| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 478, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 8.432772549922241, | |
| "learning_rate": 1.0416666666666666e-08, | |
| "logits/chosen": -2.5992650985717773, | |
| "logits/rejected": -2.567516326904297, | |
| "logps/chosen": -272.1844482421875, | |
| "logps/rejected": -362.26898193359375, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 9.993362324491976, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -2.4130637645721436, | |
| "logits/rejected": -2.3763909339904785, | |
| "logps/chosen": -268.1092834472656, | |
| "logps/rejected": -252.8538360595703, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.00011327523679938167, | |
| "rewards/margins": 0.00018297109636478126, | |
| "rewards/rejected": -6.969591049710289e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 8.212783868686264, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -2.472649097442627, | |
| "logits/rejected": -2.4103596210479736, | |
| "logps/chosen": -283.3070068359375, | |
| "logps/rejected": -297.09979248046875, | |
| "loss": 0.6928, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.001224780222401023, | |
| "rewards/margins": 0.0005858406075276434, | |
| "rewards/rejected": 0.0006389396148733795, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 7.684799704050697, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -2.508185386657715, | |
| "logits/rejected": -2.415645122528076, | |
| "logps/chosen": -301.5997619628906, | |
| "logps/rejected": -265.80426025390625, | |
| "loss": 0.6906, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.0062199728563427925, | |
| "rewards/margins": 0.0035330094397068024, | |
| "rewards/rejected": 0.002686963649466634, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 7.49113948544429, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -2.404700756072998, | |
| "logits/rejected": -2.350811243057251, | |
| "logps/chosen": -268.32647705078125, | |
| "logps/rejected": -247.6831817626953, | |
| "loss": 0.6849, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.0182647742331028, | |
| "rewards/margins": 0.020485591143369675, | |
| "rewards/rejected": -0.0022208169102668762, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 7.812080735900241, | |
| "learning_rate": 4.999733114418725e-07, | |
| "logits/chosen": -2.3245081901550293, | |
| "logits/rejected": -2.2887587547302246, | |
| "logps/chosen": -280.1948547363281, | |
| "logps/rejected": -293.09405517578125, | |
| "loss": 0.6753, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.022629689425230026, | |
| "rewards/margins": 0.031411103904247284, | |
| "rewards/rejected": -0.00878141075372696, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 8.653367610484782, | |
| "learning_rate": 4.990398100856366e-07, | |
| "logits/chosen": -2.3789138793945312, | |
| "logits/rejected": -2.3128437995910645, | |
| "logps/chosen": -271.4405822753906, | |
| "logps/rejected": -303.5579528808594, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.029351558536291122, | |
| "rewards/margins": 0.06681646406650543, | |
| "rewards/rejected": -0.03746490180492401, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 9.793859330498844, | |
| "learning_rate": 4.967775735898179e-07, | |
| "logits/chosen": -2.1905629634857178, | |
| "logits/rejected": -2.1994009017944336, | |
| "logps/chosen": -267.55340576171875, | |
| "logps/rejected": -273.4862976074219, | |
| "loss": 0.6361, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.0027915718965232372, | |
| "rewards/margins": 0.1346043348312378, | |
| "rewards/rejected": -0.13739590346813202, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 13.837088401780129, | |
| "learning_rate": 4.931986719649298e-07, | |
| "logits/chosen": -2.3278985023498535, | |
| "logits/rejected": -2.243424892425537, | |
| "logps/chosen": -337.2379455566406, | |
| "logps/rejected": -292.844970703125, | |
| "loss": 0.6223, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.09643656760454178, | |
| "rewards/margins": 0.17360267043113708, | |
| "rewards/rejected": -0.27003923058509827, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 17.06156274259609, | |
| "learning_rate": 4.883222001996351e-07, | |
| "logits/chosen": -2.1236023902893066, | |
| "logits/rejected": -2.0597236156463623, | |
| "logps/chosen": -276.4493103027344, | |
| "logps/rejected": -299.7818908691406, | |
| "loss": 0.5987, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.15940961241722107, | |
| "rewards/margins": 0.3067048490047455, | |
| "rewards/rejected": -0.46611452102661133, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 15.017538794455808, | |
| "learning_rate": 4.821741763807186e-07, | |
| "logits/chosen": -2.0527923107147217, | |
| "logits/rejected": -1.9835008382797241, | |
| "logps/chosen": -294.0035400390625, | |
| "logps/rejected": -321.8211669921875, | |
| "loss": 0.5965, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.27848348021507263, | |
| "rewards/margins": 0.3403889834880829, | |
| "rewards/rejected": -0.6188725233078003, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_logits/chosen": -2.1194002628326416, | |
| "eval_logits/rejected": -2.0640361309051514, | |
| "eval_logps/chosen": -308.44342041015625, | |
| "eval_logps/rejected": -344.93780517578125, | |
| "eval_loss": 0.6008175015449524, | |
| "eval_rewards/accuracies": 0.71484375, | |
| "eval_rewards/chosen": -0.43487486243247986, | |
| "eval_rewards/margins": 0.3607807159423828, | |
| "eval_rewards/rejected": -0.7956556081771851, | |
| "eval_runtime": 39.9329, | |
| "eval_samples_per_second": 50.084, | |
| "eval_steps_per_second": 0.801, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 13.512613063149377, | |
| "learning_rate": 4.747874028753375e-07, | |
| "logits/chosen": -2.1186444759368896, | |
| "logits/rejected": -1.960219383239746, | |
| "logps/chosen": -346.79327392578125, | |
| "logps/rejected": -330.68634033203125, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.373442143201828, | |
| "rewards/margins": 0.33256274461746216, | |
| "rewards/rejected": -0.7060048580169678, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 16.771221195438027, | |
| "learning_rate": 4.662012913161997e-07, | |
| "logits/chosen": -1.878488302230835, | |
| "logits/rejected": -1.82696533203125, | |
| "logps/chosen": -322.34173583984375, | |
| "logps/rejected": -339.63104248046875, | |
| "loss": 0.5803, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.3693729043006897, | |
| "rewards/margins": 0.3852415680885315, | |
| "rewards/rejected": -0.754614531993866, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 14.00034800920836, | |
| "learning_rate": 4.5646165232345103e-07, | |
| "logits/chosen": -1.8512026071548462, | |
| "logits/rejected": -1.7661195993423462, | |
| "logps/chosen": -322.70599365234375, | |
| "logps/rejected": -353.8482360839844, | |
| "loss": 0.5644, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.3978124260902405, | |
| "rewards/margins": 0.42583903670310974, | |
| "rewards/rejected": -0.8236514925956726, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 15.3625036150752, | |
| "learning_rate": 4.456204510851956e-07, | |
| "logits/chosen": -1.7981727123260498, | |
| "logits/rejected": -1.7398284673690796, | |
| "logps/chosen": -359.68994140625, | |
| "logps/rejected": -386.440185546875, | |
| "loss": 0.5654, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.40198105573654175, | |
| "rewards/margins": 0.43909168243408203, | |
| "rewards/rejected": -0.841072678565979, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 20.80601584306436, | |
| "learning_rate": 4.337355301007335e-07, | |
| "logits/chosen": -1.7028295993804932, | |
| "logits/rejected": -1.5830708742141724, | |
| "logps/chosen": -345.12286376953125, | |
| "logps/rejected": -376.55859375, | |
| "loss": 0.5775, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.5939025282859802, | |
| "rewards/margins": 0.37951546907424927, | |
| "rewards/rejected": -0.9734179377555847, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 18.137407218927724, | |
| "learning_rate": 4.2087030056579986e-07, | |
| "logits/chosen": -1.6537895202636719, | |
| "logits/rejected": -1.4418971538543701, | |
| "logps/chosen": -318.0480651855469, | |
| "logps/rejected": -349.61431884765625, | |
| "loss": 0.5693, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.43722066283226013, | |
| "rewards/margins": 0.5587003827095032, | |
| "rewards/rejected": -0.9959210157394409, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 16.981581647441832, | |
| "learning_rate": 4.070934040463998e-07, | |
| "logits/chosen": -1.5120352506637573, | |
| "logits/rejected": -1.4046074151992798, | |
| "logps/chosen": -306.04840087890625, | |
| "logps/rejected": -330.2176208496094, | |
| "loss": 0.5659, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.538715124130249, | |
| "rewards/margins": 0.42584919929504395, | |
| "rewards/rejected": -0.9645644426345825, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 14.617848943306955, | |
| "learning_rate": 3.9247834624635404e-07, | |
| "logits/chosen": -1.3031604290008545, | |
| "logits/rejected": -1.1622366905212402, | |
| "logps/chosen": -317.8174743652344, | |
| "logps/rejected": -331.2264404296875, | |
| "loss": 0.5424, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.6144936084747314, | |
| "rewards/margins": 0.4146398603916168, | |
| "rewards/rejected": -1.0291334390640259, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 18.83954708764831, | |
| "learning_rate": 3.7710310482256523e-07, | |
| "logits/chosen": -1.3724639415740967, | |
| "logits/rejected": -1.2839093208312988, | |
| "logps/chosen": -320.2606506347656, | |
| "logps/rejected": -353.50677490234375, | |
| "loss": 0.5567, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.47087445855140686, | |
| "rewards/margins": 0.4510224461555481, | |
| "rewards/rejected": -0.9218968152999878, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 27.897186532435434, | |
| "learning_rate": 3.610497133404795e-07, | |
| "logits/chosen": -1.14837646484375, | |
| "logits/rejected": -1.072177767753601, | |
| "logps/chosen": -318.0636291503906, | |
| "logps/rejected": -362.46044921875, | |
| "loss": 0.5688, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.5620325207710266, | |
| "rewards/margins": 0.5343230962753296, | |
| "rewards/rejected": -1.096355676651001, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "eval_logits/chosen": -1.2653636932373047, | |
| "eval_logits/rejected": -1.1455148458480835, | |
| "eval_logps/chosen": -328.60369873046875, | |
| "eval_logps/rejected": -382.0739440917969, | |
| "eval_loss": 0.558937132358551, | |
| "eval_rewards/accuracies": 0.73828125, | |
| "eval_rewards/chosen": -0.6364771723747253, | |
| "eval_rewards/margins": 0.5305400490760803, | |
| "eval_rewards/rejected": -1.1670172214508057, | |
| "eval_runtime": 39.8802, | |
| "eval_samples_per_second": 50.15, | |
| "eval_steps_per_second": 0.802, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 15.625329725139888, | |
| "learning_rate": 3.4440382358952115e-07, | |
| "logits/chosen": -1.0910618305206299, | |
| "logits/rejected": -0.9091793298721313, | |
| "logps/chosen": -360.6944885253906, | |
| "logps/rejected": -366.00146484375, | |
| "loss": 0.5724, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.6661044359207153, | |
| "rewards/margins": 0.46460071206092834, | |
| "rewards/rejected": -1.1307051181793213, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 19.24668760178638, | |
| "learning_rate": 3.272542485937368e-07, | |
| "logits/chosen": -0.9850679636001587, | |
| "logits/rejected": -0.7914190292358398, | |
| "logps/chosen": -309.50775146484375, | |
| "logps/rejected": -338.098876953125, | |
| "loss": 0.5582, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.44118037819862366, | |
| "rewards/margins": 0.524976372718811, | |
| "rewards/rejected": -0.9661566019058228, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 19.97223623454459, | |
| "learning_rate": 3.096924887558854e-07, | |
| "logits/chosen": -0.4917120039463043, | |
| "logits/rejected": -0.298466295003891, | |
| "logps/chosen": -313.9906005859375, | |
| "logps/rejected": -375.44989013671875, | |
| "loss": 0.5541, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.6258713603019714, | |
| "rewards/margins": 0.6441494226455688, | |
| "rewards/rejected": -1.2700207233428955, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 21.91087703960587, | |
| "learning_rate": 2.9181224366319943e-07, | |
| "logits/chosen": 0.09583790600299835, | |
| "logits/rejected": 0.32567495107650757, | |
| "logps/chosen": -339.2015380859375, | |
| "logps/rejected": -384.8148498535156, | |
| "loss": 0.521, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.7344726324081421, | |
| "rewards/margins": 0.6358748078346252, | |
| "rewards/rejected": -1.370347499847412, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 19.954418058737403, | |
| "learning_rate": 2.7370891215954565e-07, | |
| "logits/chosen": 0.5116527676582336, | |
| "logits/rejected": 0.8739731907844543, | |
| "logps/chosen": -363.95684814453125, | |
| "logps/rejected": -394.17877197265625, | |
| "loss": 0.5327, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.6146451830863953, | |
| "rewards/margins": 0.7260497808456421, | |
| "rewards/rejected": -1.3406950235366821, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 29.50918535565258, | |
| "learning_rate": 2.55479083351317e-07, | |
| "logits/chosen": 0.7538167834281921, | |
| "logits/rejected": 1.1193482875823975, | |
| "logps/chosen": -365.6874694824219, | |
| "logps/rejected": -389.31396484375, | |
| "loss": 0.5412, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.6788283586502075, | |
| "rewards/margins": 0.6221181154251099, | |
| "rewards/rejected": -1.3009464740753174, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 24.53746609446516, | |
| "learning_rate": 2.3722002126275822e-07, | |
| "logits/chosen": 1.132846474647522, | |
| "logits/rejected": 1.5623472929000854, | |
| "logps/chosen": -345.86700439453125, | |
| "logps/rejected": -378.31719970703125, | |
| "loss": 0.5414, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.6558988690376282, | |
| "rewards/margins": 0.5384365320205688, | |
| "rewards/rejected": -1.1943353414535522, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 27.660029561250692, | |
| "learning_rate": 2.19029145890313e-07, | |
| "logits/chosen": 1.715608835220337, | |
| "logits/rejected": 2.1731343269348145, | |
| "logps/chosen": -352.11346435546875, | |
| "logps/rejected": -406.60772705078125, | |
| "loss": 0.5375, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8623917698860168, | |
| "rewards/margins": 0.7531214952468872, | |
| "rewards/rejected": -1.6155132055282593, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 25.659061335686694, | |
| "learning_rate": 2.0100351342479216e-07, | |
| "logits/chosen": 1.6510066986083984, | |
| "logits/rejected": 1.7990186214447021, | |
| "logps/chosen": -337.61016845703125, | |
| "logps/rejected": -396.5470886230469, | |
| "loss": 0.5336, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.8643373250961304, | |
| "rewards/margins": 0.6685428023338318, | |
| "rewards/rejected": -1.5328800678253174, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 22.694419610449454, | |
| "learning_rate": 1.8323929841460178e-07, | |
| "logits/chosen": 1.5950249433517456, | |
| "logits/rejected": 2.302058696746826, | |
| "logps/chosen": -376.0797424316406, | |
| "logps/rejected": -401.6100769042969, | |
| "loss": 0.5121, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.7354680299758911, | |
| "rewards/margins": 0.6922025680541992, | |
| "rewards/rejected": -1.4276707172393799, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_logits/chosen": 1.7225008010864258, | |
| "eval_logits/rejected": 2.1388766765594482, | |
| "eval_logps/chosen": -334.2620849609375, | |
| "eval_logps/rejected": -418.3771667480469, | |
| "eval_loss": 0.5288156270980835, | |
| "eval_rewards/accuracies": 0.76171875, | |
| "eval_rewards/chosen": -0.6930612325668335, | |
| "eval_rewards/margins": 0.8369885683059692, | |
| "eval_rewards/rejected": -1.5300499200820923, | |
| "eval_runtime": 39.9288, | |
| "eval_samples_per_second": 50.089, | |
| "eval_steps_per_second": 0.801, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 24.37280438119094, | |
| "learning_rate": 1.6583128063291573e-07, | |
| "logits/chosen": 2.1118528842926025, | |
| "logits/rejected": 2.5268707275390625, | |
| "logps/chosen": -376.37969970703125, | |
| "logps/rejected": -417.34869384765625, | |
| "loss": 0.507, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.778560996055603, | |
| "rewards/margins": 0.7997097969055176, | |
| "rewards/rejected": -1.5782709121704102, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 26.080136074985454, | |
| "learning_rate": 1.488723393865766e-07, | |
| "logits/chosen": 2.5625953674316406, | |
| "logits/rejected": 3.1481173038482666, | |
| "logps/chosen": -383.0509338378906, | |
| "logps/rejected": -411.533935546875, | |
| "loss": 0.5013, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.8686873316764832, | |
| "rewards/margins": 0.7724698781967163, | |
| "rewards/rejected": -1.6411571502685547, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 27.228237079063305, | |
| "learning_rate": 1.3245295796480788e-07, | |
| "logits/chosen": 2.7803778648376465, | |
| "logits/rejected": 3.247398853302002, | |
| "logps/chosen": -351.1916809082031, | |
| "logps/rejected": -420.370849609375, | |
| "loss": 0.5142, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.8757610321044922, | |
| "rewards/margins": 0.7058261632919312, | |
| "rewards/rejected": -1.5815874338150024, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 28.892127434127993, | |
| "learning_rate": 1.1666074087171627e-07, | |
| "logits/chosen": 2.754971504211426, | |
| "logits/rejected": 3.230527400970459, | |
| "logps/chosen": -377.8298645019531, | |
| "logps/rejected": -465.24761962890625, | |
| "loss": 0.5138, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.8463465571403503, | |
| "rewards/margins": 0.987470269203186, | |
| "rewards/rejected": -1.8338168859481812, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 26.281931375691812, | |
| "learning_rate": 1.0157994641835734e-07, | |
| "logits/chosen": 2.723754405975342, | |
| "logits/rejected": 3.361722230911255, | |
| "logps/chosen": -351.22900390625, | |
| "logps/rejected": -415.8351135253906, | |
| "loss": 0.4828, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.8969828486442566, | |
| "rewards/margins": 0.857469916343689, | |
| "rewards/rejected": -1.7544529438018799, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 24.286834238524502, | |
| "learning_rate": 8.729103716819111e-08, | |
| "logits/chosen": 2.8787496089935303, | |
| "logits/rejected": 3.6532554626464844, | |
| "logps/chosen": -402.9510192871094, | |
| "logps/rejected": -443.6593322753906, | |
| "loss": 0.5325, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.9707611203193665, | |
| "rewards/margins": 0.8318966627120972, | |
| "rewards/rejected": -1.8026577234268188, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 23.72120672745611, | |
| "learning_rate": 7.387025063449081e-08, | |
| "logits/chosen": 3.308849811553955, | |
| "logits/rejected": 3.967015504837036, | |
| "logps/chosen": -388.5707702636719, | |
| "logps/rejected": -417.2923889160156, | |
| "loss": 0.5145, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.0928517580032349, | |
| "rewards/margins": 0.6804816722869873, | |
| "rewards/rejected": -1.7733335494995117, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 24.194836344161388, | |
| "learning_rate": 6.138919252022435e-08, | |
| "logits/chosen": 3.4659945964813232, | |
| "logits/rejected": 3.6677188873291016, | |
| "logps/chosen": -360.42303466796875, | |
| "logps/rejected": -468.11322021484375, | |
| "loss": 0.5064, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.1740145683288574, | |
| "rewards/margins": 0.8953350186347961, | |
| "rewards/rejected": -2.069349765777588, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 33.960328274537595, | |
| "learning_rate": 4.991445467064689e-08, | |
| "logits/chosen": 3.0402557849884033, | |
| "logits/rejected": 3.3952622413635254, | |
| "logps/chosen": -395.9051208496094, | |
| "logps/rejected": -456.98162841796875, | |
| "loss": 0.5003, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.9203785061836243, | |
| "rewards/margins": 0.77605140209198, | |
| "rewards/rejected": -1.6964296102523804, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 24.96336693295718, | |
| "learning_rate": 3.9507259776993954e-08, | |
| "logits/chosen": 3.402864456176758, | |
| "logits/rejected": 3.9089291095733643, | |
| "logps/chosen": -373.3275146484375, | |
| "logps/rejected": -453.81109619140625, | |
| "loss": 0.5208, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.0015218257904053, | |
| "rewards/margins": 0.86052405834198, | |
| "rewards/rejected": -1.8620456457138062, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_logits/chosen": 2.9372177124023438, | |
| "eval_logits/rejected": 3.4323720932006836, | |
| "eval_logps/chosen": -352.0043029785156, | |
| "eval_logps/rejected": -445.87408447265625, | |
| "eval_loss": 0.5152841210365295, | |
| "eval_rewards/accuracies": 0.7578125, | |
| "eval_rewards/chosen": -0.8704833984375, | |
| "eval_rewards/margins": 0.9345353841781616, | |
| "eval_rewards/rejected": -1.8050185441970825, | |
| "eval_runtime": 39.8891, | |
| "eval_samples_per_second": 50.139, | |
| "eval_steps_per_second": 0.802, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 26.45544298088616, | |
| "learning_rate": 3.022313472693447e-08, | |
| "logits/chosen": 3.266558885574341, | |
| "logits/rejected": 4.045865535736084, | |
| "logps/chosen": -393.68505859375, | |
| "logps/rejected": -454.821044921875, | |
| "loss": 0.5226, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.9244238138198853, | |
| "rewards/margins": 0.9223299026489258, | |
| "rewards/rejected": -1.8467538356781006, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 24.753221828065943, | |
| "learning_rate": 2.2111614344599684e-08, | |
| "logits/chosen": 3.017789363861084, | |
| "logits/rejected": 3.619795322418213, | |
| "logps/chosen": -396.9893798828125, | |
| "logps/rejected": -455.4769592285156, | |
| "loss": 0.5062, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.9954174160957336, | |
| "rewards/margins": 0.8088730573654175, | |
| "rewards/rejected": -1.804290533065796, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 28.64072501651785, | |
| "learning_rate": 1.521597710086439e-08, | |
| "logits/chosen": 3.4113173484802246, | |
| "logits/rejected": 3.839292526245117, | |
| "logps/chosen": -386.08624267578125, | |
| "logps/rejected": -451.10894775390625, | |
| "loss": 0.4865, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -1.0454809665679932, | |
| "rewards/margins": 0.8698482513427734, | |
| "rewards/rejected": -1.9153292179107666, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 28.408098716357102, | |
| "learning_rate": 9.57301420397924e-09, | |
| "logits/chosen": 2.9448680877685547, | |
| "logits/rejected": 3.614654541015625, | |
| "logps/chosen": -380.95782470703125, | |
| "logps/rejected": -454.02191162109375, | |
| "loss": 0.5045, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.9367244839668274, | |
| "rewards/margins": 0.8637276887893677, | |
| "rewards/rejected": -1.8004519939422607, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 27.41806430030018, | |
| "learning_rate": 5.212833302556258e-09, | |
| "logits/chosen": 3.149013042449951, | |
| "logits/rejected": 3.4816536903381348, | |
| "logps/chosen": -401.301513671875, | |
| "logps/rejected": -495.34002685546875, | |
| "loss": 0.5059, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.096161127090454, | |
| "rewards/margins": 0.7614067792892456, | |
| "rewards/rejected": -1.8575680255889893, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 30.97252598966743, | |
| "learning_rate": 2.158697848236607e-09, | |
| "logits/chosen": 3.359788417816162, | |
| "logits/rejected": 3.880640745162964, | |
| "logps/chosen": -376.32879638671875, | |
| "logps/rejected": -425.67010498046875, | |
| "loss": 0.5099, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.0073617696762085, | |
| "rewards/margins": 0.8302923440933228, | |
| "rewards/rejected": -1.8376541137695312, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 23.032101494576157, | |
| "learning_rate": 4.269029751107489e-10, | |
| "logits/chosen": 3.195591688156128, | |
| "logits/rejected": 3.7321903705596924, | |
| "logps/chosen": -378.28631591796875, | |
| "logps/rejected": -458.43096923828125, | |
| "loss": 0.4995, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -1.0239256620407104, | |
| "rewards/margins": 0.7799959778785706, | |
| "rewards/rejected": -1.8039219379425049, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 478, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5581105443723032, | |
| "train_runtime": 5172.5891, | |
| "train_samples_per_second": 11.819, | |
| "train_steps_per_second": 0.092 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 478, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |