{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.996886351842242, "eval_steps": 240, "global_step": 962, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0020757654385054488, "grad_norm": 7.452356109448157, "learning_rate": 5.154639175257731e-09, "logits/chosen": -2.730942726135254, "logits/rejected": -2.654609203338623, "logps/chosen": -350.489990234375, "logps/rejected": -325.546875, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.02075765438505449, "grad_norm": 7.866099026608714, "learning_rate": 5.154639175257731e-08, "logits/chosen": -2.7327775955200195, "logits/rejected": -2.734964609146118, "logps/chosen": -366.4884033203125, "logps/rejected": -412.2764892578125, "loss": 0.6931, "rewards/accuracies": 0.4826388955116272, "rewards/chosen": 6.40038269921206e-05, "rewards/margins": 0.00040091515984386206, "rewards/rejected": -0.00033691132557578385, "step": 10 }, { "epoch": 0.04151530877010898, "grad_norm": 7.559166611883559, "learning_rate": 1.0309278350515462e-07, "logits/chosen": -2.717351198196411, "logits/rejected": -2.69411563873291, "logps/chosen": -378.9507141113281, "logps/rejected": -404.0965270996094, "loss": 0.6921, "rewards/accuracies": 0.59375, "rewards/chosen": 0.00041852169670164585, "rewards/margins": 0.0022167633287608624, "rewards/rejected": -0.0017982417484745383, "step": 20 }, { "epoch": 0.062272963155163466, "grad_norm": 7.615638482493831, "learning_rate": 1.5463917525773197e-07, "logits/chosen": -2.719243288040161, "logits/rejected": -2.7044143676757812, "logps/chosen": -365.67889404296875, "logps/rejected": -388.8330993652344, "loss": 0.6864, "rewards/accuracies": 0.793749988079071, "rewards/chosen": 0.004637600388377905, "rewards/margins": 0.013907420448958874, "rewards/rejected": -0.009269820526242256, "step": 30 }, { "epoch": 0.08303061754021795, "grad_norm": 7.7707957616247505, "learning_rate": 2.0618556701030925e-07, "logits/chosen": -2.7188639640808105, "logits/rejected": -2.6840219497680664, "logps/chosen": -350.5911865234375, "logps/rejected": -365.0216979980469, "loss": 0.6693, "rewards/accuracies": 0.871874988079071, "rewards/chosen": 0.027326706796884537, "rewards/margins": 0.049114055931568146, "rewards/rejected": -0.02178734540939331, "step": 40 }, { "epoch": 0.10378827192527244, "grad_norm": 8.90458099154434, "learning_rate": 2.5773195876288655e-07, "logits/chosen": -2.7294702529907227, "logits/rejected": -2.7243564128875732, "logps/chosen": -358.9928894042969, "logps/rejected": -397.3983459472656, "loss": 0.63, "rewards/accuracies": 0.862500011920929, "rewards/chosen": 0.07257182896137238, "rewards/margins": 0.14314065873622894, "rewards/rejected": -0.07056883722543716, "step": 50 }, { "epoch": 0.12454592631032693, "grad_norm": 9.98989696476554, "learning_rate": 3.0927835051546394e-07, "logits/chosen": -2.6951613426208496, "logits/rejected": -2.678609848022461, "logps/chosen": -348.3409729003906, "logps/rejected": -426.9418029785156, "loss": 0.5482, "rewards/accuracies": 0.878125011920929, "rewards/chosen": -0.011659199371933937, "rewards/margins": 0.36892449855804443, "rewards/rejected": -0.3805837035179138, "step": 60 }, { "epoch": 0.14530358069538143, "grad_norm": 14.516753369132225, "learning_rate": 3.608247422680412e-07, "logits/chosen": -2.7372078895568848, "logits/rejected": -2.6939454078674316, "logps/chosen": -451.07293701171875, "logps/rejected": -535.2464599609375, "loss": 0.4282, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -0.5447245836257935, "rewards/margins": 0.7660267353057861, "rewards/rejected": -1.3107513189315796, "step": 70 }, { "epoch": 0.1660612350804359, "grad_norm": 15.352248807358325, "learning_rate": 4.123711340206185e-07, "logits/chosen": -2.7028908729553223, "logits/rejected": -2.6827831268310547, "logps/chosen": -547.4126586914062, "logps/rejected": -697.9911499023438, "loss": 0.3596, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -1.5639097690582275, "rewards/margins": 1.3926727771759033, "rewards/rejected": -2.9565823078155518, "step": 80 }, { "epoch": 0.1868188894654904, "grad_norm": 17.157617837316515, "learning_rate": 4.639175257731959e-07, "logits/chosen": -2.664567708969116, "logits/rejected": -2.664062023162842, "logps/chosen": -561.7244262695312, "logps/rejected": -796.8342895507812, "loss": 0.3035, "rewards/accuracies": 0.878125011920929, "rewards/chosen": -1.9646422863006592, "rewards/margins": 2.149301052093506, "rewards/rejected": -4.113943576812744, "step": 90 }, { "epoch": 0.2075765438505449, "grad_norm": 15.724313068127296, "learning_rate": 4.999851606199919e-07, "logits/chosen": -2.607901096343994, "logits/rejected": -2.587402582168579, "logps/chosen": -576.0076904296875, "logps/rejected": -909.0657348632812, "loss": 0.263, "rewards/accuracies": 0.903124988079071, "rewards/chosen": -2.0050268173217773, "rewards/margins": 3.0768344402313232, "rewards/rejected": -5.0818610191345215, "step": 100 }, { "epoch": 0.2283341982355994, "grad_norm": 22.849604961849657, "learning_rate": 4.997213984244138e-07, "logits/chosen": -2.399500608444214, "logits/rejected": -2.2552199363708496, "logps/chosen": -620.6517333984375, "logps/rejected": -1053.1802978515625, "loss": 0.2333, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -2.491640329360962, "rewards/margins": 4.089051246643066, "rewards/rejected": -6.580691337585449, "step": 110 }, { "epoch": 0.24909185262065386, "grad_norm": 15.418275685296619, "learning_rate": 4.991282726678214e-07, "logits/chosen": -2.1379497051239014, "logits/rejected": -1.6612409353256226, "logps/chosen": -636.1170654296875, "logps/rejected": -1038.722412109375, "loss": 0.2114, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -2.4115161895751953, "rewards/margins": 3.842604875564575, "rewards/rejected": -6.254120826721191, "step": 120 }, { "epoch": 0.26984950700570837, "grad_norm": 22.122903405439125, "learning_rate": 4.982065656380468e-07, "logits/chosen": -1.8730089664459229, "logits/rejected": -1.1086432933807373, "logps/chosen": -624.3699951171875, "logps/rejected": -1083.5306396484375, "loss": 0.1978, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -2.4477767944335938, "rewards/margins": 4.317242622375488, "rewards/rejected": -6.76501989364624, "step": 130 }, { "epoch": 0.29060716139076287, "grad_norm": 13.724914658926766, "learning_rate": 4.969574929966689e-07, "logits/chosen": -1.063720941543579, "logits/rejected": 0.11310062557458878, "logps/chosen": -654.7889404296875, "logps/rejected": -1111.530517578125, "loss": 0.2171, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -2.8432438373565674, "rewards/margins": 4.339225769042969, "rewards/rejected": -7.182469844818115, "step": 140 }, { "epoch": 0.3113648157758173, "grad_norm": 20.111379920464472, "learning_rate": 4.953827021756488e-07, "logits/chosen": -1.147434949874878, "logits/rejected": 0.20881839096546173, "logps/chosen": -609.7511596679688, "logps/rejected": -1106.64990234375, "loss": 0.1719, "rewards/accuracies": 0.921875, "rewards/chosen": -2.4953980445861816, "rewards/margins": 4.772416591644287, "rewards/rejected": -7.267814636230469, "step": 150 }, { "epoch": 0.3321224701608718, "grad_norm": 20.712950851230556, "learning_rate": 4.93484270204492e-07, "logits/chosen": -0.3996439576148987, "logits/rejected": 0.9938270449638367, "logps/chosen": -663.2901000976562, "logps/rejected": -1266.5684814453125, "loss": 0.1789, "rewards/accuracies": 0.921875, "rewards/chosen": -2.7938601970672607, "rewards/margins": 5.7776288986206055, "rewards/rejected": -8.571489334106445, "step": 160 }, { "epoch": 0.3528801245459263, "grad_norm": 13.955815229008401, "learning_rate": 4.91264700970804e-07, "logits/chosen": -0.3857128620147705, "logits/rejected": 1.0256187915802002, "logps/chosen": -606.496337890625, "logps/rejected": -1150.380126953125, "loss": 0.1822, "rewards/accuracies": 0.9156249761581421, "rewards/chosen": -2.4765748977661133, "rewards/margins": 5.2469162940979, "rewards/rejected": -7.723490238189697, "step": 170 }, { "epoch": 0.3736377789309808, "grad_norm": 19.91766661692759, "learning_rate": 4.88726921917853e-07, "logits/chosen": -0.7232545614242554, "logits/rejected": 0.9214665293693542, "logps/chosen": -608.4518432617188, "logps/rejected": -1221.78955078125, "loss": 0.1832, "rewards/accuracies": 0.909375011920929, "rewards/chosen": -2.418121337890625, "rewards/margins": 5.937131881713867, "rewards/rejected": -8.355253219604492, "step": 180 }, { "epoch": 0.39439543331603527, "grad_norm": 16.024239064339824, "learning_rate": 4.858742801834942e-07, "logits/chosen": -0.32018381357192993, "logits/rejected": 1.0246493816375732, "logps/chosen": -667.6340942382812, "logps/rejected": -1306.54833984375, "loss": 0.1733, "rewards/accuracies": 0.934374988079071, "rewards/chosen": -2.8495688438415527, "rewards/margins": 6.024672985076904, "rewards/rejected": -8.874241828918457, "step": 190 }, { "epoch": 0.4151530877010898, "grad_norm": 12.349261100548174, "learning_rate": 4.827105381855496e-07, "logits/chosen": 0.25890278816223145, "logits/rejected": 1.507102370262146, "logps/chosen": -618.0540161132812, "logps/rejected": -1187.555419921875, "loss": 0.1662, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -2.504819393157959, "rewards/margins": 5.406851291656494, "rewards/rejected": -7.911670684814453, "step": 200 }, { "epoch": 0.4359107420861443, "grad_norm": 13.66810823452425, "learning_rate": 4.79239868659464e-07, "logits/chosen": -0.016678806394338608, "logits/rejected": 1.5381004810333252, "logps/chosen": -641.5460815429688, "logps/rejected": -1252.699951171875, "loss": 0.1584, "rewards/accuracies": 0.934374988079071, "rewards/chosen": -2.724513530731201, "rewards/margins": 5.9185709953308105, "rewards/rejected": -8.643084526062012, "step": 210 }, { "epoch": 0.4566683964711988, "grad_norm": 22.723399659159355, "learning_rate": 4.7546684915478443e-07, "logits/chosen": -0.5056034326553345, "logits/rejected": 1.3583004474639893, "logps/chosen": -627.7324829101562, "logps/rejected": -1324.6092529296875, "loss": 0.1696, "rewards/accuracies": 0.9156249761581421, "rewards/chosen": -2.6251590251922607, "rewards/margins": 6.788819789886475, "rewards/rejected": -9.413978576660156, "step": 220 }, { "epoch": 0.4774260508562532, "grad_norm": 25.512034717506292, "learning_rate": 4.7139645599771953e-07, "logits/chosen": -0.05699120834469795, "logits/rejected": 1.986132025718689, "logps/chosen": -660.3882446289062, "logps/rejected": -1403.30322265625, "loss": 0.1656, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -2.9933114051818848, "rewards/margins": 7.2729973793029785, "rewards/rejected": -10.266307830810547, "step": 230 }, { "epoch": 0.49818370524130773, "grad_norm": 19.374014963474004, "learning_rate": 4.6703405772774325e-07, "logits/chosen": -0.2160125970840454, "logits/rejected": 2.073151111602783, "logps/chosen": -610.6514892578125, "logps/rejected": -1350.219482421875, "loss": 0.1586, "rewards/accuracies": 0.9468749761581421, "rewards/chosen": -2.5249383449554443, "rewards/margins": 7.157525539398193, "rewards/rejected": -9.682464599609375, "step": 240 }, { "epoch": 0.49818370524130773, "eval_logits/chosen": -0.19751745462417603, "eval_logits/rejected": 1.597546935081482, "eval_logps/chosen": -692.0928344726562, "eval_logps/rejected": -1332.0889892578125, "eval_loss": 0.1689271330833435, "eval_rewards/accuracies": 0.9252451062202454, "eval_rewards/chosen": -2.6868460178375244, "eval_rewards/margins": 6.174641132354736, "eval_rewards/rejected": -8.861486434936523, "eval_runtime": 297.8773, "eval_samples_per_second": 21.791, "eval_steps_per_second": 0.342, "step": 240 }, { "epoch": 0.5189413596263622, "grad_norm": 12.036861238820864, "learning_rate": 4.6238540801689896e-07, "logits/chosen": -0.4214223325252533, "logits/rejected": 1.8403441905975342, "logps/chosen": -610.5186767578125, "logps/rejected": -1277.1844482421875, "loss": 0.1604, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -2.307227373123169, "rewards/margins": 6.4557318687438965, "rewards/rejected": -8.762959480285645, "step": 250 }, { "epoch": 0.5396990140114167, "grad_norm": 20.090786603415435, "learning_rate": 4.5745663808114316e-07, "logits/chosen": -0.743881106376648, "logits/rejected": 1.5020240545272827, "logps/chosen": -657.1211547851562, "logps/rejected": -1360.107666015625, "loss": 0.1638, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -2.839501142501831, "rewards/margins": 6.807010650634766, "rewards/rejected": -9.646512031555176, "step": 260 }, { "epoch": 0.5604566683964712, "grad_norm": 14.890524731187542, "learning_rate": 4.5225424859373684e-07, "logits/chosen": -0.38103678822517395, "logits/rejected": 1.8405994176864624, "logps/chosen": -600.785888671875, "logps/rejected": -1403.572998046875, "loss": 0.1493, "rewards/accuracies": 0.9468749761581421, "rewards/chosen": -2.4715404510498047, "rewards/margins": 7.713797569274902, "rewards/rejected": -10.185338020324707, "step": 270 }, { "epoch": 0.5812143227815257, "grad_norm": 15.737222130316104, "learning_rate": 4.467851011113515e-07, "logits/chosen": 0.3215886056423187, "logits/rejected": 2.3104662895202637, "logps/chosen": -629.4906005859375, "logps/rejected": -1383.751953125, "loss": 0.1571, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.5350632667541504, "rewards/margins": 7.314475059509277, "rewards/rejected": -9.849536895751953, "step": 280 }, { "epoch": 0.6019719771665801, "grad_norm": 23.328559288837457, "learning_rate": 4.410564090241966e-07, "logits/chosen": 0.4208109378814697, "logits/rejected": 2.2348339557647705, "logps/chosen": -675.2015991210938, "logps/rejected": -1364.4970703125, "loss": 0.1635, "rewards/accuracies": 0.9156249761581421, "rewards/chosen": -3.0044052600860596, "rewards/margins": 6.818602085113525, "rewards/rejected": -9.823007583618164, "step": 290 }, { "epoch": 0.6227296315516346, "grad_norm": 14.02162805247057, "learning_rate": 4.35075728042106e-07, "logits/chosen": -0.0718420147895813, "logits/rejected": 1.6788543462753296, "logps/chosen": -612.244873046875, "logps/rejected": -1253.4449462890625, "loss": 0.1528, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -2.3619751930236816, "rewards/margins": 6.16156530380249, "rewards/rejected": -8.523540496826172, "step": 300 }, { "epoch": 0.6434872859366891, "grad_norm": 13.851968293163813, "learning_rate": 4.2885094622913016e-07, "logits/chosen": -0.06367097049951553, "logits/rejected": 1.8669401407241821, "logps/chosen": -630.0944213867188, "logps/rejected": -1326.2972412109375, "loss": 0.1541, "rewards/accuracies": 0.90625, "rewards/chosen": -2.6027626991271973, "rewards/margins": 6.887024879455566, "rewards/rejected": -9.489786148071289, "step": 310 }, { "epoch": 0.6642449403217436, "grad_norm": 22.056173099963235, "learning_rate": 4.223902735997788e-07, "logits/chosen": -0.3920370638370514, "logits/rejected": 1.7357165813446045, "logps/chosen": -582.3004150390625, "logps/rejected": -1314.4351806640625, "loss": 0.1538, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -2.171776294708252, "rewards/margins": 7.103475093841553, "rewards/rejected": -9.275251388549805, "step": 320 }, { "epoch": 0.6850025947067981, "grad_norm": 10.33520679821666, "learning_rate": 4.157022312906352e-07, "logits/chosen": -1.2140147686004639, "logits/rejected": 1.2578856945037842, "logps/chosen": -645.0560302734375, "logps/rejected": -1458.661376953125, "loss": 0.1492, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -2.7255163192749023, "rewards/margins": 7.840676784515381, "rewards/rejected": -10.566194534301758, "step": 330 }, { "epoch": 0.7057602490918526, "grad_norm": 12.698623571384912, "learning_rate": 4.0879564032162425e-07, "logits/chosen": -1.3428138494491577, "logits/rejected": 1.8070862293243408, "logps/chosen": -719.0316162109375, "logps/rejected": -1663.4603271484375, "loss": 0.143, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -3.4557037353515625, "rewards/margins": 9.146575927734375, "rewards/rejected": -12.602280616760254, "step": 340 }, { "epoch": 0.7265179034769071, "grad_norm": 16.68939070839139, "learning_rate": 4.016796099617569e-07, "logits/chosen": -1.3726019859313965, "logits/rejected": 1.8845252990722656, "logps/chosen": -672.85693359375, "logps/rejected": -1494.5582275390625, "loss": 0.1538, "rewards/accuracies": 0.893750011920929, "rewards/chosen": -3.1178722381591797, "rewards/margins": 7.914468288421631, "rewards/rejected": -11.032341003417969, "step": 350 }, { "epoch": 0.7472755578619616, "grad_norm": 14.783186894273188, "learning_rate": 3.9436352571469577e-07, "logits/chosen": -0.5545839071273804, "logits/rejected": 3.833822727203369, "logps/chosen": -830.2039184570312, "logps/rejected": -1744.631103515625, "loss": 0.1444, "rewards/accuracies": 0.934374988079071, "rewards/chosen": -4.657293796539307, "rewards/margins": 9.05556583404541, "rewards/rejected": -13.712858200073242, "step": 360 }, { "epoch": 0.768033212247016, "grad_norm": 15.755226269776646, "learning_rate": 3.868570369399893e-07, "logits/chosen": -0.6274186372756958, "logits/rejected": 3.4528141021728516, "logps/chosen": -799.2054443359375, "logps/rejected": -1723.1851806640625, "loss": 0.1458, "rewards/accuracies": 0.921875, "rewards/chosen": -4.323949337005615, "rewards/margins": 9.110595703125, "rewards/rejected": -13.434545516967773, "step": 370 }, { "epoch": 0.7887908666320705, "grad_norm": 11.452118871998977, "learning_rate": 3.791700441262987e-07, "logits/chosen": 0.8511954545974731, "logits/rejected": 4.691292762756348, "logps/chosen": -939.5437622070312, "logps/rejected": -2012.4154052734375, "loss": 0.1417, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -5.871208667755127, "rewards/margins": 10.141061782836914, "rewards/rejected": -16.012271881103516, "step": 380 }, { "epoch": 0.809548521017125, "grad_norm": 13.459005174890807, "learning_rate": 3.7131268583340515e-07, "logits/chosen": 0.3982798457145691, "logits/rejected": 4.274272441864014, "logps/chosen": -845.7362060546875, "logps/rejected": -1642.904541015625, "loss": 0.1523, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -4.7007269859313965, "rewards/margins": 7.76672887802124, "rewards/rejected": -12.467455863952637, "step": 390 }, { "epoch": 0.8303061754021795, "grad_norm": 13.869817366585716, "learning_rate": 3.632953253202198e-07, "logits/chosen": 1.5343601703643799, "logits/rejected": 4.5011491775512695, "logps/chosen": -881.2744140625, "logps/rejected": -1823.2939453125, "loss": 0.1449, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -5.160914421081543, "rewards/margins": 9.128256797790527, "rewards/rejected": -14.289172172546387, "step": 400 }, { "epoch": 0.851063829787234, "grad_norm": 14.502067608546383, "learning_rate": 3.551285368764321e-07, "logits/chosen": 1.5195695161819458, "logits/rejected": 4.168100357055664, "logps/chosen": -846.6906127929688, "logps/rejected": -1635.856689453125, "loss": 0.1445, "rewards/accuracies": 0.918749988079071, "rewards/chosen": -4.812326431274414, "rewards/margins": 7.744412422180176, "rewards/rejected": -12.556737899780273, "step": 410 }, { "epoch": 0.8718214841722886, "grad_norm": 30.730295160965017, "learning_rate": 3.468230918758242e-07, "logits/chosen": -0.11532606929540634, "logits/rejected": 3.6724467277526855, "logps/chosen": -856.9993286132812, "logps/rejected": -1789.368408203125, "loss": 0.1521, "rewards/accuracies": 0.9281250238418579, "rewards/chosen": -4.82938289642334, "rewards/margins": 9.0027494430542, "rewards/rejected": -13.832133293151855, "step": 420 }, { "epoch": 0.892579138557343, "grad_norm": 15.361401417475852, "learning_rate": 3.383899445696477e-07, "logits/chosen": -0.19606542587280273, "logits/rejected": 3.2190608978271484, "logps/chosen": -914.5608520507812, "logps/rejected": -1852.5074462890625, "loss": 0.1317, "rewards/accuracies": 0.9281250238418579, "rewards/chosen": -5.41568660736084, "rewards/margins": 9.153519630432129, "rewards/rejected": -14.569204330444336, "step": 430 }, { "epoch": 0.9133367929423976, "grad_norm": 15.988866292668206, "learning_rate": 3.2984021763879756e-07, "logits/chosen": 0.24461348354816437, "logits/rejected": 3.7112534046173096, "logps/chosen": -1049.1207275390625, "logps/rejected": -2157.442138671875, "loss": 0.1472, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -6.739071846008301, "rewards/margins": 10.936814308166504, "rewards/rejected": -17.675886154174805, "step": 440 }, { "epoch": 0.934094447327452, "grad_norm": 16.98533210114085, "learning_rate": 3.211851875238408e-07, "logits/chosen": 0.16562503576278687, "logits/rejected": 4.111753940582275, "logps/chosen": -1058.0418701171875, "logps/rejected": -2127.619873046875, "loss": 0.1444, "rewards/accuracies": 0.9375, "rewards/chosen": -6.909078121185303, "rewards/margins": 10.602731704711914, "rewards/rejected": -17.511810302734375, "step": 450 }, { "epoch": 0.9548521017125065, "grad_norm": 14.967379090940785, "learning_rate": 3.124362695522476e-07, "logits/chosen": 1.7813737392425537, "logits/rejected": 4.8395795822143555, "logps/chosen": -1155.03125, "logps/rejected": -2117.0537109375, "loss": 0.1396, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -7.852016448974609, "rewards/margins": 9.481895446777344, "rewards/rejected": -17.333911895751953, "step": 460 }, { "epoch": 0.975609756097561, "grad_norm": 14.307026095079875, "learning_rate": 3.036050028824415e-07, "logits/chosen": 1.5212550163269043, "logits/rejected": 4.2941412925720215, "logps/chosen": -1053.684814453125, "logps/rejected": -1923.693359375, "loss": 0.125, "rewards/accuracies": 0.9281250238418579, "rewards/chosen": -6.825259208679199, "rewards/margins": 8.460533142089844, "rewards/rejected": -15.285791397094727, "step": 470 }, { "epoch": 0.9963674104826155, "grad_norm": 11.916307943626432, "learning_rate": 2.9470303528452547e-07, "logits/chosen": 1.5992648601531982, "logits/rejected": 4.383261680603027, "logps/chosen": -1080.980224609375, "logps/rejected": -2100.404541015625, "loss": 0.1408, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -7.290544033050537, "rewards/margins": 9.89597225189209, "rewards/rejected": -17.1865177154541, "step": 480 }, { "epoch": 0.9963674104826155, "eval_logits/chosen": 0.5014842748641968, "eval_logits/rejected": 3.342353582382202, "eval_logps/chosen": -1083.2030029296875, "eval_logps/rejected": -2038.5933837890625, "eval_loss": 0.14824749529361725, "eval_rewards/accuracies": 0.9325980544090271, "eval_rewards/chosen": -6.597947597503662, "eval_rewards/margins": 9.328583717346191, "eval_rewards/rejected": -15.926531791687012, "eval_runtime": 296.0381, "eval_samples_per_second": 21.926, "eval_steps_per_second": 0.345, "step": 480 }, { "epoch": 1.01712506486767, "grad_norm": 14.456304387791999, "learning_rate": 2.8574210777775755e-07, "logits/chosen": 0.9456971287727356, "logits/rejected": 4.39624547958374, "logps/chosen": -1059.8154296875, "logps/rejected": -2144.65576171875, "loss": 0.0951, "rewards/accuracies": 0.953125, "rewards/chosen": -6.945745944976807, "rewards/margins": 10.675562858581543, "rewards/rejected": -17.621309280395508, "step": 490 }, { "epoch": 1.0378827192527245, "grad_norm": 17.681966171705884, "learning_rate": 2.767340391450384e-07, "logits/chosen": 1.6467583179473877, "logits/rejected": 5.070549011230469, "logps/chosen": -1184.4014892578125, "logps/rejected": -2638.23291015625, "loss": 0.0919, "rewards/accuracies": 0.9468749761581421, "rewards/chosen": -8.20818042755127, "rewards/margins": 14.303576469421387, "rewards/rejected": -22.511754989624023, "step": 500 }, { "epoch": 1.058640373637779, "grad_norm": 13.802679882016555, "learning_rate": 2.6769071034483407e-07, "logits/chosen": 1.1824378967285156, "logits/rejected": 4.564583778381348, "logps/chosen": -947.0963745117188, "logps/rejected": -2079.13916015625, "loss": 0.0879, "rewards/accuracies": 0.9781249761581421, "rewards/chosen": -5.809744358062744, "rewards/margins": 11.030488967895508, "rewards/rejected": -16.840232849121094, "step": 510 }, { "epoch": 1.0793980280228335, "grad_norm": 22.098742924019817, "learning_rate": 2.5862404884109365e-07, "logits/chosen": 1.4323368072509766, "logits/rejected": 4.83809757232666, "logps/chosen": -1047.9368896484375, "logps/rejected": -2401.321533203125, "loss": 0.0826, "rewards/accuracies": 0.965624988079071, "rewards/chosen": -6.900972843170166, "rewards/margins": 13.123873710632324, "rewards/rejected": -20.02484703063965, "step": 520 }, { "epoch": 1.100155682407888, "grad_norm": 14.888695199480608, "learning_rate": 2.495460128718305e-07, "logits/chosen": 0.8169673681259155, "logits/rejected": 4.492055416107178, "logps/chosen": -1088.3392333984375, "logps/rejected": -2334.029052734375, "loss": 0.0775, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -7.015921592712402, "rewards/margins": 12.27374267578125, "rewards/rejected": -19.289663314819336, "step": 530 }, { "epoch": 1.1209133367929425, "grad_norm": 13.711286256989611, "learning_rate": 2.404685756771143e-07, "logits/chosen": 0.3249141275882721, "logits/rejected": 4.28688383102417, "logps/chosen": -1050.6568603515625, "logps/rejected": -2440.43212890625, "loss": 0.0831, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -6.925228118896484, "rewards/margins": 13.674964904785156, "rewards/rejected": -20.60019302368164, "step": 540 }, { "epoch": 1.141670991177997, "grad_norm": 12.310015587533016, "learning_rate": 2.314037097072764e-07, "logits/chosen": 0.325296014547348, "logits/rejected": 4.037835121154785, "logps/chosen": -1070.4940185546875, "logps/rejected": -2429.653076171875, "loss": 0.0774, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.072981357574463, "rewards/margins": 13.413686752319336, "rewards/rejected": -20.48666763305664, "step": 550 }, { "epoch": 1.1624286455630513, "grad_norm": 18.024933264685323, "learning_rate": 2.2236337083215723e-07, "logits/chosen": 0.973385214805603, "logits/rejected": 4.8588714599609375, "logps/chosen": -1165.290283203125, "logps/rejected": -2533.037841796875, "loss": 0.0806, "rewards/accuracies": 0.965624988079071, "rewards/chosen": -7.950223445892334, "rewards/margins": 13.607198715209961, "rewards/rejected": -21.557422637939453, "step": 560 }, { "epoch": 1.183186299948106, "grad_norm": 39.199036754410095, "learning_rate": 2.13359482572222e-07, "logits/chosen": 0.9793283343315125, "logits/rejected": 4.486080169677734, "logps/chosen": -1073.2510986328125, "logps/rejected": -2294.072509765625, "loss": 0.0855, "rewards/accuracies": 0.9468749761581421, "rewards/chosen": -7.108448028564453, "rewards/margins": 11.980100631713867, "rewards/rejected": -19.088550567626953, "step": 570 }, { "epoch": 1.2039439543331603, "grad_norm": 13.491261159834455, "learning_rate": 2.044039203723423e-07, "logits/chosen": 0.23347489535808563, "logits/rejected": 3.7263119220733643, "logps/chosen": -997.2879028320312, "logps/rejected": -2139.56298828125, "loss": 0.0886, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -6.343294143676758, "rewards/margins": 11.091756820678711, "rewards/rejected": -17.435049057006836, "step": 580 }, { "epoch": 1.2247016087182148, "grad_norm": 22.750928538479595, "learning_rate": 1.955084959389864e-07, "logits/chosen": 0.4886883795261383, "logits/rejected": 4.1630377769470215, "logps/chosen": -1106.54052734375, "logps/rejected": -2491.51416015625, "loss": 0.0749, "rewards/accuracies": 0.9781249761581421, "rewards/chosen": -7.559281826019287, "rewards/margins": 13.47680377960205, "rewards/rejected": -21.036083221435547, "step": 590 }, { "epoch": 1.2454592631032693, "grad_norm": 20.593372405978258, "learning_rate": 1.866849416614753e-07, "logits/chosen": 0.24794098734855652, "logits/rejected": 4.261725425720215, "logps/chosen": -1190.895751953125, "logps/rejected": -2684.16943359375, "loss": 0.0762, "rewards/accuracies": 0.9781249761581421, "rewards/chosen": -8.191116333007812, "rewards/margins": 14.663787841796875, "rewards/rejected": -22.854902267456055, "step": 600 }, { "epoch": 1.2662169174883238, "grad_norm": 23.08708475257412, "learning_rate": 1.7794489513785227e-07, "logits/chosen": -0.35908642411231995, "logits/rejected": 3.719221830368042, "logps/chosen": -1067.587158203125, "logps/rejected": -2360.556396484375, "loss": 0.0824, "rewards/accuracies": 0.96875, "rewards/chosen": -6.811041355133057, "rewards/margins": 12.751296997070312, "rewards/rejected": -19.562339782714844, "step": 610 }, { "epoch": 1.2869745718733783, "grad_norm": 11.158342147450194, "learning_rate": 1.692998838257744e-07, "logits/chosen": -0.18085989356040955, "logits/rejected": 3.5346386432647705, "logps/chosen": -1038.219970703125, "logps/rejected": -2255.41064453125, "loss": 0.0769, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -6.670246124267578, "rewards/margins": 11.845807075500488, "rewards/rejected": -18.51605224609375, "step": 620 }, { "epoch": 1.3077322262584328, "grad_norm": 14.33248304938314, "learning_rate": 1.6076130983867191e-07, "logits/chosen": -0.1000831350684166, "logits/rejected": 4.0322370529174805, "logps/chosen": -1037.5277099609375, "logps/rejected": -2512.465087890625, "loss": 0.0774, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -6.664424896240234, "rewards/margins": 14.430699348449707, "rewards/rejected": -21.095125198364258, "step": 630 }, { "epoch": 1.3284898806434873, "grad_norm": 18.5366504084505, "learning_rate": 1.5234043490722587e-07, "logits/chosen": -0.20596277713775635, "logits/rejected": 3.613173246383667, "logps/chosen": -1079.144287109375, "logps/rejected": -2312.60791015625, "loss": 0.0821, "rewards/accuracies": 0.965624988079071, "rewards/chosen": -7.167913913726807, "rewards/margins": 12.191060066223145, "rewards/rejected": -19.35897445678711, "step": 640 }, { "epoch": 1.3492475350285418, "grad_norm": 18.228054814234156, "learning_rate": 1.44048365526001e-07, "logits/chosen": 0.0638003945350647, "logits/rejected": 3.9387619495391846, "logps/chosen": -1077.5164794921875, "logps/rejected": -2451.49462890625, "loss": 0.0867, "rewards/accuracies": 0.96875, "rewards/chosen": -7.139754295349121, "rewards/margins": 13.3494873046875, "rewards/rejected": -20.489240646362305, "step": 650 }, { "epoch": 1.3700051894135963, "grad_norm": 16.31608491567421, "learning_rate": 1.3589603830482243e-07, "logits/chosen": -0.08248420059680939, "logits/rejected": 4.198277950286865, "logps/chosen": -1142.9498291015625, "logps/rejected": -2611.378662109375, "loss": 0.0703, "rewards/accuracies": 0.96875, "rewards/chosen": -7.643826484680176, "rewards/margins": 14.516764640808105, "rewards/rejected": -22.160587310791016, "step": 660 }, { "epoch": 1.3907628437986508, "grad_norm": 21.850286306048936, "learning_rate": 1.2789420554421821e-07, "logits/chosen": -0.33774855732917786, "logits/rejected": 3.5582516193389893, "logps/chosen": -1178.3541259765625, "logps/rejected": -2523.93701171875, "loss": 0.0845, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -8.179405212402344, "rewards/margins": 13.182377815246582, "rewards/rejected": -21.36178207397461, "step": 670 }, { "epoch": 1.4115204981837053, "grad_norm": 15.996775283562163, "learning_rate": 1.200534210539509e-07, "logits/chosen": -0.7213211059570312, "logits/rejected": 3.0649337768554688, "logps/chosen": -1151.722412109375, "logps/rejected": -2552.525634765625, "loss": 0.0892, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.825603485107422, "rewards/margins": 13.686927795410156, "rewards/rejected": -21.512531280517578, "step": 680 }, { "epoch": 1.4322781525687598, "grad_norm": 16.75949934654318, "learning_rate": 1.1238402623334492e-07, "logits/chosen": -0.9597622752189636, "logits/rejected": 2.867584228515625, "logps/chosen": -1066.507568359375, "logps/rejected": -2423.34423828125, "loss": 0.0898, "rewards/accuracies": 0.953125, "rewards/chosen": -6.92177677154541, "rewards/margins": 13.205133438110352, "rewards/rejected": -20.126911163330078, "step": 690 }, { "epoch": 1.4530358069538143, "grad_norm": 11.338012307523448, "learning_rate": 1.0489613643176479e-07, "logits/chosen": -0.6678389310836792, "logits/rejected": 3.2197043895721436, "logps/chosen": -1113.594482421875, "logps/rejected": -2410.258544921875, "loss": 0.0722, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -7.322343349456787, "rewards/margins": 12.825822830200195, "rewards/rejected": -20.14816665649414, "step": 700 }, { "epoch": 1.4737934613388686, "grad_norm": 15.856495888079317, "learning_rate": 9.759962760723855e-08, "logits/chosen": -0.16049222648143768, "logits/rejected": 4.048083305358887, "logps/chosen": -1173.0849609375, "logps/rejected": -2468.91845703125, "loss": 0.0662, "rewards/accuracies": 0.96875, "rewards/chosen": -8.210199356079102, "rewards/margins": 12.80036735534668, "rewards/rejected": -21.01056671142578, "step": 710 }, { "epoch": 1.4945511157239233, "grad_norm": 31.07426456754591, "learning_rate": 9.050412330081883e-08, "logits/chosen": -0.20700784027576447, "logits/rejected": 3.9535317420959473, "logps/chosen": -1277.49755859375, "logps/rejected": -2737.533447265625, "loss": 0.0852, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -9.044589042663574, "rewards/margins": 14.331390380859375, "rewards/rejected": -23.375978469848633, "step": 720 }, { "epoch": 1.4945511157239233, "eval_logits/chosen": 0.06359091401100159, "eval_logits/rejected": 3.48725962638855, "eval_logps/chosen": -1424.8193359375, "eval_logps/rejected": -2816.588623046875, "eval_loss": 0.16444814205169678, "eval_rewards/accuracies": 0.9227941036224365, "eval_rewards/chosen": -10.01410961151123, "eval_rewards/margins": 13.69237232208252, "eval_rewards/rejected": -23.706483840942383, "eval_runtime": 297.3735, "eval_samples_per_second": 21.828, "eval_steps_per_second": 0.343, "step": 720 }, { "epoch": 1.5153087701089776, "grad_norm": 21.935108549149014, "learning_rate": 8.36189819438625e-08, "logits/chosen": -0.2683785557746887, "logits/rejected": 3.702204465866089, "logps/chosen": -1242.103759765625, "logps/rejected": -2603.80615234375, "loss": 0.0822, "rewards/accuracies": 0.965624988079071, "rewards/chosen": -8.634117126464844, "rewards/margins": 13.532336235046387, "rewards/rejected": -22.166454315185547, "step": 730 }, { "epoch": 1.5360664244940323, "grad_norm": 23.902586381058878, "learning_rate": 7.69532845149711e-08, "logits/chosen": -0.5570476055145264, "logits/rejected": 3.649244785308838, "logps/chosen": -1223.75439453125, "logps/rejected": -2637.007080078125, "loss": 0.0769, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -8.374210357666016, "rewards/margins": 13.947656631469727, "rewards/rejected": -22.321866989135742, "step": 740 }, { "epoch": 1.5568240788790866, "grad_norm": 20.338315184147856, "learning_rate": 7.051582256286929e-08, "logits/chosen": -0.5207056999206543, "logits/rejected": 3.2942442893981934, "logps/chosen": -1216.895263671875, "logps/rejected": -2651.10107421875, "loss": 0.0706, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -8.260024070739746, "rewards/margins": 14.131828308105469, "rewards/rejected": -22.39185333251953, "step": 750 }, { "epoch": 1.5775817332641413, "grad_norm": 19.826108952738736, "learning_rate": 6.431508661101954e-08, "logits/chosen": -0.35470911860466003, "logits/rejected": 3.6188712120056152, "logps/chosen": -1201.515625, "logps/rejected": -2452.108642578125, "loss": 0.0906, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -8.28996467590332, "rewards/margins": 12.43807315826416, "rewards/rejected": -20.728038787841797, "step": 760 }, { "epoch": 1.5983393876491956, "grad_norm": 18.731154290431807, "learning_rate": 5.8359254959266826e-08, "logits/chosen": -0.3598397374153137, "logits/rejected": 3.461977481842041, "logps/chosen": -1123.501220703125, "logps/rejected": -2401.59423828125, "loss": 0.0757, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -7.732508659362793, "rewards/margins": 12.514566421508789, "rewards/rejected": -20.247074127197266, "step": 770 }, { "epoch": 1.61909704203425, "grad_norm": 14.021094730035166, "learning_rate": 5.265618289728199e-08, "logits/chosen": -0.2899594306945801, "logits/rejected": 3.6835436820983887, "logps/chosen": -1112.9346923828125, "logps/rejected": -2395.99560546875, "loss": 0.0748, "rewards/accuracies": 0.965624988079071, "rewards/chosen": -7.564324378967285, "rewards/margins": 12.517146110534668, "rewards/rejected": -20.081470489501953, "step": 780 }, { "epoch": 1.6398546964193046, "grad_norm": 15.755848122162748, "learning_rate": 4.721339234403121e-08, "logits/chosen": -0.4539187550544739, "logits/rejected": 3.4312031269073486, "logps/chosen": -1102.984375, "logps/rejected": -2424.76904296875, "loss": 0.0889, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -7.416151523590088, "rewards/margins": 12.970865249633789, "rewards/rejected": -20.387014389038086, "step": 790 }, { "epoch": 1.660612350804359, "grad_norm": 10.116183369840204, "learning_rate": 4.203806192693587e-08, "logits/chosen": -0.16934530436992645, "logits/rejected": 3.3178775310516357, "logps/chosen": -1117.904541015625, "logps/rejected": -2234.350341796875, "loss": 0.0806, "rewards/accuracies": 0.971875011920929, "rewards/chosen": -7.489673614501953, "rewards/margins": 10.942729949951172, "rewards/rejected": -18.432403564453125, "step": 800 }, { "epoch": 1.6813700051894136, "grad_norm": 12.314909116730863, "learning_rate": 3.7137017513808544e-08, "logits/chosen": -0.2919425368309021, "logits/rejected": 3.503552198410034, "logps/chosen": -1133.6949462890625, "logps/rejected": -2370.295166015625, "loss": 0.0813, "rewards/accuracies": 0.9593750238418579, "rewards/chosen": -7.621180057525635, "rewards/margins": 12.205830574035645, "rewards/rejected": -19.827011108398438, "step": 810 }, { "epoch": 1.702127659574468, "grad_norm": 13.100920531092996, "learning_rate": 3.251672321005147e-08, "logits/chosen": -0.3816925585269928, "logits/rejected": 3.223738193511963, "logps/chosen": -1128.720458984375, "logps/rejected": -2432.7451171875, "loss": 0.0799, "rewards/accuracies": 0.96875, "rewards/chosen": -7.558645725250244, "rewards/margins": 12.71537971496582, "rewards/rejected": -20.274024963378906, "step": 820 }, { "epoch": 1.7228853139595226, "grad_norm": 13.542985324247502, "learning_rate": 2.8183272832992267e-08, "logits/chosen": -0.4070394039154053, "logits/rejected": 3.100729465484619, "logps/chosen": -1098.7197265625, "logps/rejected": -2404.990234375, "loss": 0.0769, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -7.4235520362854, "rewards/margins": 12.707655906677246, "rewards/rejected": -20.131206512451172, "step": 830 }, { "epoch": 1.743642968344577, "grad_norm": 19.414761044152183, "learning_rate": 2.414238187460191e-08, "logits/chosen": -0.3680698275566101, "logits/rejected": 3.6575589179992676, "logps/chosen": -1149.003662109375, "logps/rejected": -2435.11474609375, "loss": 0.0801, "rewards/accuracies": 0.971875011920929, "rewards/chosen": -7.7628068923950195, "rewards/margins": 12.7561674118042, "rewards/rejected": -20.51897621154785, "step": 840 }, { "epoch": 1.7644006227296316, "grad_norm": 18.249543754589453, "learning_rate": 2.0399379963194713e-08, "logits/chosen": -0.15194618701934814, "logits/rejected": 3.823453903198242, "logps/chosen": -1178.7374267578125, "logps/rejected": -2647.753173828125, "loss": 0.0742, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -8.07176685333252, "rewards/margins": 14.458259582519531, "rewards/rejected": -22.530025482177734, "step": 850 }, { "epoch": 1.7851582771146859, "grad_norm": 13.687621613308757, "learning_rate": 1.695920383405322e-08, "logits/chosen": -0.1904297024011612, "logits/rejected": 3.487968921661377, "logps/chosen": -1222.4801025390625, "logps/rejected": -2662.45654296875, "loss": 0.0862, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.289772033691406, "rewards/margins": 14.194554328918457, "rewards/rejected": -22.484325408935547, "step": 860 }, { "epoch": 1.8059159314997406, "grad_norm": 20.833924751057065, "learning_rate": 1.3826390818249434e-08, "logits/chosen": -0.10482398420572281, "logits/rejected": 3.5555343627929688, "logps/chosen": -1184.642822265625, "logps/rejected": -2578.11376953125, "loss": 0.0847, "rewards/accuracies": 0.965624988079071, "rewards/chosen": -8.188472747802734, "rewards/margins": 13.555729866027832, "rewards/rejected": -21.74420166015625, "step": 870 }, { "epoch": 1.826673585884795, "grad_norm": 12.098887002832749, "learning_rate": 1.1005072858249614e-08, "logits/chosen": -0.21788470447063446, "logits/rejected": 3.596818447113037, "logps/chosen": -1198.203125, "logps/rejected": -2579.16455078125, "loss": 0.0742, "rewards/accuracies": 0.9781249761581421, "rewards/chosen": -8.193432807922363, "rewards/margins": 13.623662948608398, "rewards/rejected": -21.817094802856445, "step": 880 }, { "epoch": 1.8474312402698496, "grad_norm": 22.649648963902354, "learning_rate": 8.498971058195886e-09, "logits/chosen": -0.2010045051574707, "logits/rejected": 3.6617989540100098, "logps/chosen": -1205.736328125, "logps/rejected": -2611.108642578125, "loss": 0.0685, "rewards/accuracies": 0.9781249761581421, "rewards/chosen": -8.295499801635742, "rewards/margins": 13.880526542663574, "rewards/rejected": -22.176025390625, "step": 890 }, { "epoch": 1.868188894654904, "grad_norm": 14.001123095428785, "learning_rate": 6.311390776052527e-09, "logits/chosen": -0.26746782660484314, "logits/rejected": 3.480095386505127, "logps/chosen": -1162.368408203125, "logps/rejected": -2528.66650390625, "loss": 0.0735, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -7.984139442443848, "rewards/margins": 13.350987434387207, "rewards/rejected": -21.335126876831055, "step": 900 }, { "epoch": 1.8889465490399586, "grad_norm": 17.75181453766059, "learning_rate": 4.445217264089751e-09, "logits/chosen": -0.3570159077644348, "logits/rejected": 3.7746682167053223, "logps/chosen": -1153.1602783203125, "logps/rejected": -2572.643798828125, "loss": 0.0662, "rewards/accuracies": 0.96875, "rewards/chosen": -7.910887718200684, "rewards/margins": 13.993339538574219, "rewards/rejected": -21.904226303100586, "step": 910 }, { "epoch": 1.909704203425013, "grad_norm": 19.779224657698173, "learning_rate": 2.902911863455121e-09, "logits/chosen": -0.23463687300682068, "logits/rejected": 3.8247523307800293, "logps/chosen": -1181.158447265625, "logps/rejected": -2677.92041015625, "loss": 0.0761, "rewards/accuracies": 0.9781249761581421, "rewards/chosen": -8.19698429107666, "rewards/margins": 14.724853515625, "rewards/rejected": -22.92184066772461, "step": 920 }, { "epoch": 1.9304618578100676, "grad_norm": 22.12956855667942, "learning_rate": 1.686508757851507e-09, "logits/chosen": -0.2650212049484253, "logits/rejected": 3.5063083171844482, "logps/chosen": -1194.435791015625, "logps/rejected": -2615.01708984375, "loss": 0.0854, "rewards/accuracies": 0.956250011920929, "rewards/chosen": -8.247761726379395, "rewards/margins": 13.95422077178955, "rewards/rejected": -22.201980590820312, "step": 930 }, { "epoch": 1.951219512195122, "grad_norm": 20.722879655487088, "learning_rate": 7.976122906031557e-10, "logits/chosen": -0.23710790276527405, "logits/rejected": 3.5924439430236816, "logps/chosen": -1181.043212890625, "logps/rejected": -2523.64794921875, "loss": 0.088, "rewards/accuracies": 0.9437500238418579, "rewards/chosen": -8.123555183410645, "rewards/margins": 13.19268798828125, "rewards/rejected": -21.31624412536621, "step": 940 }, { "epoch": 1.9719771665801764, "grad_norm": 12.897702624327897, "learning_rate": 2.37394848648792e-10, "logits/chosen": -0.3340440094470978, "logits/rejected": 3.253843307495117, "logps/chosen": -1149.8099365234375, "logps/rejected": -2490.43994140625, "loss": 0.0801, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -7.880959510803223, "rewards/margins": 13.208969116210938, "rewards/rejected": -21.089927673339844, "step": 950 }, { "epoch": 1.992734820965231, "grad_norm": 17.31058461781814, "learning_rate": 6.5953162521614755e-12, "logits/chosen": -0.3153567910194397, "logits/rejected": 3.3712565898895264, "logps/chosen": -1199.875244140625, "logps/rejected": -2640.38525390625, "loss": 0.0743, "rewards/accuracies": 0.981249988079071, "rewards/chosen": -8.392365455627441, "rewards/margins": 14.206585884094238, "rewards/rejected": -22.598949432373047, "step": 960 }, { "epoch": 1.992734820965231, "eval_logits/chosen": -0.1724339723587036, "eval_logits/rejected": 3.016493320465088, "eval_logps/chosen": -1310.28955078125, "eval_logps/rejected": -2616.585205078125, "eval_loss": 0.15784408152103424, "eval_rewards/accuracies": 0.9203431606292725, "eval_rewards/chosen": -8.868812561035156, "eval_rewards/margins": 12.837637901306152, "eval_rewards/rejected": -21.706451416015625, "eval_runtime": 296.2637, "eval_samples_per_second": 21.91, "eval_steps_per_second": 0.344, "step": 960 }, { "epoch": 1.996886351842242, "step": 962, "total_flos": 0.0, "train_loss": 0.1585804910288655, "train_runtime": 28225.2793, "train_samples_per_second": 8.737, "train_steps_per_second": 0.034 } ], "logging_steps": 10, "max_steps": 962, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 240, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }