 6,
      "timing/generation_ms": 3378.591795451939,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 3378.591795451939,
      "tokens/completion": 178.73828125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 105.60203862190247
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8531922698020935,
      "epoch": 0.009333333333333334,
      "grad_norm": 0.018354067998903482,
      "importance_ratio": 0.9980432391166687,
      "learning_rate": 5e-06,
      "loss": -0.0002,
      "mismatch_kl": 0.003655636915937066,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 7,
      "timing/generation_ms": 12279.695899225771,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 12279.695899225771,
      "tokens/completion": 631.35546875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 104.63563537597656
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7072162628173828,
      "epoch": 0.010666666666666666,
      "grad_norm": 0.005552532749027135,
      "importance_ratio": 0.9982293844223022,
      "learning_rate": 5e-06,
      "loss": -0.0014,
      "mismatch_kl": 0.0029395928140729666,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 8,
      "timing/generation_ms": 6614.743183366954,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 6614.743183366954,
      "tokens/completion": 339.1640625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 132.164165019989
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8156145215034485,
      "epoch": 0.012,
      "grad_norm": 0.008176505226750404,
      "importance_ratio": 0.9981797933578491,
      "learning_rate": 5e-06,
      "loss": 0.0027,
      "mismatch_kl": 0.0031279518734663725,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 9,
      "timing/generation_ms": 8826.908372342587,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 8826.908372342587,
      "tokens/completion": 444.53515625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 144.61542773246765
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8708666563034058,
      "epoch": 0.013333333333333334,
      "grad_norm": 0.009382372847258274,
      "importance_ratio": 0.9981642961502075,
      "learning_rate": 5e-06,
      "loss": 0.0126,
      "mismatch_kl": 0.0030885515734553337,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 10,
      "timing/generation_ms": 7367.4805322662,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 7367.4805322662,
      "tokens/completion": 400.74609375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 176.57727003097534
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.6906348466873169,
      "epoch": 0.014666666666666666,
      "grad_norm": 0.007616251351947248,
      "importance_ratio": 1.0045424699783325,
      "learning_rate": 5e-06,
      "loss": 0.0542,
      "mismatch_kl": 0.03194786608219147,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 11,
      "timing/generation_ms": 26879.562875255942,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 26879.562875255942,
      "tokens/completion": 1682.0546875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 190.3386266231537
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.6506091356277466,
      "epoch": 0.016,
      "grad_norm": 0.004382353798954015,
      "importance_ratio": 0.9982648491859436,
      "learning_rate": 5e-06,
      "loss": 0.043,
      "mismatch_kl": 0.02482638508081436,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 12,
      "timing/generation_ms": 22301.60311050713,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 22301.60311050713,
      "tokens/completion": 1387.734375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 177.6784646511078
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.143129825592041,
      "epoch": 0.017333333333333333,
      "grad_norm": 0.009321138085104996,
      "importance_ratio": 1.001217007637024,
      "learning_rate": 5e-06,
      "loss": -0.0139,
      "mismatch_kl": 0.0036374337505549192,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 13,
      "timing/generation_ms": 6277.724616229534,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 6277.724616229534,
      "tokens/completion": 432.66796875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 129.69259929656982
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.650863766670227,
      "epoch": 0.018666666666666668,
      "grad_norm": 0.0076614251264825245,
      "importance_ratio": 0.9983827471733093,
      "learning_rate": 5e-06,
      "loss": 0.0049,
      "mismatch_kl": 0.0027237425092607737,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 14,
      "timing/generation_ms": 7103.812717832625,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 7103.812717832625,
      "tokens/completion": 404.96484375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 57.65379452705383
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7439635992050171,
      "epoch": 0.02,
      "grad_norm": 0.009401568464987338,
      "importance_ratio": 0.9981654286384583,
      "learning_rate": 5e-06,
      "loss": 0.0109,
      "mismatch_kl": 0.002909082220867276,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 15,
      "timing/generation_ms": 8292.532542720437,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 8292.532542720437,
      "tokens/completion": 459.85546875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 92.21157336235046
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7638830542564392,
      "epoch": 0.021333333333333333,
      "grad_norm": 0.010374572910211358,
      "importance_ratio": 0.9969711899757385,
      "learning_rate": 5e-06,
      "loss": -0.005,
      "mismatch_kl": 0.0034673516638576984,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 16,
      "timing/generation_ms": 5712.000676430762,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 5712.000676430762,
      "tokens/completion": 308.4296875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 52.6866238117218
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.18189160525798798,
      "epoch": 0.02266666666666667,
      "grad_norm": 0.00257455457059234,
      "importance_ratio": 0.9984971880912781,
      "learning_rate": 5e-06,
      "loss": 0.0681,
      "mismatch_kl": 0.018514186143875122,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 17,
      "timing/generation_ms": 14281.423358246684,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 14281.423358246684,
      "tokens/completion": 1101.546875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 129.24327325820923
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.5917271375656128,
      "epoch": 0.024,
      "grad_norm": 0.005668903472483887,
      "importance_ratio": 0.999828577041626,
      "learning_rate": 5e-06,
      "loss": 0.027,
      "mismatch_kl": 0.002100760815665126,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 18,
      "timing/generation_ms": 24175.399120897055,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 24175.399120897055,
      "tokens/completion": 1504.0390625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 158.57504653930664
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.3282008171081543,
      "epoch": 0.025333333333333333,
      "grad_norm": 0.006636047431888786,
      "importance_ratio": 1.0022344589233398,
      "learning_rate": 5e-06,
      "loss": -0.0015,
      "mismatch_kl": 0.004634195473045111,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 19,
      "timing/generation_ms": 15713.139976374805,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 15713.139976374805,
      "tokens/completion": 764.3203125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 78.56244468688965
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.02470862865448,
      "epoch": 0.02666666666666667,
      "grad_norm": 0.00833481021786943,
      "importance_ratio": 1.0026451349258423,
      "learning_rate": 5e-06,
      "loss": -0.0052,
      "mismatch_kl": 0.004158638883382082,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 20,
      "timing/generation_ms": 6632.851202040911,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 6632.851202040911,
      "tokens/completion": 382.6171875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 199.43552422523499
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.07275530695915222,
      "epoch": 0.028,
      "grad_norm": 0.005944388738403685,
      "importance_ratio": 0.9988561868667603,
      "learning_rate": 5e-06,
      "loss": 0.0452,
      "mismatch_kl": 0.00023643655003979802,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 21,
      "timing/generation_ms": 119174.6030151844,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 119174.6030151844,
      "tokens/completion": 4008.38671875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 685.4423098564148
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.83598792552948,
      "epoch": 0.029333333333333333,
      "grad_norm": 0.00977617477475085,
      "importance_ratio": 0.995696485042572,
      "learning_rate": 5e-06,
      "loss": 0.0066,
      "mismatch_kl": 0.003957619424909353,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 22,
      "timing/generation_ms": 12322.44247943163,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 12322.44247943163,
      "tokens/completion": 442.85546875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 94.37256598472595
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8219886422157288,
      "epoch": 0.030666666666666665,
      "grad_norm": 0.0057449599218849946,
      "importance_ratio": 0.9990558624267578,
      "learning_rate": 5e-06,
      "loss": 0.0388,
      "mismatch_kl": 0.031180420890450478,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 23,
      "timing/generation_ms": 29090.628595091403,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 29090.628595091403,
      "tokens/completion": 1716.3515625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 169.45334482192993
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.0089167356491089,
      "epoch": 0.032,
      "grad_norm": 0.009762837519367,
      "importance_ratio": 0.9979202151298523,
      "learning_rate": 5e-06,
      "loss": 0.0012,
      "mismatch_kl": 0.00405939482152462,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 24,
      "timing/generation_ms": 17154.327374882996,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 17154.327374882996,
      "tokens/completion": 883.390625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 130.7891206741333
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.5053093433380127,
      "epoch": 0.03333333333333333,
      "grad_norm": 0.007416974683241316,
      "importance_ratio": 0.9982149600982666,
      "learning_rate": 5e-06,
      "loss": -0.0068,
      "mismatch_kl": 0.0024536694400012493,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 25,
      "timing/generation_ms": 28463.361867703497,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 28463.361867703497,
      "tokens/completion": 1409.54296875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 178.60342526435852
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.4973055422306061,
      "epoch": 0.034666666666666665,
      "grad_norm": 0.004048717808220336,
      "importance_ratio": 1.0012173652648926,
      "learning_rate": 5e-06,
      "loss": 0.0547,
      "mismatch_kl": 0.03473234549164772,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 26,
      "timing/generation_ms": 18848.746892996132,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 18848.746892996132,
      "tokens/completion": 1286.6875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 181.75563287734985
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.5914682149887085,
      "epoch": 0.036,
      "grad_norm": 0.010568088931367656,
      "importance_ratio": 0.9986244440078735,
      "learning_rate": 5e-06,
      "loss": -0.0214,
      "mismatch_kl": 0.002536088228225708,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 27,
      "timing/generation_ms": 11602.461927570403,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 11602.461927570403,
      "tokens/completion": 734.80078125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 188.88015818595886
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.4526905119419098,
      "epoch": 0.037333333333333336,
      "grad_norm": 0.0035728175606856527,
      "importance_ratio": 0.9999799728393555,
      "learning_rate": 5e-06,
      "loss": 0.0026,
      "mismatch_kl": 0.0024842985440045595,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 28,
      "timing/generation_ms": 30549.59301650524,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 30549.59301650524,
      "tokens/completion": 1536.96875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 328.0478210449219
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.32794511318206787,
      "epoch": 0.03866666666666667,
      "grad_norm": 0.003333518820406266,
      "importance_ratio": 0.9995192885398865,
      "learning_rate": 5e-06,
      "loss": 0.056,
      "mismatch_kl": 0.028769802302122116,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 29,
      "timing/generation_ms": 18838.333567604423,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 18838.333567604423,
      "tokens/completion": 1263.640625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 290.5948350429535
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.9063822031021118,
      "epoch": 0.04,
      "grad_norm": 0.007342388496075293,
      "importance_ratio": 0.9953157901763916,
      "learning_rate": 5e-06,
      "loss": 0.0025,
      "mismatch_kl": 0.004266439005732536,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 30,
      "timing/generation_ms": 9477.213966660202,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 9477.213966660202,
      "tokens/completion": 473.26953125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 62.30127143859863
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7977282404899597,
      "epoch": 0.04133333333333333,
      "grad_norm": 0.00884043332375607,
      "importance_ratio": 0.9971498847007751,
      "learning_rate": 5e-06,
      "loss": -0.0029,
      "mismatch_kl": 0.004033135715872049,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 31,
      "timing/generation_ms": 18995.201839134097,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 18995.201839134097,
      "tokens/completion": 958.625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 88.6347918510437
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8451470732688904,
      "epoch": 0.042666666666666665,
      "grad_norm": 0.018842389370386323,
      "importance_ratio": 0.9982671141624451,
      "learning_rate": 5e-06,
      "loss": 0.0369,
      "mismatch_kl": 0.003600390162318945,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 32,
      "timing/generation_ms": 5587.277088314295,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 5587.277088314295,
      "tokens/completion": 407.12109375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 140.4788475036621
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.1521912813186646,
      "epoch": 0.044,
      "grad_norm": 0.006379742039913797,
      "importance_ratio": 0.997858464717865,
      "learning_rate": 5e-06,
      "loss": -0.0065,
      "mismatch_kl": 0.005035887472331524,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 33,
      "timing/generation_ms": 18916.152058169246,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 18916.152058169246,
      "tokens/completion": 966.55859375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 90.41954302787781
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.1553761959075928,
      "epoch": 0.04533333333333334,
      "grad_norm": 0.010733713274389883,
      "importance_ratio": 1.0111567974090576,
      "learning_rate": 5e-06,
      "loss": 0.0014,
      "mismatch_kl": 0.006704343948513269,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 34,
      "timing/generation_ms": 17302.85968258977,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 17302.85968258977,
      "tokens/completion": 864.44140625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 143.8659963607788
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.3105199635028839,
      "epoch": 0.04666666666666667,
      "grad_norm": 0.003940130100379767,
      "importance_ratio": 1.0006911754608154,
      "learning_rate": 5e-06,
      "loss": 0.0315,
      "mismatch_kl": 0.022524980828166008,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 35,
      "timing/generation_ms": 29806.298807263374,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 29806.298807263374,
      "tokens/completion": 1672.48828125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 164.04821372032166
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7972971200942993,
      "epoch": 0.048,
      "grad_norm": 0.008409173142054645,
      "importance_ratio": 0.9948906898498535,
      "learning_rate": 5e-06,
      "loss": 0.004,
      "mismatch_kl": 0.004282351583242416,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 36,
      "timing/generation_ms": 14936.399303376675,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 14936.399303376675,
      "tokens/completion": 787.78125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 196.79586815834045
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.5769950747489929,
      "epoch": 0.04933333333333333,
      "grad_norm": 0.009636377939254703,
      "importance_ratio": 0.9972301721572876,
      "learning_rate": 5e-06,
      "loss": -0.0011,
      "mismatch_kl": 0.003603809280321002,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 37,
      "timing/generation_ms": 13729.571803472936,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 13729.571803472936,
      "tokens/completion": 697.64453125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 76.77378511428833
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.715777575969696,
      "epoch": 0.050666666666666665,
      "grad_norm": 0.005305945077729364,
      "importance_ratio": 0.9969701766967773,
      "learning_rate": 5e-06,
      "loss": 0.0093,
      "mismatch_kl": 0.004232620354741812,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 38,
      "timing/generation_ms": 26689.202761277556,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 26689.202761277556,
      "tokens/completion": 1302.75390625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 119.53459739685059
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.108477234840393,
      "epoch": 0.052,
      "grad_norm": 0.01158876392732835,
      "importance_ratio": 0.9918505549430847,
      "learning_rate": 5e-06,
      "loss": 0.0069,
      "mismatch_kl": 0.0055715711787343025,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 39,
      "timing/generation_ms": 9316.26115180552,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 9316.26115180552,
      "tokens/completion": 510.88671875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 134.0537760257721
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.0468562841415405,
      "epoch": 0.05333333333333334,
      "grad_norm": 0.006250915142780056,
      "importance_ratio": 0.993874192237854,
      "learning_rate": 5e-06,
      "loss": -0.004,
      "mismatch_kl": 0.00569565873593092,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 40,
      "timing/generation_ms": 22657.30178449303,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 22657.30178449303,
      "tokens/completion": 1117.0390625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 104.31990480422974
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 1.0242066383361816,
      "epoch": 0.05466666666666667,
      "grad_norm": 0.009730238448609988,
      "importance_ratio": 1.0014866590499878,
      "learning_rate": 5e-06,
      "loss": 0.0029,
      "mismatch_kl": 0.006813807878643274,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 41,
      "timing/generation_ms": 15266.15516282618,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 15266.15516282618,
      "tokens/completion": 789.296875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 137.6475269794464
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.9917812943458557,
      "epoch": 0.056,
      "grad_norm": 0.015130940878153589,
      "importance_ratio": 0.9915910959243774,
      "learning_rate": 5e-06,
      "loss": -0.0016,
      "mismatch_kl": 0.006494010798633099,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 42,
      "timing/generation_ms": 8552.51188017428,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 8552.51188017428,
      "tokens/completion": 427.00390625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 58.246270418167114
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7529230117797852,
      "epoch": 0.05733333333333333,
      "grad_norm": 0.017225340266775354,
      "importance_ratio": 0.9983583092689514,
      "learning_rate": 5e-06,
      "loss": 0.0017,
      "mismatch_kl": 0.005849814508110285,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 43,
      "timing/generation_ms": 5776.03021170944,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 5776.03021170944,
      "tokens/completion": 292.6484375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 138.62879586219788
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.9116057753562927,
      "epoch": 0.058666666666666666,
      "grad_norm": 0.013240792131345649,
      "importance_ratio": 0.993713915348053,
      "learning_rate": 5e-06,
      "loss": -0.006,
      "mismatch_kl": 0.00599726801738143,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 44,
      "timing/generation_ms": 4909.729053266346,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 4909.729053266346,
      "tokens/completion": 252.2890625,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 53.461458683013916
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.6952740550041199,
      "epoch": 0.06,
      "grad_norm": 0.007271643900369788,
      "importance_ratio": 0.9978048205375671,
      "learning_rate": 5e-06,
      "loss": -0.0094,
      "mismatch_kl": 0.004028764553368092,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 45,
      "timing/generation_ms": 12042.251928709447,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 12042.251928709447,
      "tokens/completion": 668.03125,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 77.72424340248108
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8735002279281616,
      "epoch": 0.06133333333333333,
      "grad_norm": 0.00817327643152143,
      "importance_ratio": 1.0016076564788818,
      "learning_rate": 5e-06,
      "loss": 0.0002,
      "mismatch_kl": 0.004535754211246967,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 46,
      "timing/generation_ms": 8553.523855283856,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 8553.523855283856,
      "tokens/completion": 459.87109375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 104.68091750144958
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7288662195205688,
      "epoch": 0.06266666666666666,
      "grad_norm": 0.016180435920793518,
      "importance_ratio": 1.0001567602157593,
      "learning_rate": 5e-06,
      "loss": 0.0002,
      "mismatch_kl": 0.006666674744337797,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 47,
      "timing/generation_ms": 7466.575676575303,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 7466.575676575303,
      "tokens/completion": 361.484375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 59.11225175857544
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.6449630856513977,
      "epoch": 0.064,
      "grad_norm": 0.004581873635760183,
      "importance_ratio": 1.0026441812515259,
      "learning_rate": 5e-06,
      "loss": 0.0588,
      "mismatch_kl": 0.059744831174612045,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 48,
      "timing/generation_ms": 14945.35976741463,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 14945.35976741463,
      "tokens/completion": 1044.0,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 182.3247947692871
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.8048098683357239,
      "epoch": 0.06533333333333333,
      "grad_norm": 0.0052364810032066635,
      "importance_ratio": 0.9973055720329285,
      "learning_rate": 5e-06,
      "loss": 0.0347,
      "mismatch_kl": 0.0451083704829216,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 49,
      "timing/generation_ms": 28440.53523708135,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 28440.53523708135,
      "tokens/completion": 1630.4296875,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 221.0147523880005
    },
    {
      "advantage/absmean": 0.12451171875,
      "entropy": 0.7735000252723694,
      "epoch": 0.06666666666666667,
      "grad_norm": 0.015103596816141955,
      "importance_ratio": 0.9899436831474304,
      "learning_rate": 5e-06,
      "loss": -0.0022,
      "mismatch_kl": 0.008240272291004658,
      "reward": 0.12451171875,
      "reward/std": 0.1738164722919464,
      "step": 50,
      "timing/generation_ms": 6061.147706583142,
      "timing/scoring_ms": 0.0,
      "timing/total_ms": 6061.147706583142,
      "tokens/completion": 331.859375,
      "tokens/masked_fraction": 0.0,
      "wall_clock/generate_s": 108.05560183525085
    }
  ],
  "logging_steps": 1,
  "max_steps": 750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 5,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}