{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 501, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.001996007984031936, "grad_norm": 16.492788975984936, "learning_rate": 1.9607843137254902e-08, "logits/chosen": 1.3011724948883057, "logits/rejected": 1.3879833221435547, "logps/chosen": -0.9583934545516968, "logps/rejected": -0.908146858215332, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.00998003992015968, "grad_norm": 18.616408754077472, "learning_rate": 9.80392156862745e-08, "logits/chosen": 1.230096697807312, "logits/rejected": 1.2526068687438965, "logps/chosen": -0.8566818833351135, "logps/rejected": -0.9451415538787842, "loss": 0.6931, "rewards/accuracies": 0.34375, "rewards/chosen": -0.00325668603181839, "rewards/margins": -0.0027664820663630962, "rewards/rejected": -0.0004902039654552937, "step": 5 }, { "epoch": 0.01996007984031936, "grad_norm": 19.72943287901331, "learning_rate": 1.96078431372549e-07, "logits/chosen": 1.1732714176177979, "logits/rejected": 1.2325665950775146, "logps/chosen": -0.8571804165840149, "logps/rejected": -0.9293950200080872, "loss": 0.6935, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -0.0004458148032426834, "rewards/margins": 0.00015245657414197922, "rewards/rejected": -0.0005982713773846626, "step": 10 }, { "epoch": 0.029940119760479042, "grad_norm": 15.989882260008397, "learning_rate": 2.941176470588235e-07, "logits/chosen": 1.1804090738296509, "logits/rejected": 1.227670669555664, "logps/chosen": -0.8887054324150085, "logps/rejected": -0.9415630102157593, "loss": 0.6942, "rewards/accuracies": 0.4124999940395355, "rewards/chosen": -0.005856434814631939, "rewards/margins": -0.004528886638581753, "rewards/rejected": -0.0013275481760501862, "step": 15 }, { "epoch": 0.03992015968063872, "grad_norm": 21.617808491664096, "learning_rate": 3.92156862745098e-07, "logits/chosen": 1.2326724529266357, "logits/rejected": 1.239334225654602, "logps/chosen": -1.025779366493225, "logps/rejected": -0.9815893173217773, "loss": 0.6924, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -0.011453075334429741, "rewards/margins": -0.006528028752654791, "rewards/rejected": -0.004925046116113663, "step": 20 }, { "epoch": 0.0499001996007984, "grad_norm": 14.85293949325317, "learning_rate": 4.901960784313725e-07, "logits/chosen": 1.1621646881103516, "logits/rejected": 1.2202112674713135, "logps/chosen": -0.9435383081436157, "logps/rejected": -1.0095245838165283, "loss": 0.6926, "rewards/accuracies": 0.4749999940395355, "rewards/chosen": -0.008930020034313202, "rewards/margins": 0.011791445314884186, "rewards/rejected": -0.020721465349197388, "step": 25 }, { "epoch": 0.059880239520958084, "grad_norm": 18.28049213488445, "learning_rate": 5.88235294117647e-07, "logits/chosen": 1.2528660297393799, "logits/rejected": 1.3036034107208252, "logps/chosen": -0.9394888877868652, "logps/rejected": -0.9887625575065613, "loss": 0.6928, "rewards/accuracies": 0.5, "rewards/chosen": -0.020373398438096046, "rewards/margins": -0.0037106983363628387, "rewards/rejected": -0.016662701964378357, "step": 30 }, { "epoch": 0.06986027944111776, "grad_norm": 15.999817261895291, "learning_rate": 6.862745098039216e-07, "logits/chosen": 1.1911170482635498, "logits/rejected": 1.2343870401382446, "logps/chosen": -0.9691197276115417, "logps/rejected": -0.9998400807380676, "loss": 0.6912, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -0.01592147909104824, "rewards/margins": 0.0014309640973806381, "rewards/rejected": -0.01735244318842888, "step": 35 }, { "epoch": 0.07984031936127745, "grad_norm": 14.867532956796651, "learning_rate": 7.84313725490196e-07, "logits/chosen": 1.2250399589538574, "logits/rejected": 1.2877371311187744, "logps/chosen": -0.9882305860519409, "logps/rejected": -1.013457179069519, "loss": 0.6866, "rewards/accuracies": 0.625, "rewards/chosen": -0.014133378863334656, "rewards/margins": 0.023214522749185562, "rewards/rejected": -0.03734790161252022, "step": 40 }, { "epoch": 0.08982035928143713, "grad_norm": 19.324457876267314, "learning_rate": 8.823529411764705e-07, "logits/chosen": 1.2738926410675049, "logits/rejected": 1.2850474119186401, "logps/chosen": -0.9675604701042175, "logps/rejected": -1.0223758220672607, "loss": 0.6804, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -0.010748568922281265, "rewards/margins": 0.0401780791580677, "rewards/rejected": -0.05092664435505867, "step": 45 }, { "epoch": 0.0998003992015968, "grad_norm": 18.41965619240879, "learning_rate": 9.80392156862745e-07, "logits/chosen": 1.2459691762924194, "logits/rejected": 1.3021132946014404, "logps/chosen": -0.9164802432060242, "logps/rejected": -1.013667345046997, "loss": 0.6739, "rewards/accuracies": 0.637499988079071, "rewards/chosen": -0.026862669736146927, "rewards/margins": 0.07785479724407196, "rewards/rejected": -0.10471747070550919, "step": 50 }, { "epoch": 0.10978043912175649, "grad_norm": 16.897558680765105, "learning_rate": 9.99805057520177e-07, "logits/chosen": 1.1725871562957764, "logits/rejected": 1.173532247543335, "logps/chosen": -0.9711742401123047, "logps/rejected": -0.9595904350280762, "loss": 0.6738, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.03100317157804966, "rewards/margins": 0.11643954366445541, "rewards/rejected": -0.14744271337985992, "step": 55 }, { "epoch": 0.11976047904191617, "grad_norm": 14.265898399249132, "learning_rate": 9.990133642141357e-07, "logits/chosen": 1.1897612810134888, "logits/rejected": 1.2000809907913208, "logps/chosen": -0.9207404255867004, "logps/rejected": -0.9565631747245789, "loss": 0.6693, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -0.06663166731595993, "rewards/margins": 0.18790733814239502, "rewards/rejected": -0.25453901290893555, "step": 60 }, { "epoch": 0.12974051896207583, "grad_norm": 19.11690615448299, "learning_rate": 9.976136999909155e-07, "logits/chosen": 1.261150598526001, "logits/rejected": 1.2706950902938843, "logps/chosen": -0.9832024574279785, "logps/rejected": -0.9762998819351196, "loss": 0.6736, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -0.08731752634048462, "rewards/margins": 0.037145014852285385, "rewards/rejected": -0.1244625449180603, "step": 65 }, { "epoch": 0.13972055888223553, "grad_norm": 16.308947856088572, "learning_rate": 9.956077701257707e-07, "logits/chosen": 1.1949059963226318, "logits/rejected": 1.2373930215835571, "logps/chosen": -0.923254132270813, "logps/rejected": -0.9537776708602905, "loss": 0.6588, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -0.0689881294965744, "rewards/margins": 0.0771615132689476, "rewards/rejected": -0.1461496502161026, "step": 70 }, { "epoch": 0.1497005988023952, "grad_norm": 18.31597399561215, "learning_rate": 9.929980185352525e-07, "logits/chosen": 1.1303855180740356, "logits/rejected": 1.166472315788269, "logps/chosen": -1.0079195499420166, "logps/rejected": -1.0075769424438477, "loss": 0.6486, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.026683980599045753, "rewards/margins": 0.0837903693318367, "rewards/rejected": -0.1104743480682373, "step": 75 }, { "epoch": 0.1596806387225549, "grad_norm": 16.142069213484454, "learning_rate": 9.89787624799672e-07, "logits/chosen": 1.1199300289154053, "logits/rejected": 1.1778942346572876, "logps/chosen": -0.9652664065361023, "logps/rejected": -1.0307899713516235, "loss": 0.6496, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.0005544237792491913, "rewards/margins": 0.10419625043869019, "rewards/rejected": -0.10475067049264908, "step": 80 }, { "epoch": 0.16966067864271456, "grad_norm": 15.306215682029483, "learning_rate": 9.859805002892731e-07, "logits/chosen": 1.1645634174346924, "logits/rejected": 1.1519969701766968, "logps/chosen": -0.9917243719100952, "logps/rejected": -1.060416340827942, "loss": 0.661, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -0.09171438962221146, "rewards/margins": 0.20642364025115967, "rewards/rejected": -0.2981380820274353, "step": 85 }, { "epoch": 0.17964071856287425, "grad_norm": 15.254186655042064, "learning_rate": 9.81581283398829e-07, "logits/chosen": 1.103563904762268, "logits/rejected": 1.1300289630889893, "logps/chosen": -0.969647228717804, "logps/rejected": -0.9916777610778809, "loss": 0.6337, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": 0.013281228020787239, "rewards/margins": 0.1793346405029297, "rewards/rejected": -0.1660534292459488, "step": 90 }, { "epoch": 0.18962075848303392, "grad_norm": 15.282160453652757, "learning_rate": 9.765953338964734e-07, "logits/chosen": 1.0921884775161743, "logits/rejected": 1.1226933002471924, "logps/chosen": -1.0730102062225342, "logps/rejected": -1.1024601459503174, "loss": 0.6385, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": 0.0058188484981656075, "rewards/margins": 0.21344614028930664, "rewards/rejected": -0.2076272964477539, "step": 95 }, { "epoch": 0.1996007984031936, "grad_norm": 21.61748232589367, "learning_rate": 9.710287263936483e-07, "logits/chosen": 1.0795061588287354, "logits/rejected": 1.0822367668151855, "logps/chosen": -1.0821864604949951, "logps/rejected": -1.1186153888702393, "loss": 0.621, "rewards/accuracies": 0.6875, "rewards/chosen": 0.07217474281787872, "rewards/margins": 0.3310089707374573, "rewards/rejected": -0.25883427262306213, "step": 100 }, { "epoch": 0.20958083832335328, "grad_norm": 23.129340339628143, "learning_rate": 9.648882429441256e-07, "logits/chosen": 1.0758105516433716, "logits/rejected": 1.0959206819534302, "logps/chosen": -0.9559111595153809, "logps/rejected": -0.9702944755554199, "loss": 0.6278, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.027358528226614, "rewards/margins": 0.17237558960914612, "rewards/rejected": -0.1997341364622116, "step": 105 }, { "epoch": 0.21956087824351297, "grad_norm": 15.179993551058994, "learning_rate": 9.581813647811197e-07, "logits/chosen": 1.094158411026001, "logits/rejected": 1.114650011062622, "logps/chosen": -0.9534612894058228, "logps/rejected": -0.9674752354621887, "loss": 0.6124, "rewards/accuracies": 0.75, "rewards/chosen": 0.058080364018678665, "rewards/margins": 0.30136996507644653, "rewards/rejected": -0.24328958988189697, "step": 110 }, { "epoch": 0.22954091816367264, "grad_norm": 13.705886523650493, "learning_rate": 9.509162632025569e-07, "logits/chosen": 1.0607202053070068, "logits/rejected": 1.045424461364746, "logps/chosen": -1.1245949268341064, "logps/rejected": -1.1509120464324951, "loss": 0.6167, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -0.07293939590454102, "rewards/margins": 0.17659255862236023, "rewards/rejected": -0.24953195452690125, "step": 115 }, { "epoch": 0.23952095808383234, "grad_norm": 27.63745397695, "learning_rate": 9.431017896156073e-07, "logits/chosen": 1.0884307622909546, "logits/rejected": 1.101423978805542, "logps/chosen": -0.9731513857841492, "logps/rejected": -1.023944616317749, "loss": 0.6061, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": 0.055571239441633224, "rewards/margins": 0.36377543210983276, "rewards/rejected": -0.3082042336463928, "step": 120 }, { "epoch": 0.249500998003992, "grad_norm": 14.89222449165367, "learning_rate": 9.347474647526095e-07, "logits/chosen": 1.0270758867263794, "logits/rejected": 1.061629056930542, "logps/chosen": -0.9520384073257446, "logps/rejected": -0.9609392285346985, "loss": 0.5982, "rewards/accuracies": 0.75, "rewards/chosen": -0.03607760742306709, "rewards/margins": 0.205668643116951, "rewards/rejected": -0.24174626171588898, "step": 125 }, { "epoch": 0.25948103792415167, "grad_norm": 16.598481366896863, "learning_rate": 9.258634670715237e-07, "logits/chosen": 0.9625661969184875, "logits/rejected": 0.9923114776611328, "logps/chosen": -0.9781230092048645, "logps/rejected": -1.0380465984344482, "loss": 0.6033, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -0.08731977641582489, "rewards/margins": 0.32325008511543274, "rewards/rejected": -0.41056984663009644, "step": 130 }, { "epoch": 0.2694610778443114, "grad_norm": 17.332548482467526, "learning_rate": 9.164606203550497e-07, "logits/chosen": 1.0289146900177002, "logits/rejected": 1.0656378269195557, "logps/chosen": -0.898147463798523, "logps/rejected": -0.949881911277771, "loss": 0.602, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.01896672695875168, "rewards/margins": 0.33230799436569214, "rewards/rejected": -0.351274698972702, "step": 135 }, { "epoch": 0.27944111776447106, "grad_norm": 16.198930205544908, "learning_rate": 9.065503805235137e-07, "logits/chosen": 1.0169068574905396, "logits/rejected": 1.0443648099899292, "logps/chosen": -0.9837234616279602, "logps/rejected": -1.015891432762146, "loss": 0.5962, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.11362324655056, "rewards/margins": 0.29118040204048157, "rewards/rejected": -0.40480366349220276, "step": 140 }, { "epoch": 0.2894211576846307, "grad_norm": 14.26896893631501, "learning_rate": 8.961448216775953e-07, "logits/chosen": 0.9983326196670532, "logits/rejected": 0.984410285949707, "logps/chosen": -1.0478496551513672, "logps/rejected": -1.0458667278289795, "loss": 0.6106, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -0.384859174489975, "rewards/margins": 0.23550406098365784, "rewards/rejected": -0.6203632354736328, "step": 145 }, { "epoch": 0.2994011976047904, "grad_norm": 17.114721290662626, "learning_rate": 8.852566213878946e-07, "logits/chosen": 0.9434808492660522, "logits/rejected": 0.9709761738777161, "logps/chosen": -0.946380615234375, "logps/rejected": -1.011891484260559, "loss": 0.6051, "rewards/accuracies": 0.8125, "rewards/chosen": -0.10636933147907257, "rewards/margins": 0.38454684615135193, "rewards/rejected": -0.4909161627292633, "step": 150 }, { "epoch": 0.3093812375249501, "grad_norm": 12.532368784046184, "learning_rate": 8.73899045249266e-07, "logits/chosen": 0.9879257082939148, "logits/rejected": 1.0284427404403687, "logps/chosen": -0.9887377619743347, "logps/rejected": -1.0425301790237427, "loss": 0.5785, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.18314163386821747, "rewards/margins": 0.3873257339000702, "rewards/rejected": -0.5704673528671265, "step": 155 }, { "epoch": 0.3193612774451098, "grad_norm": 20.653306879612042, "learning_rate": 8.620859307187338e-07, "logits/chosen": 0.9365812540054321, "logits/rejected": 0.9702633619308472, "logps/chosen": -0.9792872667312622, "logps/rejected": -1.0119134187698364, "loss": 0.5915, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.28814950585365295, "rewards/margins": 0.3159632086753845, "rewards/rejected": -0.6041126847267151, "step": 160 }, { "epoch": 0.32934131736526945, "grad_norm": 46.72961396087624, "learning_rate": 8.498316702566826e-07, "logits/chosen": 1.0088741779327393, "logits/rejected": 1.0628801584243774, "logps/chosen": -1.0191477537155151, "logps/rejected": -1.0589340925216675, "loss": 0.5784, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.43470463156700134, "rewards/margins": 0.3699150085449219, "rewards/rejected": -0.8046196103096008, "step": 165 }, { "epoch": 0.3393213572854291, "grad_norm": 17.875828780004216, "learning_rate": 8.371511937918617e-07, "logits/chosen": 0.8878538012504578, "logits/rejected": 0.9223688840866089, "logps/chosen": -0.9879400134086609, "logps/rejected": -1.029184103012085, "loss": 0.6033, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.3266729712486267, "rewards/margins": 0.5894866585731506, "rewards/rejected": -0.9161597490310669, "step": 170 }, { "epoch": 0.34930139720558884, "grad_norm": 16.753564996079323, "learning_rate": 8.240599505315654e-07, "logits/chosen": 0.8993169665336609, "logits/rejected": 0.931827187538147, "logps/chosen": -1.0300993919372559, "logps/rejected": -1.104100227355957, "loss": 0.5719, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.26795631647109985, "rewards/margins": 0.4897838234901428, "rewards/rejected": -0.7577401399612427, "step": 175 }, { "epoch": 0.3592814371257485, "grad_norm": 37.583807172799176, "learning_rate": 8.105738901391551e-07, "logits/chosen": 0.9503185153007507, "logits/rejected": 0.9612258076667786, "logps/chosen": -0.9984599947929382, "logps/rejected": -1.0410314798355103, "loss": 0.5749, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -0.3058738708496094, "rewards/margins": 0.4171958565711975, "rewards/rejected": -0.7230697274208069, "step": 180 }, { "epoch": 0.36926147704590817, "grad_norm": 31.25023262450056, "learning_rate": 7.967094433018508e-07, "logits/chosen": 0.8538301587104797, "logits/rejected": 0.8765438199043274, "logps/chosen": -1.0639922618865967, "logps/rejected": -1.1690170764923096, "loss": 0.5628, "rewards/accuracies": 0.75, "rewards/chosen": -0.31266123056411743, "rewards/margins": 0.6013716459274292, "rewards/rejected": -0.9140329360961914, "step": 185 }, { "epoch": 0.37924151696606784, "grad_norm": 21.697159728289254, "learning_rate": 7.82483501712469e-07, "logits/chosen": 0.8337196111679077, "logits/rejected": 0.8647937774658203, "logps/chosen": -1.0461093187332153, "logps/rejected": -1.0427898168563843, "loss": 0.5701, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.48062148690223694, "rewards/margins": 0.4651462435722351, "rewards/rejected": -0.9457677602767944, "step": 190 }, { "epoch": 0.38922155688622756, "grad_norm": 13.452601846054984, "learning_rate": 7.679133974894982e-07, "logits/chosen": 0.8574188351631165, "logits/rejected": 0.8771845102310181, "logps/chosen": -1.0106494426727295, "logps/rejected": -1.1300721168518066, "loss": 0.5649, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -0.5489331483840942, "rewards/margins": 0.6363680958747864, "rewards/rejected": -1.1853011846542358, "step": 195 }, { "epoch": 0.3992015968063872, "grad_norm": 14.045999687736995, "learning_rate": 7.530168820605818e-07, "logits/chosen": 0.8429155349731445, "logits/rejected": 0.8586641550064087, "logps/chosen": -1.1243480443954468, "logps/rejected": -1.126988172531128, "loss": 0.5556, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.611405074596405, "rewards/margins": 0.5212133526802063, "rewards/rejected": -1.1326184272766113, "step": 200 }, { "epoch": 0.4091816367265469, "grad_norm": 11.17840118621038, "learning_rate": 7.378121045351377e-07, "logits/chosen": 0.8805230855941772, "logits/rejected": 0.8616092801094055, "logps/chosen": -1.025155782699585, "logps/rejected": -1.1043505668640137, "loss": 0.5274, "rewards/accuracies": 0.75, "rewards/chosen": -0.5864971280097961, "rewards/margins": 0.6259642839431763, "rewards/rejected": -1.2124613523483276, "step": 205 }, { "epoch": 0.41916167664670656, "grad_norm": 17.02714464481709, "learning_rate": 7.223175895924637e-07, "logits/chosen": 0.8006958961486816, "logits/rejected": 0.8346986770629883, "logps/chosen": -1.0029652118682861, "logps/rejected": -1.0653495788574219, "loss": 0.5437, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -0.5938572287559509, "rewards/margins": 0.5961542725563049, "rewards/rejected": -1.1900113821029663, "step": 210 }, { "epoch": 0.4291417165668663, "grad_norm": 12.708513730083673, "learning_rate": 7.065522149122709e-07, "logits/chosen": 0.7250524759292603, "logits/rejected": 0.7448651790618896, "logps/chosen": -1.0977203845977783, "logps/rejected": -1.1671373844146729, "loss": 0.5296, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -0.8822553753852844, "rewards/margins": 0.5623547434806824, "rewards/rejected": -1.4446099996566772, "step": 215 }, { "epoch": 0.43912175648702595, "grad_norm": 23.02316300872982, "learning_rate": 6.905351881751371e-07, "logits/chosen": 0.7448990345001221, "logits/rejected": 0.759792685508728, "logps/chosen": -1.0590717792510986, "logps/rejected": -1.1264311075210571, "loss": 0.5411, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.7196047306060791, "rewards/margins": 0.6396269798278809, "rewards/rejected": -1.35923171043396, "step": 220 }, { "epoch": 0.4491017964071856, "grad_norm": 15.317013674505024, "learning_rate": 6.742860236609076e-07, "logits/chosen": 0.7387903928756714, "logits/rejected": 0.7482835054397583, "logps/chosen": -1.0746328830718994, "logps/rejected": -1.0935697555541992, "loss": 0.5362, "rewards/accuracies": 0.75, "rewards/chosen": -0.91168612241745, "rewards/margins": 0.43347805738449097, "rewards/rejected": -1.3451642990112305, "step": 225 }, { "epoch": 0.4590818363273453, "grad_norm": 20.979897267224555, "learning_rate": 6.578245184735512e-07, "logits/chosen": 0.7530170679092407, "logits/rejected": 0.7971817851066589, "logps/chosen": -1.0877180099487305, "logps/rejected": -1.1133688688278198, "loss": 0.5297, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.7731893062591553, "rewards/margins": 0.6359795331954956, "rewards/rejected": -1.4091689586639404, "step": 230 }, { "epoch": 0.469061876247505, "grad_norm": 23.654443057551042, "learning_rate": 6.411707284214383e-07, "logits/chosen": 0.7936985492706299, "logits/rejected": 0.8071283102035522, "logps/chosen": -1.1007978916168213, "logps/rejected": -1.175083041191101, "loss": 0.5189, "rewards/accuracies": 0.75, "rewards/chosen": -0.8833421468734741, "rewards/margins": 0.6522087454795837, "rewards/rejected": -1.535550832748413, "step": 235 }, { "epoch": 0.47904191616766467, "grad_norm": 17.617163550183303, "learning_rate": 6.243449435824276e-07, "logits/chosen": 0.7208539247512817, "logits/rejected": 0.7419721484184265, "logps/chosen": -0.9637038111686707, "logps/rejected": -1.0924652814865112, "loss": 0.5239, "rewards/accuracies": 0.75, "rewards/chosen": -0.823961079120636, "rewards/margins": 0.9264790415763855, "rewards/rejected": -1.750440239906311, "step": 240 }, { "epoch": 0.48902195608782434, "grad_norm": 23.429183199427385, "learning_rate": 6.073676635835316e-07, "logits/chosen": 0.7787541747093201, "logits/rejected": 0.8052628636360168, "logps/chosen": -1.0102448463439941, "logps/rejected": -1.2205474376678467, "loss": 0.5087, "rewards/accuracies": 0.8125, "rewards/chosen": -0.7600202560424805, "rewards/margins": 1.2399606704711914, "rewards/rejected": -1.9999806880950928, "step": 245 }, { "epoch": 0.499001996007984, "grad_norm": 15.897882119964894, "learning_rate": 5.9025957262528e-07, "logits/chosen": 0.674079179763794, "logits/rejected": 0.6560205221176147, "logps/chosen": -1.0422561168670654, "logps/rejected": -1.15647292137146, "loss": 0.5129, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -0.9762313961982727, "rewards/margins": 0.9589719772338867, "rewards/rejected": -1.9352035522460938, "step": 250 }, { "epoch": 0.5089820359281437, "grad_norm": 23.926407573258526, "learning_rate": 5.730415142812058e-07, "logits/chosen": 0.6435045003890991, "logits/rejected": 0.6540420055389404, "logps/chosen": -1.1076724529266357, "logps/rejected": -1.2272051572799683, "loss": 0.5207, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -1.1593824625015259, "rewards/margins": 0.9408347010612488, "rewards/rejected": -2.10021710395813, "step": 255 }, { "epoch": 0.5189620758483033, "grad_norm": 25.873872774153636, "learning_rate": 5.557344661031627e-07, "logits/chosen": 0.6078227758407593, "logits/rejected": 0.6234728097915649, "logps/chosen": -1.1247365474700928, "logps/rejected": -1.1961596012115479, "loss": 0.5558, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.4092304706573486, "rewards/margins": 0.7921189665794373, "rewards/rejected": -2.2013492584228516, "step": 260 }, { "epoch": 0.5289421157684631, "grad_norm": 14.32961070454234, "learning_rate": 5.383595140634093e-07, "logits/chosen": 0.634459912776947, "logits/rejected": 0.6405037641525269, "logps/chosen": -1.054018259048462, "logps/rejected": -1.1594079732894897, "loss": 0.5108, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.3187201023101807, "rewards/margins": 0.7670294046401978, "rewards/rejected": -2.085749626159668, "step": 265 }, { "epoch": 0.5389221556886228, "grad_norm": 12.765239103380393, "learning_rate": 5.209378268645997e-07, "logits/chosen": 0.6027650237083435, "logits/rejected": 0.6287568211555481, "logps/chosen": -1.008739948272705, "logps/rejected": -1.1450139284133911, "loss": 0.5318, "rewards/accuracies": 0.75, "rewards/chosen": -1.2432196140289307, "rewards/margins": 0.6896325349807739, "rewards/rejected": -1.932852029800415, "step": 270 }, { "epoch": 0.5489021956087824, "grad_norm": 17.632252940280708, "learning_rate": 5.034906301489807e-07, "logits/chosen": 0.6728609204292297, "logits/rejected": 0.6657983064651489, "logps/chosen": -1.1084893941879272, "logps/rejected": -1.182081937789917, "loss": 0.5413, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.4515244960784912, "rewards/margins": 0.8624111413955688, "rewards/rejected": -2.3139355182647705, "step": 275 }, { "epoch": 0.5588822355289421, "grad_norm": 22.324779245169193, "learning_rate": 4.860391806382156e-07, "logits/chosen": 0.6076717972755432, "logits/rejected": 0.6105688810348511, "logps/chosen": -1.0900273323059082, "logps/rejected": -1.1191543340682983, "loss": 0.512, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -1.21976900100708, "rewards/margins": 0.7310382723808289, "rewards/rejected": -1.9508073329925537, "step": 280 }, { "epoch": 0.5688622754491018, "grad_norm": 18.31410229389494, "learning_rate": 4.686047402353433e-07, "logits/chosen": 0.6535500288009644, "logits/rejected": 0.6188856959342957, "logps/chosen": -1.0886776447296143, "logps/rejected": -1.1758826971054077, "loss": 0.4871, "rewards/accuracies": 0.8125, "rewards/chosen": -1.381490707397461, "rewards/margins": 0.8018454313278198, "rewards/rejected": -2.1833362579345703, "step": 285 }, { "epoch": 0.5788423153692615, "grad_norm": 28.66415855277427, "learning_rate": 4.512085501204253e-07, "logits/chosen": 0.7206050157546997, "logits/rejected": 0.7634114623069763, "logps/chosen": -1.116304636001587, "logps/rejected": -1.1659752130508423, "loss": 0.553, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.4132583141326904, "rewards/margins": 0.657631516456604, "rewards/rejected": -2.070889711380005, "step": 290 }, { "epoch": 0.5888223552894212, "grad_norm": 17.845472804681062, "learning_rate": 4.338718048714387e-07, "logits/chosen": 0.6159842610359192, "logits/rejected": 0.6279784440994263, "logps/chosen": -1.0876820087432861, "logps/rejected": -1.1220331192016602, "loss": 0.5134, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -1.233598232269287, "rewards/margins": 0.5510263442993164, "rewards/rejected": -1.784624695777893, "step": 295 }, { "epoch": 0.5988023952095808, "grad_norm": 22.693877555997958, "learning_rate": 4.166156266419489e-07, "logits/chosen": 0.6069675087928772, "logits/rejected": 0.6444044709205627, "logps/chosen": -1.0315536260604858, "logps/rejected": -1.1916632652282715, "loss": 0.5015, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -1.4063842296600342, "rewards/margins": 0.8028200268745422, "rewards/rejected": -2.2092041969299316, "step": 300 }, { "epoch": 0.6087824351297405, "grad_norm": 29.176362707547757, "learning_rate": 3.9946103942701775e-07, "logits/chosen": 0.6731375455856323, "logits/rejected": 0.6558948755264282, "logps/chosen": -1.0202916860580444, "logps/rejected": -1.1923191547393799, "loss": 0.4922, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -1.134216547012329, "rewards/margins": 0.8526105880737305, "rewards/rejected": -1.9868271350860596, "step": 305 }, { "epoch": 0.6187624750499002, "grad_norm": 16.68482157863775, "learning_rate": 3.8242894344870495e-07, "logits/chosen": 0.6814749836921692, "logits/rejected": 0.6544811129570007, "logps/chosen": -1.1450976133346558, "logps/rejected": -1.2438080310821533, "loss": 0.518, "rewards/accuracies": 0.875, "rewards/chosen": -1.6444076299667358, "rewards/margins": 0.7449675798416138, "rewards/rejected": -2.3893752098083496, "step": 310 }, { "epoch": 0.6287425149700598, "grad_norm": 24.235832517472502, "learning_rate": 3.6554008969236715e-07, "logits/chosen": 0.6473463773727417, "logits/rejected": 0.6526838541030884, "logps/chosen": -1.1155353784561157, "logps/rejected": -1.2412601709365845, "loss": 0.5036, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -1.5284216403961182, "rewards/margins": 0.7738613486289978, "rewards/rejected": -2.3022828102111816, "step": 315 }, { "epoch": 0.6387225548902196, "grad_norm": 14.637858980431309, "learning_rate": 3.488150546247778e-07, "logits/chosen": 0.641592264175415, "logits/rejected": 0.6488014459609985, "logps/chosen": -1.0486619472503662, "logps/rejected": -1.117785930633545, "loss": 0.5076, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -1.3681632280349731, "rewards/margins": 0.7144045829772949, "rewards/rejected": -2.0825679302215576, "step": 320 }, { "epoch": 0.6487025948103793, "grad_norm": 14.611695255214977, "learning_rate": 3.3227421512487255e-07, "logits/chosen": 0.6020836234092712, "logits/rejected": 0.5903235673904419, "logps/chosen": -1.0694186687469482, "logps/rejected": -1.1450960636138916, "loss": 0.4988, "rewards/accuracies": 0.75, "rewards/chosen": -1.495102047920227, "rewards/margins": 0.6063732504844666, "rewards/rejected": -2.101475238800049, "step": 325 }, { "epoch": 0.6586826347305389, "grad_norm": 53.83501877641933, "learning_rate": 3.15937723657661e-07, "logits/chosen": 0.5937372446060181, "logits/rejected": 0.5983537435531616, "logps/chosen": -1.0911985635757446, "logps/rejected": -1.1970782279968262, "loss": 0.5292, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -1.5829055309295654, "rewards/margins": 0.6139482259750366, "rewards/rejected": -2.1968536376953125, "step": 330 }, { "epoch": 0.6686626746506986, "grad_norm": 22.263382020354737, "learning_rate": 2.9982548372155256e-07, "logits/chosen": 0.5615344047546387, "logits/rejected": 0.5818489789962769, "logps/chosen": -1.0336883068084717, "logps/rejected": -1.1681592464447021, "loss": 0.5036, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -1.2369213104248047, "rewards/margins": 1.0152759552001953, "rewards/rejected": -2.252197265625, "step": 335 }, { "epoch": 0.6786427145708582, "grad_norm": 28.648845585620847, "learning_rate": 2.8395712559900874e-07, "logits/chosen": 0.5865073204040527, "logits/rejected": 0.5887445211410522, "logps/chosen": -1.2165284156799316, "logps/rejected": -1.285627007484436, "loss": 0.5287, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.876123070716858, "rewards/margins": 0.8445480465888977, "rewards/rejected": -2.7206711769104004, "step": 340 }, { "epoch": 0.688622754491018, "grad_norm": 27.7967265314739, "learning_rate": 2.683519824400692e-07, "logits/chosen": 0.6274746656417847, "logits/rejected": 0.6499799489974976, "logps/chosen": -1.1229360103607178, "logps/rejected": -1.2630244493484497, "loss": 0.5228, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.6735212802886963, "rewards/margins": 0.8694586753845215, "rewards/rejected": -2.5429797172546387, "step": 345 }, { "epoch": 0.6986027944111777, "grad_norm": 22.50466227883414, "learning_rate": 2.530290667078846e-07, "logits/chosen": 0.620844841003418, "logits/rejected": 0.6195460557937622, "logps/chosen": -1.1165497303009033, "logps/rejected": -1.1850183010101318, "loss": 0.5019, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -1.5181663036346436, "rewards/margins": 0.7825000286102295, "rewards/rejected": -2.300666332244873, "step": 350 }, { "epoch": 0.7085828343313373, "grad_norm": 21.26239260918309, "learning_rate": 2.380070470149605e-07, "logits/chosen": 0.6597515940666199, "logits/rejected": 0.6680246591567993, "logps/chosen": -1.1044973134994507, "logps/rejected": -1.2354698181152344, "loss": 0.4914, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -1.4524997472763062, "rewards/margins": 1.0394423007965088, "rewards/rejected": -2.4919424057006836, "step": 355 }, { "epoch": 0.718562874251497, "grad_norm": 30.986440285036473, "learning_rate": 2.23304225378328e-07, "logits/chosen": 0.5663326978683472, "logits/rejected": 0.5595775842666626, "logps/chosen": -1.2192121744155884, "logps/rejected": -1.2986266613006592, "loss": 0.4991, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -1.6858431100845337, "rewards/margins": 1.0494027137756348, "rewards/rejected": -2.735245704650879, "step": 360 }, { "epoch": 0.7285429141716567, "grad_norm": 31.154865644231116, "learning_rate": 2.0893851492135532e-07, "logits/chosen": 0.5782791376113892, "logits/rejected": 0.5733534097671509, "logps/chosen": -1.108449935913086, "logps/rejected": -1.2320600748062134, "loss": 0.4969, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -1.4567809104919434, "rewards/margins": 1.0569483041763306, "rewards/rejected": -2.5137288570404053, "step": 365 }, { "epoch": 0.7385229540918163, "grad_norm": 25.665361579684205, "learning_rate": 1.9492741804936618e-07, "logits/chosen": 0.5590307116508484, "logits/rejected": 0.5289372205734253, "logps/chosen": -1.135460615158081, "logps/rejected": -1.1982572078704834, "loss": 0.5123, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -1.5751354694366455, "rewards/margins": 0.7708719968795776, "rewards/rejected": -2.346007823944092, "step": 370 }, { "epoch": 0.7485029940119761, "grad_norm": 13.466456049238714, "learning_rate": 1.812880051256551e-07, "logits/chosen": 0.5437551736831665, "logits/rejected": 0.540005087852478, "logps/chosen": -1.1057159900665283, "logps/rejected": -1.1711170673370361, "loss": 0.4956, "rewards/accuracies": 0.6875, "rewards/chosen": -1.607832670211792, "rewards/margins": 0.5974467396736145, "rewards/rejected": -2.205279588699341, "step": 375 }, { "epoch": 0.7584830339321357, "grad_norm": 22.830693166932168, "learning_rate": 1.6803689367387918e-07, "logits/chosen": 0.5521084666252136, "logits/rejected": 0.563881516456604, "logps/chosen": -1.0188974142074585, "logps/rejected": -1.0934759378433228, "loss": 0.5364, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": -1.6915432214736938, "rewards/margins": 0.41655927896499634, "rewards/rejected": -2.108102560043335, "step": 380 }, { "epoch": 0.7684630738522954, "grad_norm": 16.717527209109356, "learning_rate": 1.551902281321651e-07, "logits/chosen": 0.6741904020309448, "logits/rejected": 0.6558393836021423, "logps/chosen": -1.153136134147644, "logps/rejected": -1.2626793384552002, "loss": 0.4885, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -1.4042468070983887, "rewards/margins": 1.1682214736938477, "rewards/rejected": -2.5724682807922363, "step": 385 }, { "epoch": 0.7784431137724551, "grad_norm": 21.35977175588003, "learning_rate": 1.4276366018359842e-07, "logits/chosen": 0.5667208433151245, "logits/rejected": 0.5196897387504578, "logps/chosen": -1.1266697645187378, "logps/rejected": -1.2171615362167358, "loss": 0.5164, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -1.5344607830047607, "rewards/margins": 0.9302822947502136, "rewards/rejected": -2.464743137359619, "step": 390 }, { "epoch": 0.7884231536926147, "grad_norm": 21.011004494402407, "learning_rate": 1.3077232968705805e-07, "logits/chosen": 0.581541121006012, "logits/rejected": 0.5735315084457397, "logps/chosen": -1.1142101287841797, "logps/rejected": -1.204474687576294, "loss": 0.4938, "rewards/accuracies": 0.8125, "rewards/chosen": -1.4655622243881226, "rewards/margins": 0.7967506647109985, "rewards/rejected": -2.262312650680542, "step": 395 }, { "epoch": 0.7984031936127745, "grad_norm": 21.73473549415687, "learning_rate": 1.192308462316317e-07, "logits/chosen": 0.49174603819847107, "logits/rejected": 0.5455900430679321, "logps/chosen": -1.1133276224136353, "logps/rejected": -1.2326544523239136, "loss": 0.5169, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -1.5614674091339111, "rewards/margins": 0.7484425902366638, "rewards/rejected": -2.309910297393799, "step": 400 }, { "epoch": 0.8083832335329342, "grad_norm": 18.213898305037567, "learning_rate": 1.0815327133708013e-07, "logits/chosen": 0.5780737400054932, "logits/rejected": 0.6099402904510498, "logps/chosen": -1.0353389978408813, "logps/rejected": -1.1462222337722778, "loss": 0.5117, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -1.678412675857544, "rewards/margins": 0.7521916031837463, "rewards/rejected": -2.4306042194366455, "step": 405 }, { "epoch": 0.8183632734530938, "grad_norm": 14.585406444054382, "learning_rate": 9.755310132204297e-08, "logits/chosen": 0.5654971599578857, "logits/rejected": 0.5592461824417114, "logps/chosen": -1.0622889995574951, "logps/rejected": -1.2071908712387085, "loss": 0.5324, "rewards/accuracies": 0.75, "rewards/chosen": -1.5269814729690552, "rewards/margins": 1.1922402381896973, "rewards/rejected": -2.719221591949463, "step": 410 }, { "epoch": 0.8283433133732535, "grad_norm": 23.842758561266724, "learning_rate": 8.744325086085247e-08, "logits/chosen": 0.5762808322906494, "logits/rejected": 0.5668230652809143, "logps/chosen": -1.1735143661499023, "logps/rejected": -1.2483824491500854, "loss": 0.504, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -1.7859411239624023, "rewards/margins": 0.8882120847702026, "rewards/rejected": -2.6741533279418945, "step": 415 }, { "epoch": 0.8383233532934131, "grad_norm": 41.62435891120011, "learning_rate": 7.783603724899257e-08, "logits/chosen": 0.5172094106674194, "logits/rejected": 0.5516868829727173, "logps/chosen": -1.145578145980835, "logps/rejected": -1.248124599456787, "loss": 0.5313, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -1.742923378944397, "rewards/margins": 0.8347417712211609, "rewards/rejected": -2.577665090560913, "step": 420 }, { "epoch": 0.8483033932135728, "grad_norm": 26.49206528554136, "learning_rate": 6.874316539637126e-08, "logits/chosen": 0.5917887091636658, "logits/rejected": 0.6128644943237305, "logps/chosen": -1.0141854286193848, "logps/rejected": -1.1528804302215576, "loss": 0.4854, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -1.2433063983917236, "rewards/margins": 0.826849639415741, "rewards/rejected": -2.0701558589935303, "step": 425 }, { "epoch": 0.8582834331337326, "grad_norm": 14.052261790037898, "learning_rate": 6.017571356669182e-08, "logits/chosen": 0.5848813652992249, "logits/rejected": 0.5729445219039917, "logps/chosen": -1.1584407091140747, "logps/rejected": -1.304329752922058, "loss": 0.5097, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -1.7920862436294556, "rewards/margins": 0.670229971408844, "rewards/rejected": -2.4623162746429443, "step": 430 }, { "epoch": 0.8682634730538922, "grad_norm": 18.538952507210343, "learning_rate": 5.2144119880293544e-08, "logits/chosen": 0.5895706415176392, "logits/rejected": 0.5678045749664307, "logps/chosen": -1.075687050819397, "logps/rejected": -1.2755447626113892, "loss": 0.5187, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -1.3742207288742065, "rewards/margins": 1.1585171222686768, "rewards/rejected": -2.532737970352173, "step": 435 }, { "epoch": 0.8782435129740519, "grad_norm": 17.852951128940067, "learning_rate": 4.465816959691149e-08, "logits/chosen": 0.5652838945388794, "logits/rejected": 0.6052907705307007, "logps/chosen": -1.2037353515625, "logps/rejected": -1.3117984533309937, "loss": 0.5009, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -1.7123981714248657, "rewards/margins": 1.019058108329773, "rewards/rejected": -2.7314560413360596, "step": 440 }, { "epoch": 0.8882235528942116, "grad_norm": 14.249216111089481, "learning_rate": 3.7726983193843485e-08, "logits/chosen": 0.517819344997406, "logits/rejected": 0.5205559730529785, "logps/chosen": -1.0231516361236572, "logps/rejected": -1.148457646369934, "loss": 0.5199, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -1.4192359447479248, "rewards/margins": 0.889240562915802, "rewards/rejected": -2.308476448059082, "step": 445 }, { "epoch": 0.8982035928143712, "grad_norm": 17.926055468482602, "learning_rate": 3.135900525405427e-08, "logits/chosen": 0.5483254194259644, "logits/rejected": 0.5569725036621094, "logps/chosen": -1.160649299621582, "logps/rejected": -1.2841194868087769, "loss": 0.5171, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.7122529745101929, "rewards/margins": 0.9245160222053528, "rewards/rejected": -2.6367690563201904, "step": 450 }, { "epoch": 0.908183632734531, "grad_norm": 15.183540302368447, "learning_rate": 2.5561994177751732e-08, "logits/chosen": 0.6299312114715576, "logits/rejected": 0.6476989984512329, "logps/chosen": -1.0939618349075317, "logps/rejected": -1.187743902206421, "loss": 0.4643, "rewards/accuracies": 0.75, "rewards/chosen": -1.5630865097045898, "rewards/margins": 0.8125492334365845, "rewards/rejected": -2.3756356239318848, "step": 455 }, { "epoch": 0.9181636726546906, "grad_norm": 15.769731536967763, "learning_rate": 2.0343012729971243e-08, "logits/chosen": 0.5560911893844604, "logits/rejected": 0.563130259513855, "logps/chosen": -1.1199856996536255, "logps/rejected": -1.2025185823440552, "loss": 0.4948, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -1.6438310146331787, "rewards/margins": 0.8590097427368164, "rewards/rejected": -2.502840518951416, "step": 460 }, { "epoch": 0.9281437125748503, "grad_norm": 21.7191545301948, "learning_rate": 1.570841943568446e-08, "logits/chosen": 0.6068538427352905, "logits/rejected": 0.6409121751785278, "logps/chosen": -1.211014747619629, "logps/rejected": -1.3489148616790771, "loss": 0.4823, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -1.6730806827545166, "rewards/margins": 1.099481463432312, "rewards/rejected": -2.772562026977539, "step": 465 }, { "epoch": 0.93812375249501, "grad_norm": 23.87949640105234, "learning_rate": 1.166386083291604e-08, "logits/chosen": 0.5642676949501038, "logits/rejected": 0.5937134027481079, "logps/chosen": -1.0865347385406494, "logps/rejected": -1.2261168956756592, "loss": 0.5012, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -1.5355206727981567, "rewards/margins": 0.9762239456176758, "rewards/rejected": -2.511744260787964, "step": 470 }, { "epoch": 0.9481037924151696, "grad_norm": 11.821044673852356, "learning_rate": 8.214264593307096e-09, "logits/chosen": 0.5303646922111511, "logits/rejected": 0.550706684589386, "logps/chosen": -1.212477207183838, "logps/rejected": -1.27964186668396, "loss": 0.4881, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.8850059509277344, "rewards/margins": 0.7182316780090332, "rewards/rejected": -2.603238105773926, "step": 475 }, { "epoch": 0.9580838323353293, "grad_norm": 18.686466292842862, "learning_rate": 5.3638335185058335e-09, "logits/chosen": 0.5736523866653442, "logits/rejected": 0.625065267086029, "logps/chosen": -1.2084397077560425, "logps/rejected": -1.3002406358718872, "loss": 0.5266, "rewards/accuracies": 0.8125, "rewards/chosen": -1.7023632526397705, "rewards/margins": 0.9809869527816772, "rewards/rejected": -2.6833503246307373, "step": 480 }, { "epoch": 0.9680638722554891, "grad_norm": 26.284460492309865, "learning_rate": 3.116040419701815e-09, "logits/chosen": 0.5889401435852051, "logits/rejected": 0.5715131759643555, "logps/chosen": -1.1137585639953613, "logps/rejected": -1.2124450206756592, "loss": 0.5171, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -1.5402872562408447, "rewards/margins": 0.7413524389266968, "rewards/rejected": -2.281639575958252, "step": 485 }, { "epoch": 0.9780439121756487, "grad_norm": 30.03419204450396, "learning_rate": 1.4736238865398765e-09, "logits/chosen": 0.6326584815979004, "logits/rejected": 0.6649195551872253, "logps/chosen": -1.1648736000061035, "logps/rejected": -1.2586228847503662, "loss": 0.519, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.403749942779541, "rewards/margins": 1.0453355312347412, "rewards/rejected": -2.4490854740142822, "step": 490 }, { "epoch": 0.9880239520958084, "grad_norm": 33.25313115299262, "learning_rate": 4.3858495057080836e-10, "logits/chosen": 0.5985925197601318, "logits/rejected": 0.5628663301467896, "logps/chosen": -1.2969096899032593, "logps/rejected": -1.4230639934539795, "loss": 0.4764, "rewards/accuracies": 0.8125, "rewards/chosen": -1.819708228111267, "rewards/margins": 1.287508487701416, "rewards/rejected": -3.1072165966033936, "step": 495 }, { "epoch": 0.998003992015968, "grad_norm": 30.55644687750284, "learning_rate": 1.2184647302626582e-11, "logits/chosen": 0.5701574087142944, "logits/rejected": 0.5943929553031921, "logps/chosen": -1.1318881511688232, "logps/rejected": -1.2142232656478882, "loss": 0.4976, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -1.6143200397491455, "rewards/margins": 0.6937130093574524, "rewards/rejected": -2.3080332279205322, "step": 500 }, { "epoch": 1.0, "step": 501, "total_flos": 0.0, "train_loss": 0.559710203114146, "train_runtime": 2572.2381, "train_samples_per_second": 24.931, "train_steps_per_second": 0.195 } ], "logging_steps": 5, "max_steps": 501, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 101, "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": 4, "trial_name": null, "trial_params": null }