{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 625, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 7.936507936507937e-08, "logits/chosen": 0.24662643671035767, "logits/rejected": 0.3199857771396637, "logps/chosen": -398.9810791015625, "logps/rejected": -445.0414123535156, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.02, "learning_rate": 7.936507936507937e-07, "logits/chosen": 0.2540830075740814, "logits/rejected": 0.28116142749786377, "logps/chosen": -492.2332763671875, "logps/rejected": -497.7977294921875, "loss": 0.6931, "rewards/accuracies": 0.3958333432674408, "rewards/chosen": -0.0001207808600156568, "rewards/margins": -3.325760189909488e-05, "rewards/rejected": -8.752326539251953e-05, "step": 10 }, { "epoch": 0.03, "learning_rate": 1.5873015873015873e-06, "logits/chosen": 0.19658036530017853, "logits/rejected": 0.304654985666275, "logps/chosen": -462.8489685058594, "logps/rejected": -485.5413513183594, "loss": 0.693, "rewards/accuracies": 0.4375, "rewards/chosen": -0.001320437528192997, "rewards/margins": 0.0002517173998057842, "rewards/rejected": -0.001572155044414103, "step": 20 }, { "epoch": 0.05, "learning_rate": 2.380952380952381e-06, "logits/chosen": 0.233014315366745, "logits/rejected": 0.2545226514339447, "logps/chosen": -516.826171875, "logps/rejected": -515.098876953125, "loss": 0.6921, "rewards/accuracies": 0.59375, "rewards/chosen": -0.008494555950164795, "rewards/margins": 0.0018449639901518822, "rewards/rejected": -0.010339519940316677, "step": 30 }, { "epoch": 0.06, "learning_rate": 3.1746031746031746e-06, "logits/chosen": 0.17288276553153992, "logits/rejected": 0.2692878246307373, "logps/chosen": -518.7865600585938, "logps/rejected": -552.9195556640625, "loss": 0.6906, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -0.03214402496814728, "rewards/margins": 0.004966778215020895, "rewards/rejected": -0.03711080551147461, "step": 40 }, { "epoch": 0.08, "learning_rate": 3.968253968253968e-06, "logits/chosen": 0.2141023576259613, "logits/rejected": 0.2770322263240814, "logps/chosen": -574.9302978515625, "logps/rejected": -587.6199951171875, "loss": 0.6899, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -0.07075769454240799, "rewards/margins": 0.009018848650157452, "rewards/rejected": -0.07977654784917831, "step": 50 }, { "epoch": 0.1, "learning_rate": 4.761904761904762e-06, "logits/chosen": 0.20746545493602753, "logits/rejected": 0.22541038691997528, "logps/chosen": -621.58935546875, "logps/rejected": -625.3094482421875, "loss": 0.6895, "rewards/accuracies": 0.48124998807907104, "rewards/chosen": -0.10110817104578018, "rewards/margins": 0.006856895051896572, "rewards/rejected": -0.10796506702899933, "step": 60 }, { "epoch": 0.11, "learning_rate": 4.998086282661188e-06, "logits/chosen": 0.2066463679075241, "logits/rejected": 0.218001127243042, "logps/chosen": -615.7960815429688, "logps/rejected": -646.2564697265625, "loss": 0.6859, "rewards/accuracies": 0.5, "rewards/chosen": -0.1264130175113678, "rewards/margins": 0.008419327437877655, "rewards/rejected": -0.13483233749866486, "step": 70 }, { "epoch": 0.13, "learning_rate": 4.988720025682995e-06, "logits/chosen": 0.17233335971832275, "logits/rejected": 0.15302792191505432, "logps/chosen": -658.7843017578125, "logps/rejected": -671.2628173828125, "loss": 0.6848, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -0.12070327997207642, "rewards/margins": 0.018190523609519005, "rewards/rejected": -0.13889379799365997, "step": 80 }, { "epoch": 0.14, "learning_rate": 4.9715789537359126e-06, "logits/chosen": 0.10335955768823624, "logits/rejected": 0.1700659692287445, "logps/chosen": -580.2098999023438, "logps/rejected": -629.610595703125, "loss": 0.6797, "rewards/accuracies": 0.5625, "rewards/chosen": -0.14437471330165863, "rewards/margins": 0.03023264743387699, "rewards/rejected": -0.17460736632347107, "step": 90 }, { "epoch": 0.16, "learning_rate": 4.946716615897932e-06, "logits/chosen": 0.08332869410514832, "logits/rejected": 0.09228050708770752, "logps/chosen": -648.9441528320312, "logps/rejected": -705.8828735351562, "loss": 0.6796, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -0.19636474549770355, "rewards/margins": 0.02650091052055359, "rewards/rejected": -0.22286565601825714, "step": 100 }, { "epoch": 0.18, "learning_rate": 4.9142106826480114e-06, "logits/chosen": 0.01861090026795864, "logits/rejected": 0.041032858192920685, "logps/chosen": -754.0704956054688, "logps/rejected": -777.1426391601562, "loss": 0.6776, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -0.27138790488243103, "rewards/margins": 0.0310644693672657, "rewards/rejected": -0.3024523854255676, "step": 110 }, { "epoch": 0.19, "learning_rate": 4.874162703221823e-06, "logits/chosen": -0.012570058926939964, "logits/rejected": -0.028305906802415848, "logps/chosen": -888.3677978515625, "logps/rejected": -892.1744384765625, "loss": 0.6705, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -0.34433773159980774, "rewards/margins": 0.03458678722381592, "rewards/rejected": -0.3789244592189789, "step": 120 }, { "epoch": 0.21, "learning_rate": 4.826697788369752e-06, "logits/chosen": -0.039124295115470886, "logits/rejected": 0.005070955958217382, "logps/chosen": -856.4949340820312, "logps/rejected": -902.6595458984375, "loss": 0.6731, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -0.3688521981239319, "rewards/margins": 0.05119430273771286, "rewards/rejected": -0.42004650831222534, "step": 130 }, { "epoch": 0.22, "learning_rate": 4.7719642195082224e-06, "logits/chosen": -0.051692504435777664, "logits/rejected": 0.06696543842554092, "logps/chosen": -899.3610229492188, "logps/rejected": -987.4519653320312, "loss": 0.6609, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -0.38903677463531494, "rewards/margins": 0.10688074678182602, "rewards/rejected": -0.49591749906539917, "step": 140 }, { "epoch": 0.24, "learning_rate": 4.710132985485355e-06, "logits/chosen": 0.04679523780941963, "logits/rejected": 0.16161568462848663, "logps/chosen": -1067.924072265625, "logps/rejected": -1131.458984375, "loss": 0.6745, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -0.5985130071640015, "rewards/margins": 0.07710713148117065, "rewards/rejected": -0.6756200790405273, "step": 150 }, { "epoch": 0.26, "learning_rate": 4.641397248408122e-06, "logits/chosen": 0.12509503960609436, "logits/rejected": 0.20963266491889954, "logps/chosen": -970.3092651367188, "logps/rejected": -1117.892333984375, "loss": 0.6506, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -0.4672892987728119, "rewards/margins": 0.15404292941093445, "rewards/rejected": -0.6213322877883911, "step": 160 }, { "epoch": 0.27, "learning_rate": 4.5659717401997655e-06, "logits/chosen": 0.1800670623779297, "logits/rejected": 0.20961984992027283, "logps/chosen": -1060.942626953125, "logps/rejected": -1264.063232421875, "loss": 0.6553, "rewards/accuracies": 0.625, "rewards/chosen": -0.5244041681289673, "rewards/margins": 0.17471647262573242, "rewards/rejected": -0.6991206407546997, "step": 170 }, { "epoch": 0.29, "learning_rate": 4.4840920917726425e-06, "logits/chosen": 0.25674110651016235, "logits/rejected": 0.2803070843219757, "logps/chosen": -1182.87060546875, "logps/rejected": -1315.55712890625, "loss": 0.6543, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -0.7135756611824036, "rewards/margins": 0.13786718249320984, "rewards/rejected": -0.851442813873291, "step": 180 }, { "epoch": 0.3, "learning_rate": 4.396014096912182e-06, "logits/chosen": 0.1727183610200882, "logits/rejected": 0.35109418630599976, "logps/chosen": -1045.9290771484375, "logps/rejected": -1137.6866455078125, "loss": 0.6689, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -0.5587235689163208, "rewards/margins": 0.10695569217205048, "rewards/rejected": -0.6656793355941772, "step": 190 }, { "epoch": 0.32, "learning_rate": 4.302012913171584e-06, "logits/chosen": 0.21685247123241425, "logits/rejected": 0.1864197552204132, "logps/chosen": -865.9853515625, "logps/rejected": -946.7151489257812, "loss": 0.6622, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -0.3805222511291504, "rewards/margins": 0.08247803151607513, "rewards/rejected": -0.46300023794174194, "step": 200 }, { "epoch": 0.34, "learning_rate": 4.202382202273702e-06, "logits/chosen": 0.3170011639595032, "logits/rejected": 0.26561760902404785, "logps/chosen": -860.9000854492188, "logps/rejected": -995.12353515625, "loss": 0.6505, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -0.4116226136684418, "rewards/margins": 0.10268483310937881, "rewards/rejected": -0.5143074989318848, "step": 210 }, { "epoch": 0.35, "learning_rate": 4.097433212705492e-06, "logits/chosen": 0.32917365431785583, "logits/rejected": 0.422170490026474, "logps/chosen": -1080.322021484375, "logps/rejected": -1246.759521484375, "loss": 0.6454, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -0.6017876863479614, "rewards/margins": 0.16769546270370483, "rewards/rejected": -0.7694831490516663, "step": 220 }, { "epoch": 0.37, "learning_rate": 3.987493807371033e-06, "logits/chosen": 0.3952299654483795, "logits/rejected": 0.39465999603271484, "logps/chosen": -1251.5406494140625, "logps/rejected": -1388.2440185546875, "loss": 0.6442, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -0.7216378450393677, "rewards/margins": 0.1245543584227562, "rewards/rejected": -0.8461921811103821, "step": 230 }, { "epoch": 0.38, "learning_rate": 3.872907439340758e-06, "logits/chosen": 0.4063127934932709, "logits/rejected": 0.47256404161453247, "logps/chosen": -1212.349853515625, "logps/rejected": -1397.411865234375, "loss": 0.6451, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -0.7250560522079468, "rewards/margins": 0.1564178168773651, "rewards/rejected": -0.8814738392829895, "step": 240 }, { "epoch": 0.4, "learning_rate": 3.75403207889666e-06, "logits/chosen": 0.3914504051208496, "logits/rejected": 0.3554847240447998, "logps/chosen": -1249.439208984375, "logps/rejected": -1423.4649658203125, "loss": 0.6436, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -0.7152602076530457, "rewards/margins": 0.17244547605514526, "rewards/rejected": -0.8877056837081909, "step": 250 }, { "epoch": 0.42, "learning_rate": 3.631239095225417e-06, "logits/chosen": 0.3514712452888489, "logits/rejected": 0.45444172620773315, "logps/chosen": -1238.7828369140625, "logps/rejected": -1391.1258544921875, "loss": 0.6492, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -0.7164724469184875, "rewards/margins": 0.16447243094444275, "rewards/rejected": -0.8809448480606079, "step": 260 }, { "epoch": 0.43, "learning_rate": 3.5049120962530608e-06, "logits/chosen": 0.375350683927536, "logits/rejected": 0.4519611895084381, "logps/chosen": -1169.996337890625, "logps/rejected": -1291.4034423828125, "loss": 0.6529, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -0.6447538137435913, "rewards/margins": 0.12557630240917206, "rewards/rejected": -0.7703300714492798, "step": 270 }, { "epoch": 0.45, "learning_rate": 3.3754457302455464e-06, "logits/chosen": 0.3003133237361908, "logits/rejected": 0.448310911655426, "logps/chosen": -1107.552734375, "logps/rejected": -1246.8009033203125, "loss": 0.6489, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -0.6071010828018188, "rewards/margins": 0.13486045598983765, "rewards/rejected": -0.7419615387916565, "step": 280 }, { "epoch": 0.46, "learning_rate": 3.2432444529190714e-06, "logits/chosen": 0.3579423427581787, "logits/rejected": 0.432018905878067, "logps/chosen": -1139.100830078125, "logps/rejected": -1201.1663818359375, "loss": 0.6495, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -0.6468966007232666, "rewards/margins": 0.0843476727604866, "rewards/rejected": -0.7312443852424622, "step": 290 }, { "epoch": 0.48, "learning_rate": 3.1087212639117057e-06, "logits/chosen": 0.3242015540599823, "logits/rejected": 0.4742099642753601, "logps/chosen": -1214.9383544921875, "logps/rejected": -1312.9287109375, "loss": 0.6498, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -0.6838893294334412, "rewards/margins": 0.11757048219442368, "rewards/rejected": -0.801459789276123, "step": 300 }, { "epoch": 0.5, "learning_rate": 2.9722964165636263e-06, "logits/chosen": 0.4161956310272217, "logits/rejected": 0.47839951515197754, "logps/chosen": -1124.3660888671875, "logps/rejected": -1247.5843505859375, "loss": 0.6398, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -0.6321052312850952, "rewards/margins": 0.10825280100107193, "rewards/rejected": -0.7403579950332642, "step": 310 }, { "epoch": 0.51, "learning_rate": 2.8343961050366275e-06, "logits/chosen": 0.46386805176734924, "logits/rejected": 0.5212578773498535, "logps/chosen": -1118.0859375, "logps/rejected": -1206.055908203125, "loss": 0.6473, "rewards/accuracies": 0.581250011920929, "rewards/chosen": -0.6330564022064209, "rewards/margins": 0.102098748087883, "rewards/rejected": -0.7351552248001099, "step": 320 }, { "epoch": 0.53, "learning_rate": 2.695451132874385e-06, "logits/chosen": 0.5020041465759277, "logits/rejected": 0.4887539744377136, "logps/chosen": -1234.7464599609375, "logps/rejected": -1376.104736328125, "loss": 0.6393, "rewards/accuracies": 0.53125, "rewards/chosen": -0.7160638570785522, "rewards/margins": 0.1427764594554901, "rewards/rejected": -0.85884028673172, "step": 330 }, { "epoch": 0.54, "learning_rate": 2.5558955671628964e-06, "logits/chosen": 0.44761380553245544, "logits/rejected": 0.5008188486099243, "logps/chosen": -1225.663818359375, "logps/rejected": -1363.5836181640625, "loss": 0.6549, "rewards/accuracies": 0.59375, "rewards/chosen": -0.7390362620353699, "rewards/margins": 0.1303921490907669, "rewards/rejected": -0.8694285154342651, "step": 340 }, { "epoch": 0.56, "learning_rate": 2.4161653824955654e-06, "logits/chosen": 0.502762496471405, "logits/rejected": 0.521251380443573, "logps/chosen": -1289.813720703125, "logps/rejected": -1509.394775390625, "loss": 0.6337, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -0.7973452806472778, "rewards/margins": 0.20352289080619812, "rewards/rejected": -1.0008680820465088, "step": 350 }, { "epoch": 0.58, "learning_rate": 2.2766970989791697e-06, "logits/chosen": 0.5564379096031189, "logits/rejected": 0.4828278422355652, "logps/chosen": -1317.9136962890625, "logps/rejected": -1490.851806640625, "loss": 0.6489, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -0.8256872296333313, "rewards/margins": 0.14507783949375153, "rewards/rejected": -0.9707651138305664, "step": 360 }, { "epoch": 0.59, "learning_rate": 2.1379264185356545e-06, "logits/chosen": 0.4197818338871002, "logits/rejected": 0.4475874900817871, "logps/chosen": -1211.713623046875, "logps/rejected": -1361.810791015625, "loss": 0.6479, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -0.7186134457588196, "rewards/margins": 0.15284962952136993, "rewards/rejected": -0.8714631199836731, "step": 370 }, { "epoch": 0.61, "learning_rate": 2.000286863759934e-06, "logits/chosen": 0.4016849398612976, "logits/rejected": 0.6099621057510376, "logps/chosen": -1292.288818359375, "logps/rejected": -1514.7738037109375, "loss": 0.6132, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -0.8104116320610046, "rewards/margins": 0.20091453194618225, "rewards/rejected": -1.0113261938095093, "step": 380 }, { "epoch": 0.62, "learning_rate": 1.8642084235859764e-06, "logits/chosen": 0.551924467086792, "logits/rejected": 0.6596301198005676, "logps/chosen": -1387.1436767578125, "logps/rejected": -1532.2877197265625, "loss": 0.6588, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -0.913176417350769, "rewards/margins": 0.1407497525215149, "rewards/rejected": -1.0539262294769287, "step": 390 }, { "epoch": 0.64, "learning_rate": 1.7301162099921013e-06, "logits/chosen": 0.5529540777206421, "logits/rejected": 0.5772652626037598, "logps/chosen": -1380.193115234375, "logps/rejected": -1582.646484375, "loss": 0.6341, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -0.8533477783203125, "rewards/margins": 0.18637995421886444, "rewards/rejected": -1.0397276878356934, "step": 400 }, { "epoch": 0.66, "learning_rate": 1.5984291299420117e-06, "logits/chosen": 0.47877341508865356, "logits/rejected": 0.5813654661178589, "logps/chosen": -1256.6849365234375, "logps/rejected": -1472.9185791015625, "loss": 0.6404, "rewards/accuracies": 0.5625, "rewards/chosen": -0.7804775238037109, "rewards/margins": 0.19573338329792023, "rewards/rejected": -0.97621089220047, "step": 410 }, { "epoch": 0.67, "learning_rate": 1.4695585767104092e-06, "logits/chosen": 0.41678476333618164, "logits/rejected": 0.6133105754852295, "logps/chosen": -1238.5712890625, "logps/rejected": -1406.7432861328125, "loss": 0.6409, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -0.7545350193977356, "rewards/margins": 0.1725122034549713, "rewards/rejected": -0.9270472526550293, "step": 420 }, { "epoch": 0.69, "learning_rate": 1.3439071446815452e-06, "logits/chosen": 0.4347582459449768, "logits/rejected": 0.5867087244987488, "logps/chosen": -1281.6087646484375, "logps/rejected": -1513.9439697265625, "loss": 0.6387, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -0.7874932289123535, "rewards/margins": 0.2120811492204666, "rewards/rejected": -0.9995743632316589, "step": 430 }, { "epoch": 0.7, "learning_rate": 1.2218673716356919e-06, "logits/chosen": 0.5029743909835815, "logits/rejected": 0.591041088104248, "logps/chosen": -1364.179931640625, "logps/rejected": -1468.728759765625, "loss": 0.6623, "rewards/accuracies": 0.59375, "rewards/chosen": -0.8501272201538086, "rewards/margins": 0.09424228221178055, "rewards/rejected": -0.9443694949150085, "step": 440 }, { "epoch": 0.72, "learning_rate": 1.103820512452661e-06, "logits/chosen": 0.53369140625, "logits/rejected": 0.5491160154342651, "logps/chosen": -1355.01171875, "logps/rejected": -1442.1351318359375, "loss": 0.6686, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -0.8647180795669556, "rewards/margins": 0.09323947131633759, "rewards/rejected": -0.9579576253890991, "step": 450 }, { "epoch": 0.74, "learning_rate": 9.901353480633468e-07, "logits/chosen": 0.5199618339538574, "logits/rejected": 0.4953738749027252, "logps/chosen": -1270.599853515625, "logps/rejected": -1456.319580078125, "loss": 0.6241, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -0.7807380557060242, "rewards/margins": 0.18214623630046844, "rewards/rejected": -0.9628843069076538, "step": 460 }, { "epoch": 0.75, "learning_rate": 8.811670333701544e-07, "logits/chosen": 0.5168347954750061, "logits/rejected": 0.6290295124053955, "logps/chosen": -1292.0872802734375, "logps/rejected": -1423.704345703125, "loss": 0.639, "rewards/accuracies": 0.59375, "rewards/chosen": -0.771618664264679, "rewards/margins": 0.1294311136007309, "rewards/rejected": -0.901049792766571, "step": 470 }, { "epoch": 0.77, "learning_rate": 7.772559877354341e-07, "logits/chosen": 0.5879518985748291, "logits/rejected": 0.6285650730133057, "logps/chosen": -1255.3343505859375, "logps/rejected": -1419.4129638671875, "loss": 0.6316, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -0.8007827997207642, "rewards/margins": 0.14159244298934937, "rewards/rejected": -0.9423751831054688, "step": 480 }, { "epoch": 0.78, "learning_rate": 6.787268315040604e-07, "logits/chosen": 0.45197635889053345, "logits/rejected": 0.5737031698226929, "logps/chosen": -1276.957275390625, "logps/rejected": -1509.0966796875, "loss": 0.6336, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -0.8125177621841431, "rewards/margins": 0.19976083934307098, "rewards/rejected": -1.012278437614441, "step": 490 }, { "epoch": 0.8, "learning_rate": 5.858873718824829e-07, "logits/chosen": 0.5449025630950928, "logits/rejected": 0.6064163446426392, "logps/chosen": -1345.1566162109375, "logps/rejected": -1465.4927978515625, "loss": 0.6838, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -0.8671814799308777, "rewards/margins": 0.12205106019973755, "rewards/rejected": -0.9892325401306152, "step": 500 }, { "epoch": 0.82, "learning_rate": 4.990276413423817e-07, "logits/chosen": 0.45466941595077515, "logits/rejected": 0.5977617502212524, "logps/chosen": -1280.716064453125, "logps/rejected": -1477.654541015625, "loss": 0.6267, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -0.7988293766975403, "rewards/margins": 0.19528616964817047, "rewards/rejected": -0.9941156506538391, "step": 510 }, { "epoch": 0.83, "learning_rate": 4.184189915529796e-07, "logits/chosen": 0.5602769255638123, "logits/rejected": 0.611350953578949, "logps/chosen": -1306.686279296875, "logps/rejected": -1565.746826171875, "loss": 0.6323, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -0.794352650642395, "rewards/margins": 0.22704383730888367, "rewards/rejected": -1.0213963985443115, "step": 520 }, { "epoch": 0.85, "learning_rate": 3.4431324567258176e-07, "logits/chosen": 0.5258222222328186, "logits/rejected": 0.5747347474098206, "logps/chosen": -1296.38916015625, "logps/rejected": -1511.482666015625, "loss": 0.6341, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -0.7807685136795044, "rewards/margins": 0.19532978534698486, "rewards/rejected": -0.9760982394218445, "step": 530 }, { "epoch": 0.86, "learning_rate": 2.769419116476052e-07, "logits/chosen": 0.46619558334350586, "logits/rejected": 0.6566539406776428, "logps/chosen": -1344.41748046875, "logps/rejected": -1466.8358154296875, "loss": 0.6556, "rewards/accuracies": 0.59375, "rewards/chosen": -0.8279752731323242, "rewards/margins": 0.1466597020626068, "rewards/rejected": -0.9746350049972534, "step": 540 }, { "epoch": 0.88, "learning_rate": 2.1651545897676512e-07, "logits/chosen": 0.5445398092269897, "logits/rejected": 0.5394152998924255, "logps/chosen": -1226.078125, "logps/rejected": -1361.6815185546875, "loss": 0.6609, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -0.7584909200668335, "rewards/margins": 0.11591245234012604, "rewards/rejected": -0.8744033575057983, "step": 550 }, { "epoch": 0.9, "learning_rate": 1.6322266119983222e-07, "logits/chosen": 0.5224272012710571, "logits/rejected": 0.6026440858840942, "logps/chosen": -1239.085693359375, "logps/rejected": -1359.008544921875, "loss": 0.6537, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -0.7636959552764893, "rewards/margins": 0.1270333230495453, "rewards/rejected": -0.8907292485237122, "step": 560 }, { "epoch": 0.91, "learning_rate": 1.1723000616502167e-07, "logits/chosen": 0.44439181685447693, "logits/rejected": 0.587325930595398, "logps/chosen": -1246.112548828125, "logps/rejected": -1382.4180908203125, "loss": 0.6502, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -0.7664066553115845, "rewards/margins": 0.1451866328716278, "rewards/rejected": -0.9115932583808899, "step": 570 }, { "epoch": 0.93, "learning_rate": 7.868117591737585e-08, "logits/chosen": 0.5547848343849182, "logits/rejected": 0.6335003972053528, "logps/chosen": -1389.32470703125, "logps/rejected": -1530.272705078125, "loss": 0.6396, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -0.8785858154296875, "rewards/margins": 0.14252634346485138, "rewards/rejected": -1.0211122035980225, "step": 580 }, { "epoch": 0.94, "learning_rate": 4.769659783295383e-08, "logits/chosen": 0.5736573338508606, "logits/rejected": 0.6307505369186401, "logps/chosen": -1240.878662109375, "logps/rejected": -1390.9312744140625, "loss": 0.6415, "rewards/accuracies": 0.5687500238418579, "rewards/chosen": -0.8237506747245789, "rewards/margins": 0.1187644749879837, "rewards/rejected": -0.9425151944160461, "step": 590 }, { "epoch": 0.96, "learning_rate": 2.4373068401120358e-08, "logits/chosen": 0.4911600947380066, "logits/rejected": 0.6507722735404968, "logps/chosen": -1280.173095703125, "logps/rejected": -1408.863037109375, "loss": 0.6423, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -0.7916342616081238, "rewards/margins": 0.13446274399757385, "rewards/rejected": -0.92609703540802, "step": 600 }, { "epoch": 0.98, "learning_rate": 8.78345083022425e-09, "logits/chosen": 0.438457727432251, "logits/rejected": 0.5983772277832031, "logps/chosen": -1361.9815673828125, "logps/rejected": -1418.54541015625, "loss": 0.6503, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -0.8475711941719055, "rewards/margins": 0.06368283927440643, "rewards/rejected": -0.9112540483474731, "step": 610 }, { "epoch": 0.99, "learning_rate": 9.764474213677654e-10, "logits/chosen": 0.4755234122276306, "logits/rejected": 0.6281898617744446, "logps/chosen": -1357.261962890625, "logps/rejected": -1532.4453125, "loss": 0.6579, "rewards/accuracies": 0.6187499761581421, "rewards/chosen": -0.8383301496505737, "rewards/margins": 0.17498043179512024, "rewards/rejected": -1.0133105516433716, "step": 620 }, { "epoch": 1.0, "step": 625, "total_flos": 0.0, "train_loss": 0.6552388866424561, "train_runtime": 7636.4902, "train_samples_per_second": 2.619, "train_steps_per_second": 0.082 } ], "logging_steps": 10, "max_steps": 625, "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 }