{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 385,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 792.0,
      "learning_rate": 1.282051282051282e-07,
      "logits/chosen": -2.7358343601226807,
      "logits/rejected": -2.7480404376983643,
      "logps/chosen": -27.35565757751465,
      "logps/rejected": -21.06114387512207,
      "loss": 25.0,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.03,
      "grad_norm": 1384.0,
      "learning_rate": 1.282051282051282e-06,
      "logits/chosen": -3.0097055435180664,
      "logits/rejected": -2.9980692863464355,
      "logps/chosen": -33.19136047363281,
      "logps/rejected": -31.962371826171875,
      "loss": 25.1536,
      "rewards/accuracies": 0.4166666567325592,
      "rewards/chosen": -0.0010418971069157124,
      "rewards/margins": -0.001195515040308237,
      "rewards/rejected": 0.00015361772966571152,
      "step": 10
    },
    {
      "epoch": 0.05,
      "grad_norm": 940.0,
      "learning_rate": 2.564102564102564e-06,
      "logits/chosen": -2.899199962615967,
      "logits/rejected": -2.893846273422241,
      "logps/chosen": -32.49421691894531,
      "logps/rejected": -28.94989013671875,
      "loss": 25.5664,
      "rewards/accuracies": 0.44999998807907104,
      "rewards/chosen": -0.0019120589131489396,
      "rewards/margins": -0.005109794903546572,
      "rewards/rejected": 0.0031977363396435976,
      "step": 20
    },
    {
      "epoch": 0.08,
      "grad_norm": 872.0,
      "learning_rate": 3.846153846153847e-06,
      "logits/chosen": -3.0976572036743164,
      "logits/rejected": -3.1090171337127686,
      "logps/chosen": -32.78887176513672,
      "logps/rejected": -30.141361236572266,
      "loss": 24.4486,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.011981222778558731,
      "rewards/margins": 0.006511862389743328,
      "rewards/rejected": 0.005469360388815403,
      "step": 30
    },
    {
      "epoch": 0.1,
      "grad_norm": 788.0,
      "learning_rate": 4.999896948438434e-06,
      "logits/chosen": -2.8625879287719727,
      "logits/rejected": -2.853339195251465,
      "logps/chosen": -31.531911849975586,
      "logps/rejected": -32.36937713623047,
      "loss": 22.726,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": 0.02984103001654148,
      "rewards/margins": 0.031033098697662354,
      "rewards/rejected": -0.001192068331874907,
      "step": 40
    },
    {
      "epoch": 0.13,
      "grad_norm": 536.0,
      "learning_rate": 4.987541037542187e-06,
      "logits/chosen": -2.880608558654785,
      "logits/rejected": -2.8780317306518555,
      "logps/chosen": -29.381378173828125,
      "logps/rejected": -30.077362060546875,
      "loss": 23.1409,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": 0.035878658294677734,
      "rewards/margins": 0.034720782190561295,
      "rewards/rejected": 0.0011578750563785434,
      "step": 50
    },
    {
      "epoch": 0.16,
      "grad_norm": 784.0,
      "learning_rate": 4.954691471941119e-06,
      "logits/chosen": -2.9093430042266846,
      "logits/rejected": -2.910740375518799,
      "logps/chosen": -29.860986709594727,
      "logps/rejected": -28.04913902282715,
      "loss": 22.1643,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.02687840536236763,
      "rewards/margins": 0.036224402487277985,
      "rewards/rejected": -0.00934599619358778,
      "step": 60
    },
    {
      "epoch": 0.18,
      "grad_norm": 1600.0,
      "learning_rate": 4.901618883413549e-06,
      "logits/chosen": -2.9914920330047607,
      "logits/rejected": -2.99847674369812,
      "logps/chosen": -29.173349380493164,
      "logps/rejected": -30.959280014038086,
      "loss": 24.455,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": 0.01076198648661375,
      "rewards/margins": 0.017510101199150085,
      "rewards/rejected": -0.006748115178197622,
      "step": 70
    },
    {
      "epoch": 0.21,
      "grad_norm": 1072.0,
      "learning_rate": 4.828760511501322e-06,
      "logits/chosen": -2.8084774017333984,
      "logits/rejected": -2.823390007019043,
      "logps/chosen": -29.325841903686523,
      "logps/rejected": -29.870441436767578,
      "loss": 22.0145,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": 0.021488705649971962,
      "rewards/margins": 0.041104547679424286,
      "rewards/rejected": -0.01961584761738777,
      "step": 80
    },
    {
      "epoch": 0.23,
      "grad_norm": 740.0,
      "learning_rate": 4.7367166013034295e-06,
      "logits/chosen": -2.8982763290405273,
      "logits/rejected": -2.8804843425750732,
      "logps/chosen": -32.65864181518555,
      "logps/rejected": -30.32712745666504,
      "loss": 25.4545,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.017779098823666573,
      "rewards/margins": 0.04832718148827553,
      "rewards/rejected": -0.030548080801963806,
      "step": 90
    },
    {
      "epoch": 0.26,
      "grad_norm": 720.0,
      "learning_rate": 4.626245458345211e-06,
      "logits/chosen": -3.0048553943634033,
      "logits/rejected": -3.004960298538208,
      "logps/chosen": -31.79574966430664,
      "logps/rejected": -30.897558212280273,
      "loss": 22.8693,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.026112746447324753,
      "rewards/margins": 0.034251101315021515,
      "rewards/rejected": -0.008138353936374187,
      "step": 100
    },
    {
      "epoch": 0.26,
      "eval_logits/chosen": -2.8112006187438965,
      "eval_logits/rejected": -2.8086960315704346,
      "eval_logps/chosen": -31.26334571838379,
      "eval_logps/rejected": -34.84490203857422,
      "eval_loss": 23.988525390625,
      "eval_rewards/accuracies": 0.6129568219184875,
      "eval_rewards/chosen": 0.0019106707768514752,
      "eval_rewards/margins": 0.016585489735007286,
      "eval_rewards/rejected": -0.014674817211925983,
      "eval_runtime": 113.1272,
      "eval_samples_per_second": 3.032,
      "eval_steps_per_second": 0.38,
      "step": 100
    },
    {
      "epoch": 0.29,
      "grad_norm": 1008.0,
      "learning_rate": 4.498257201263691e-06,
      "logits/chosen": -2.9598517417907715,
      "logits/rejected": -2.9358468055725098,
      "logps/chosen": -31.7904109954834,
      "logps/rejected": -31.376705169677734,
      "loss": 21.3073,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.042638350278139114,
      "rewards/margins": 0.06442411243915558,
      "rewards/rejected": -0.02178577333688736,
      "step": 110
    },
    {
      "epoch": 0.31,
      "grad_norm": 792.0,
      "learning_rate": 4.353806263777678e-06,
      "logits/chosen": -3.0428009033203125,
      "logits/rejected": -3.071193218231201,
      "logps/chosen": -28.65635108947754,
      "logps/rejected": -34.34435272216797,
      "loss": 20.586,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.05075501650571823,
      "rewards/margins": 0.07043538987636566,
      "rewards/rejected": -0.01968037523329258,
      "step": 120
    },
    {
      "epoch": 0.34,
      "grad_norm": 668.0,
      "learning_rate": 4.1940827077152755e-06,
      "logits/chosen": -2.744912624359131,
      "logits/rejected": -2.740849494934082,
      "logps/chosen": -28.600677490234375,
      "logps/rejected": -30.34041404724121,
      "loss": 21.4152,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.037400223314762115,
      "rewards/margins": 0.05806100368499756,
      "rewards/rejected": -0.020660774782299995,
      "step": 130
    },
    {
      "epoch": 0.36,
      "grad_norm": 624.0,
      "learning_rate": 4.0204024186666215e-06,
      "logits/chosen": -3.0207009315490723,
      "logits/rejected": -3.0183839797973633,
      "logps/chosen": -27.157459259033203,
      "logps/rejected": -31.912410736083984,
      "loss": 22.1262,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": 0.03287653252482414,
      "rewards/margins": 0.05844072252511978,
      "rewards/rejected": -0.02556418441236019,
      "step": 140
    },
    {
      "epoch": 0.39,
      "grad_norm": 668.0,
      "learning_rate": 3.834196265035119e-06,
      "logits/chosen": -2.8160903453826904,
      "logits/rejected": -2.8107659816741943,
      "logps/chosen": -27.24477195739746,
      "logps/rejected": -31.60110092163086,
      "loss": 18.6765,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": 0.05807359144091606,
      "rewards/margins": 0.09581951797008514,
      "rewards/rejected": -0.03774593397974968,
      "step": 150
    },
    {
      "epoch": 0.42,
      "grad_norm": 872.0,
      "learning_rate": 3.636998309800573e-06,
      "logits/chosen": -3.134445905685425,
      "logits/rejected": -3.1172776222229004,
      "logps/chosen": -31.69814109802246,
      "logps/rejected": -29.522869110107422,
      "loss": 16.4703,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.07935762405395508,
      "rewards/margins": 0.1336670219898224,
      "rewards/rejected": -0.054309405386447906,
      "step": 160
    },
    {
      "epoch": 0.44,
      "grad_norm": 688.0,
      "learning_rate": 3.4304331721118078e-06,
      "logits/chosen": -2.9469733238220215,
      "logits/rejected": -2.9541804790496826,
      "logps/chosen": -29.33526039123535,
      "logps/rejected": -31.71769142150879,
      "loss": 17.9149,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.0559355802834034,
      "rewards/margins": 0.11775630712509155,
      "rewards/rejected": -0.06182072311639786,
      "step": 170
    },
    {
      "epoch": 0.47,
      "grad_norm": 688.0,
      "learning_rate": 3.2162026428305436e-06,
      "logits/chosen": -2.798689126968384,
      "logits/rejected": -2.7966549396514893,
      "logps/chosen": -29.22808265686035,
      "logps/rejected": -30.211177825927734,
      "loss": 19.5177,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.04417658969759941,
      "rewards/margins": 0.088477224111557,
      "rewards/rejected": -0.044300634413957596,
      "step": 180
    },
    {
      "epoch": 0.49,
      "grad_norm": 496.0,
      "learning_rate": 2.996071664294641e-06,
      "logits/chosen": -2.9117610454559326,
      "logits/rejected": -2.9078633785247803,
      "logps/chosen": -29.73740577697754,
      "logps/rejected": -28.741764068603516,
      "loss": 22.4526,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": 0.044930893927812576,
      "rewards/margins": 0.08335400372743607,
      "rewards/rejected": -0.03842311352491379,
      "step": 190
    },
    {
      "epoch": 0.52,
      "grad_norm": 446.0,
      "learning_rate": 2.7718537898066833e-06,
      "logits/chosen": -2.9809372425079346,
      "logits/rejected": -2.9683079719543457,
      "logps/chosen": -32.74761962890625,
      "logps/rejected": -30.57464027404785,
      "loss": 19.5946,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.09513962268829346,
      "rewards/margins": 0.10919489711523056,
      "rewards/rejected": -0.01405528374016285,
      "step": 200
    },
    {
      "epoch": 0.52,
      "eval_logits/chosen": -2.8196890354156494,
      "eval_logits/rejected": -2.817232370376587,
      "eval_logps/chosen": -31.39967155456543,
      "eval_logps/rejected": -35.00971221923828,
      "eval_loss": 23.757455825805664,
      "eval_rewards/accuracies": 0.5718438625335693,
      "eval_rewards/chosen": -0.011722045950591564,
      "eval_rewards/margins": 0.019433511421084404,
      "eval_rewards/rejected": -0.03115556202828884,
      "eval_runtime": 112.9888,
      "eval_samples_per_second": 3.036,
      "eval_steps_per_second": 0.381,
      "step": 200
    },
    {
      "epoch": 0.55,
      "grad_norm": 756.0,
      "learning_rate": 2.5453962426402006e-06,
      "logits/chosen": -2.9178690910339355,
      "logits/rejected": -2.9186930656433105,
      "logps/chosen": -32.30227279663086,
      "logps/rejected": -34.28144073486328,
      "loss": 17.7864,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.0731135681271553,
      "rewards/margins": 0.1151106134057045,
      "rewards/rejected": -0.041997045278549194,
      "step": 210
    },
    {
      "epoch": 0.57,
      "grad_norm": 584.0,
      "learning_rate": 2.3185646976551794e-06,
      "logits/chosen": -2.8974623680114746,
      "logits/rejected": -2.9132611751556396,
      "logps/chosen": -29.29262924194336,
      "logps/rejected": -28.876379013061523,
      "loss": 18.1243,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.08854089677333832,
      "rewards/margins": 0.12335184961557388,
      "rewards/rejected": -0.03481093421578407,
      "step": 220
    },
    {
      "epoch": 0.6,
      "grad_norm": 644.0,
      "learning_rate": 2.0932279108998323e-06,
      "logits/chosen": -2.9454588890075684,
      "logits/rejected": -2.9493844509124756,
      "logps/chosen": -30.615680694580078,
      "logps/rejected": -32.01018142700195,
      "loss": 20.5727,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.05361897870898247,
      "rewards/margins": 0.07882187515497208,
      "rewards/rejected": -0.02520289644598961,
      "step": 230
    },
    {
      "epoch": 0.62,
      "grad_norm": 692.0,
      "learning_rate": 1.8712423238279358e-06,
      "logits/chosen": -3.000526189804077,
      "logits/rejected": -3.007200241088867,
      "logps/chosen": -29.994653701782227,
      "logps/rejected": -30.485973358154297,
      "loss": 16.6669,
      "rewards/accuracies": 0.800000011920929,
      "rewards/chosen": 0.09102725982666016,
      "rewards/margins": 0.11610045284032822,
      "rewards/rejected": -0.02507319487631321,
      "step": 240
    },
    {
      "epoch": 0.65,
      "grad_norm": 828.0,
      "learning_rate": 1.6544367689701824e-06,
      "logits/chosen": -2.8318004608154297,
      "logits/rejected": -2.821593761444092,
      "logps/chosen": -26.242666244506836,
      "logps/rejected": -29.69923973083496,
      "loss": 19.8026,
      "rewards/accuracies": 0.762499988079071,
      "rewards/chosen": 0.07115494459867477,
      "rewards/margins": 0.08791451901197433,
      "rewards/rejected": -0.016759568825364113,
      "step": 250
    },
    {
      "epoch": 0.68,
      "grad_norm": 482.0,
      "learning_rate": 1.4445974030621963e-06,
      "logits/chosen": -2.813377857208252,
      "logits/rejected": -2.8331103324890137,
      "logps/chosen": -28.772018432617188,
      "logps/rejected": -34.57004165649414,
      "loss": 16.4634,
      "rewards/accuracies": 0.800000011920929,
      "rewards/chosen": 0.1057942658662796,
      "rewards/margins": 0.14538946747779846,
      "rewards/rejected": -0.039595186710357666,
      "step": 260
    },
    {
      "epoch": 0.7,
      "grad_norm": 792.0,
      "learning_rate": 1.243452991757889e-06,
      "logits/chosen": -2.955620527267456,
      "logits/rejected": -2.9620866775512695,
      "logps/chosen": -30.066015243530273,
      "logps/rejected": -30.39284324645996,
      "loss": 16.2985,
      "rewards/accuracies": 0.8374999761581421,
      "rewards/chosen": 0.071534164249897,
      "rewards/margins": 0.1242922693490982,
      "rewards/rejected": -0.0527581050992012,
      "step": 270
    },
    {
      "epoch": 0.73,
      "grad_norm": 552.0,
      "learning_rate": 1.0526606671603523e-06,
      "logits/chosen": -2.9709465503692627,
      "logits/rejected": -2.957740306854248,
      "logps/chosen": -30.085498809814453,
      "logps/rejected": -28.866928100585938,
      "loss": 21.3586,
      "rewards/accuracies": 0.7875000238418579,
      "rewards/chosen": 0.06361865997314453,
      "rewards/margins": 0.08717648684978485,
      "rewards/rejected": -0.02355783060193062,
      "step": 280
    },
    {
      "epoch": 0.75,
      "grad_norm": 616.0,
      "learning_rate": 8.737922755071455e-07,
      "logits/chosen": -2.89894437789917,
      "logits/rejected": -2.881261110305786,
      "logps/chosen": -31.43949317932129,
      "logps/rejected": -31.19020652770996,
      "loss": 13.7665,
      "rewards/accuracies": 0.824999988079071,
      "rewards/chosen": 0.11968123912811279,
      "rewards/margins": 0.1863711178302765,
      "rewards/rejected": -0.0666898787021637,
      "step": 290
    },
    {
      "epoch": 0.78,
      "grad_norm": 760.0,
      "learning_rate": 7.08321427484816e-07,
      "logits/chosen": -2.9006505012512207,
      "logits/rejected": -2.896925449371338,
      "logps/chosen": -31.2346134185791,
      "logps/rejected": -27.882694244384766,
      "loss": 18.7484,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.0988042801618576,
      "rewards/margins": 0.12671498954296112,
      "rewards/rejected": -0.027910714969038963,
      "step": 300
    },
    {
      "epoch": 0.78,
      "eval_logits/chosen": -2.818225860595703,
      "eval_logits/rejected": -2.81588077545166,
      "eval_logps/chosen": -31.401628494262695,
      "eval_logps/rejected": -35.010406494140625,
      "eval_loss": 23.85563087463379,
      "eval_rewards/accuracies": 0.5718438625335693,
      "eval_rewards/chosen": -0.011917698197066784,
      "eval_rewards/margins": 0.019307522103190422,
      "eval_rewards/rejected": -0.03122522123157978,
      "eval_runtime": 113.0254,
      "eval_samples_per_second": 3.035,
      "eval_steps_per_second": 0.38,
      "step": 300
    },
    {
      "epoch": 0.81,
      "grad_norm": 784.0,
      "learning_rate": 5.576113578589035e-07,
      "logits/chosen": -2.781919002532959,
      "logits/rejected": -2.7987494468688965,
      "logps/chosen": -28.510242462158203,
      "logps/rejected": -31.248886108398438,
      "loss": 17.0117,
      "rewards/accuracies": 0.8500000238418579,
      "rewards/chosen": 0.07191774249076843,
      "rewards/margins": 0.1236056461930275,
      "rewards/rejected": -0.051687903702259064,
      "step": 310
    },
    {
      "epoch": 0.83,
      "grad_norm": 524.0,
      "learning_rate": 4.229036944380913e-07,
      "logits/chosen": -3.0269644260406494,
      "logits/rejected": -3.012279748916626,
      "logps/chosen": -29.241586685180664,
      "logps/rejected": -28.835968017578125,
      "loss": 15.086,
      "rewards/accuracies": 0.8374999761581421,
      "rewards/chosen": 0.09699983894824982,
      "rewards/margins": 0.1487189531326294,
      "rewards/rejected": -0.051719121634960175,
      "step": 320
    },
    {
      "epoch": 0.86,
      "grad_norm": 360.0,
      "learning_rate": 3.053082288996112e-07,
      "logits/chosen": -2.9412882328033447,
      "logits/rejected": -2.9239907264709473,
      "logps/chosen": -27.266399383544922,
      "logps/rejected": -31.0549259185791,
      "loss": 14.0005,
      "rewards/accuracies": 0.862500011920929,
      "rewards/chosen": 0.10956673324108124,
      "rewards/margins": 0.1758948117494583,
      "rewards/rejected": -0.06632807105779648,
      "step": 330
    },
    {
      "epoch": 0.88,
      "grad_norm": 716.0,
      "learning_rate": 2.0579377374915805e-07,
      "logits/chosen": -3.158176898956299,
      "logits/rejected": -3.163879871368408,
      "logps/chosen": -30.825565338134766,
      "logps/rejected": -33.42596435546875,
      "loss": 15.6936,
      "rewards/accuracies": 0.7875000238418579,
      "rewards/chosen": 0.08136396110057831,
      "rewards/margins": 0.15435989201068878,
      "rewards/rejected": -0.07299593091011047,
      "step": 340
    },
    {
      "epoch": 0.91,
      "grad_norm": 552.0,
      "learning_rate": 1.2518018074041684e-07,
      "logits/chosen": -3.034761428833008,
      "logits/rejected": -3.038020372390747,
      "logps/chosen": -29.743371963500977,
      "logps/rejected": -31.885990142822266,
      "loss": 17.0818,
      "rewards/accuracies": 0.7875000238418579,
      "rewards/chosen": 0.1063508540391922,
      "rewards/margins": 0.1436188519001007,
      "rewards/rejected": -0.03726799413561821,
      "step": 350
    },
    {
      "epoch": 0.94,
      "grad_norm": 608.0,
      "learning_rate": 6.41315865106129e-08,
      "logits/chosen": -2.8735318183898926,
      "logits/rejected": -2.8751091957092285,
      "logps/chosen": -27.433670043945312,
      "logps/rejected": -30.212413787841797,
      "loss": 15.0018,
      "rewards/accuracies": 0.800000011920929,
      "rewards/chosen": 0.12160570919513702,
      "rewards/margins": 0.1597532331943512,
      "rewards/rejected": -0.038147542625665665,
      "step": 360
    },
    {
      "epoch": 0.96,
      "grad_norm": 984.0,
      "learning_rate": 2.3150941078050325e-08,
      "logits/chosen": -2.950293779373169,
      "logits/rejected": -2.947977066040039,
      "logps/chosen": -29.780099868774414,
      "logps/rejected": -32.294410705566406,
      "loss": 17.366,
      "rewards/accuracies": 0.800000011920929,
      "rewards/chosen": 0.10065505653619766,
      "rewards/margins": 0.13729141652584076,
      "rewards/rejected": -0.03663638234138489,
      "step": 370
    },
    {
      "epoch": 0.99,
      "grad_norm": 556.0,
      "learning_rate": 2.575864278703266e-09,
      "logits/chosen": -2.906440258026123,
      "logits/rejected": -2.8900303840637207,
      "logps/chosen": -28.174755096435547,
      "logps/rejected": -28.421972274780273,
      "loss": 16.0238,
      "rewards/accuracies": 0.8500000238418579,
      "rewards/chosen": 0.07702798396348953,
      "rewards/margins": 0.13026344776153564,
      "rewards/rejected": -0.05323547124862671,
      "step": 380
    },
    {
      "epoch": 1.0,
      "step": 385,
      "total_flos": 0.0,
      "train_loss": 19.552858486423244,
      "train_runtime": 2721.3646,
      "train_samples_per_second": 1.131,
      "train_steps_per_second": 0.141
    }
  ],
  "logging_steps": 10,
  "max_steps": 385,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}