| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.9997605937275557, |
| "eval_steps": 500, |
| "global_step": 261, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0038305003591094086, |
| "grad_norm": 0.9418857097625732, |
| "learning_rate": 1.0000000000000002e-06, |
| "logits/chosen": -1.0421228408813477, |
| "logits/rejected": -1.0402593612670898, |
| "logps/chosen": -905.7581787109375, |
| "logps/rejected": -924.099853515625, |
| "loss": 0.6931, |
| "num_input_tokens_seen": 362496, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.007661000718218817, |
| "grad_norm": 1.0012181997299194, |
| "learning_rate": 2.0000000000000003e-06, |
| "logits/chosen": -1.0420148372650146, |
| "logits/rejected": -1.0282371044158936, |
| "logps/chosen": -898.2119140625, |
| "logps/rejected": -893.538818359375, |
| "loss": 0.6931, |
| "num_input_tokens_seen": 718144, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.011491501077328227, |
| "grad_norm": 0.9935199618339539, |
| "learning_rate": 3e-06, |
| "logits/chosen": -1.0312358140945435, |
| "logits/rejected": -1.023416519165039, |
| "logps/chosen": -918.4049072265625, |
| "logps/rejected": -903.5130615234375, |
| "loss": 0.7036, |
| "num_input_tokens_seen": 1075392, |
| "rewards/accuracies": 0.4375, |
| "rewards/chosen": -0.012404728680849075, |
| "rewards/margins": -0.017508292570710182, |
| "rewards/rejected": 0.0051035634241998196, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.015322001436437634, |
| "grad_norm": 0.9774389266967773, |
| "learning_rate": 4.000000000000001e-06, |
| "logits/chosen": -1.0343425273895264, |
| "logits/rejected": -1.036562204360962, |
| "logps/chosen": -896.435302734375, |
| "logps/rejected": -899.7354736328125, |
| "loss": 0.6913, |
| "num_input_tokens_seen": 1430080, |
| "rewards/accuracies": 0.5390625, |
| "rewards/chosen": -0.0023043155670166016, |
| "rewards/margins": 0.00819547288119793, |
| "rewards/rejected": -0.010499788448214531, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.019152501795547044, |
| "grad_norm": 0.9151395559310913, |
| "learning_rate": 5e-06, |
| "logits/chosen": -1.0331459045410156, |
| "logits/rejected": -1.0269978046417236, |
| "logps/chosen": -948.02978515625, |
| "logps/rejected": -949.6331787109375, |
| "loss": 0.6978, |
| "num_input_tokens_seen": 1786304, |
| "rewards/accuracies": 0.5, |
| "rewards/chosen": -0.003812384558841586, |
| "rewards/margins": -0.005953050684183836, |
| "rewards/rejected": 0.002140665426850319, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.022983002154656453, |
| "grad_norm": 0.9881334900856018, |
| "learning_rate": 6e-06, |
| "logits/chosen": -1.0443257093429565, |
| "logits/rejected": -1.0362162590026855, |
| "logps/chosen": -881.51611328125, |
| "logps/rejected": -884.4502563476562, |
| "loss": 0.6808, |
| "num_input_tokens_seen": 2132544, |
| "rewards/accuracies": 0.6015625, |
| "rewards/chosen": 0.005164409056305885, |
| "rewards/margins": 0.028560878708958626, |
| "rewards/rejected": -0.02339646965265274, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.02681350251376586, |
| "grad_norm": 0.9441390037536621, |
| "learning_rate": 7e-06, |
| "logits/chosen": -1.0445983409881592, |
| "logits/rejected": -1.0422747135162354, |
| "logps/chosen": -899.3116455078125, |
| "logps/rejected": -894.371337890625, |
| "loss": 0.6866, |
| "num_input_tokens_seen": 2488832, |
| "rewards/accuracies": 0.546875, |
| "rewards/chosen": -0.0004974845796823502, |
| "rewards/margins": 0.016697216778993607, |
| "rewards/rejected": -0.017194701358675957, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.03064400287287527, |
| "grad_norm": 1.0369701385498047, |
| "learning_rate": 8.000000000000001e-06, |
| "logits/chosen": -1.0434613227844238, |
| "logits/rejected": -1.0450818538665771, |
| "logps/chosen": -924.5731201171875, |
| "logps/rejected": -928.7008056640625, |
| "loss": 0.6897, |
| "num_input_tokens_seen": 2852544, |
| "rewards/accuracies": 0.515625, |
| "rewards/chosen": 0.007696032524108887, |
| "rewards/margins": 0.011814999394118786, |
| "rewards/rejected": -0.004118966404348612, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.03447450323198468, |
| "grad_norm": 0.9359210133552551, |
| "learning_rate": 9e-06, |
| "logits/chosen": -1.0368328094482422, |
| "logits/rejected": -1.0320663452148438, |
| "logps/chosen": -965.3587036132812, |
| "logps/rejected": -946.9695434570312, |
| "loss": 0.6901, |
| "num_input_tokens_seen": 3220608, |
| "rewards/accuracies": 0.5546875, |
| "rewards/chosen": 0.005610942840576172, |
| "rewards/margins": 0.009852622635662556, |
| "rewards/rejected": -0.004241681657731533, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.03830500359109409, |
| "grad_norm": 0.9655279517173767, |
| "learning_rate": 1e-05, |
| "logits/chosen": -1.0187113285064697, |
| "logits/rejected": -1.0166434049606323, |
| "logps/chosen": -953.6986083984375, |
| "logps/rejected": -927.5408935546875, |
| "loss": 0.691, |
| "num_input_tokens_seen": 3592128, |
| "rewards/accuracies": 0.53125, |
| "rewards/chosen": -0.0018309114966541529, |
| "rewards/margins": 0.008346128277480602, |
| "rewards/rejected": -0.010177039541304111, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0421355039502035, |
| "grad_norm": 0.9258586168289185, |
| "learning_rate": 9.999608360361114e-06, |
| "logits/chosen": -1.0451351404190063, |
| "logits/rejected": -1.0583577156066895, |
| "logps/chosen": -958.3239135742188, |
| "logps/rejected": -946.181640625, |
| "loss": 0.6915, |
| "num_input_tokens_seen": 3971392, |
| "rewards/accuracies": 0.5, |
| "rewards/chosen": 0.004320908337831497, |
| "rewards/margins": 0.008112478069961071, |
| "rewards/rejected": -0.003791570197790861, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.045966004309312906, |
| "grad_norm": 1.1154330968856812, |
| "learning_rate": 9.998433502797097e-06, |
| "logits/chosen": -1.0229029655456543, |
| "logits/rejected": -1.026119589805603, |
| "logps/chosen": -947.627685546875, |
| "logps/rejected": -935.3385009765625, |
| "loss": 0.691, |
| "num_input_tokens_seen": 4327552, |
| "rewards/accuracies": 0.515625, |
| "rewards/chosen": -0.007967257872223854, |
| "rewards/margins": 0.008088444359600544, |
| "rewards/rejected": -0.016055703163146973, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.049796504668422316, |
| "grad_norm": 1.0335172414779663, |
| "learning_rate": 9.996475611356265e-06, |
| "logits/chosen": -1.0391089916229248, |
| "logits/rejected": -1.0464394092559814, |
| "logps/chosen": -949.6978149414062, |
| "logps/rejected": -913.878173828125, |
| "loss": 0.6828, |
| "num_input_tokens_seen": 4701568, |
| "rewards/accuracies": 0.59375, |
| "rewards/chosen": -0.006509041413664818, |
| "rewards/margins": 0.025559592992067337, |
| "rewards/rejected": -0.0320686399936676, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.05362700502753172, |
| "grad_norm": 1.0432441234588623, |
| "learning_rate": 9.993734992753777e-06, |
| "logits/chosen": -1.0264198780059814, |
| "logits/rejected": -1.039010763168335, |
| "logps/chosen": -957.4710693359375, |
| "logps/rejected": -918.60400390625, |
| "loss": 0.6787, |
| "num_input_tokens_seen": 5056448, |
| "rewards/accuracies": 0.6640625, |
| "rewards/chosen": -0.005599617958068848, |
| "rewards/margins": 0.03315434232354164, |
| "rewards/rejected": -0.03875396400690079, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.05745750538664113, |
| "grad_norm": 1.0883227586746216, |
| "learning_rate": 9.990212076323587e-06, |
| "logits/chosen": -1.047149896621704, |
| "logits/rejected": -1.0479631423950195, |
| "logps/chosen": -911.3246459960938, |
| "logps/rejected": -907.6549072265625, |
| "loss": 0.6765, |
| "num_input_tokens_seen": 5405056, |
| "rewards/accuracies": 0.6015625, |
| "rewards/chosen": -0.0018613815773278475, |
| "rewards/margins": 0.0383245013654232, |
| "rewards/rejected": -0.04018588364124298, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.06128800574575054, |
| "grad_norm": 0.9944784045219421, |
| "learning_rate": 9.98590741395118e-06, |
| "logits/chosen": -1.0353460311889648, |
| "logits/rejected": -1.0215399265289307, |
| "logps/chosen": -875.39697265625, |
| "logps/rejected": -897.5616455078125, |
| "loss": 0.6792, |
| "num_input_tokens_seen": 5758464, |
| "rewards/accuracies": 0.6015625, |
| "rewards/chosen": -0.004133629146963358, |
| "rewards/margins": 0.034194350242614746, |
| "rewards/rejected": -0.038327984511852264, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.06511850610485995, |
| "grad_norm": 1.0136407613754272, |
| "learning_rate": 9.980821679987125e-06, |
| "logits/chosen": -1.0320684909820557, |
| "logits/rejected": -1.033224105834961, |
| "logps/chosen": -929.647216796875, |
| "logps/rejected": -911.69580078125, |
| "loss": 0.6786, |
| "num_input_tokens_seen": 6116480, |
| "rewards/accuracies": 0.59375, |
| "rewards/chosen": -0.0052222274243831635, |
| "rewards/margins": 0.036130644381046295, |
| "rewards/rejected": -0.04135286808013916, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.06894900646396936, |
| "grad_norm": 0.9946112036705017, |
| "learning_rate": 9.974955671141425e-06, |
| "logits/chosen": -1.0013747215270996, |
| "logits/rejected": -0.9874509572982788, |
| "logps/chosen": -966.796875, |
| "logps/rejected": -983.128173828125, |
| "loss": 0.6739, |
| "num_input_tokens_seen": 6482624, |
| "rewards/accuracies": 0.5703125, |
| "rewards/chosen": 0.006748510058969259, |
| "rewards/margins": 0.044564343988895416, |
| "rewards/rejected": -0.03781583532691002, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.07277950682307877, |
| "grad_norm": 1.0340311527252197, |
| "learning_rate": 9.968310306358715e-06, |
| "logits/chosen": -1.027139663696289, |
| "logits/rejected": -1.0141856670379639, |
| "logps/chosen": -920.3916625976562, |
| "logps/rejected": -924.4229736328125, |
| "loss": 0.66, |
| "num_input_tokens_seen": 6835712, |
| "rewards/accuracies": 0.6640625, |
| "rewards/chosen": 0.0031435242854058743, |
| "rewards/margins": 0.07400324940681458, |
| "rewards/rejected": -0.07085972279310226, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.07661000718218818, |
| "grad_norm": 1.0329723358154297, |
| "learning_rate": 9.960886626674302e-06, |
| "logits/chosen": -1.0163955688476562, |
| "logits/rejected": -1.0183823108673096, |
| "logps/chosen": -912.9906616210938, |
| "logps/rejected": -900.5574951171875, |
| "loss": 0.6575, |
| "num_input_tokens_seen": 7183040, |
| "rewards/accuracies": 0.6484375, |
| "rewards/chosen": 0.028547190129756927, |
| "rewards/margins": 0.07918091118335724, |
| "rewards/rejected": -0.050633713603019714, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08044050754129758, |
| "grad_norm": 0.9984006881713867, |
| "learning_rate": 9.952685795051078e-06, |
| "logits/chosen": -1.0189419984817505, |
| "logits/rejected": -1.0193971395492554, |
| "logps/chosen": -962.4049072265625, |
| "logps/rejected": -968.8514404296875, |
| "loss": 0.6635, |
| "num_input_tokens_seen": 7552000, |
| "rewards/accuracies": 0.6484375, |
| "rewards/chosen": 0.0059728133492171764, |
| "rewards/margins": 0.06682717800140381, |
| "rewards/rejected": -0.060854364186525345, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.084271007900407, |
| "grad_norm": 0.9608243107795715, |
| "learning_rate": 9.943709096197334e-06, |
| "logits/chosen": -1.0346640348434448, |
| "logits/rejected": -1.0270400047302246, |
| "logps/chosen": -943.638427734375, |
| "logps/rejected": -949.8948364257812, |
| "loss": 0.6437, |
| "num_input_tokens_seen": 7918784, |
| "rewards/accuracies": 0.7109375, |
| "rewards/chosen": 0.031516481190919876, |
| "rewards/margins": 0.11247380077838898, |
| "rewards/rejected": -0.0809573158621788, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.0881015082595164, |
| "grad_norm": 0.9745583534240723, |
| "learning_rate": 9.933957936365515e-06, |
| "logits/chosen": -1.0197912454605103, |
| "logits/rejected": -1.0225633382797241, |
| "logps/chosen": -924.9276733398438, |
| "logps/rejected": -902.6036987304688, |
| "loss": 0.6572, |
| "num_input_tokens_seen": 8275392, |
| "rewards/accuracies": 0.6953125, |
| "rewards/chosen": 0.015636922791600227, |
| "rewards/margins": 0.08563442528247833, |
| "rewards/rejected": -0.06999750435352325, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.09193200861862581, |
| "grad_norm": 1.012017846107483, |
| "learning_rate": 9.9234338431319e-06, |
| "logits/chosen": -1.0310735702514648, |
| "logits/rejected": -1.0211362838745117, |
| "logps/chosen": -943.5421752929688, |
| "logps/rejected": -943.68896484375, |
| "loss": 0.6214, |
| "num_input_tokens_seen": 8630976, |
| "rewards/accuracies": 0.734375, |
| "rewards/chosen": 0.049514107406139374, |
| "rewards/margins": 0.16673865914344788, |
| "rewards/rejected": -0.1172245517373085, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.09576250897773522, |
| "grad_norm": 1.0184485912322998, |
| "learning_rate": 9.912138465157325e-06, |
| "logits/chosen": -1.040213942527771, |
| "logits/rejected": -1.0284395217895508, |
| "logps/chosen": -944.1600952148438, |
| "logps/rejected": -950.0694580078125, |
| "loss": 0.6145, |
| "num_input_tokens_seen": 8994496, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": 0.04350156709551811, |
| "rewards/margins": 0.18440237641334534, |
| "rewards/rejected": -0.14090080559253693, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.09959300933684463, |
| "grad_norm": 0.9871705174446106, |
| "learning_rate": 9.900073571928887e-06, |
| "logits/chosen": -1.0415468215942383, |
| "logits/rejected": -1.02827787399292, |
| "logps/chosen": -930.43798828125, |
| "logps/rejected": -943.5054931640625, |
| "loss": 0.6296, |
| "num_input_tokens_seen": 9363648, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": 0.01750366762280464, |
| "rewards/margins": 0.15038560330867767, |
| "rewards/rejected": -0.13288193941116333, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.10342350969595403, |
| "grad_norm": 1.0247187614440918, |
| "learning_rate": 9.887241053482756e-06, |
| "logits/chosen": -1.0407733917236328, |
| "logits/rejected": -1.0286006927490234, |
| "logps/chosen": -901.0194702148438, |
| "logps/rejected": -897.4530029296875, |
| "loss": 0.6069, |
| "num_input_tokens_seen": 9716544, |
| "rewards/accuracies": 0.7578125, |
| "rewards/chosen": 0.0669078379869461, |
| "rewards/margins": 0.2108609676361084, |
| "rewards/rejected": -0.1439531445503235, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.10725401005506344, |
| "grad_norm": 1.011189341545105, |
| "learning_rate": 9.87364292010809e-06, |
| "logits/chosen": -1.0111464262008667, |
| "logits/rejected": -1.0069133043289185, |
| "logps/chosen": -919.2281494140625, |
| "logps/rejected": -911.4672241210938, |
| "loss": 0.5903, |
| "num_input_tokens_seen": 10062592, |
| "rewards/accuracies": 0.8046875, |
| "rewards/chosen": 0.10726039111614227, |
| "rewards/margins": 0.2591054439544678, |
| "rewards/rejected": -0.1518450379371643, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.11108451041417285, |
| "grad_norm": 0.9787644147872925, |
| "learning_rate": 9.859281302032107e-06, |
| "logits/chosen": -1.0300339460372925, |
| "logits/rejected": -1.0144660472869873, |
| "logps/chosen": -905.2968139648438, |
| "logps/rejected": -912.3560791015625, |
| "loss": 0.5922, |
| "num_input_tokens_seen": 10416320, |
| "rewards/accuracies": 0.8203125, |
| "rewards/chosen": 0.07246110588312149, |
| "rewards/margins": 0.2484126091003418, |
| "rewards/rejected": -0.1759515106678009, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.11491501077328226, |
| "grad_norm": 1.0253512859344482, |
| "learning_rate": 9.844158449086372e-06, |
| "logits/chosen": -1.0572880506515503, |
| "logits/rejected": -1.0502582788467407, |
| "logps/chosen": -945.9561767578125, |
| "logps/rejected": -944.3990478515625, |
| "loss": 0.5985, |
| "num_input_tokens_seen": 10785856, |
| "rewards/accuracies": 0.796875, |
| "rewards/chosen": 0.06250870227813721, |
| "rewards/margins": 0.23862135410308838, |
| "rewards/rejected": -0.17611265182495117, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.11874551113239167, |
| "grad_norm": 1.0238397121429443, |
| "learning_rate": 9.828276730354353e-06, |
| "logits/chosen": -1.076204776763916, |
| "logits/rejected": -1.0662384033203125, |
| "logps/chosen": -930.0040893554688, |
| "logps/rejected": -932.3923950195312, |
| "loss": 0.5576, |
| "num_input_tokens_seen": 11150656, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": 0.07753083854913712, |
| "rewards/margins": 0.3478483259677887, |
| "rewards/rejected": -0.2703174948692322, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.12257601149150107, |
| "grad_norm": 0.9527409672737122, |
| "learning_rate": 9.811638633800287e-06, |
| "logits/chosen": -1.0208882093429565, |
| "logits/rejected": -1.0182850360870361, |
| "logps/chosen": -911.5631103515625, |
| "logps/rejected": -909.9766845703125, |
| "loss": 0.5511, |
| "num_input_tokens_seen": 11493568, |
| "rewards/accuracies": 0.84375, |
| "rewards/chosen": 0.11385183781385422, |
| "rewards/margins": 0.3859643340110779, |
| "rewards/rejected": -0.27211251854896545, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.1264065118506105, |
| "grad_norm": 1.0301215648651123, |
| "learning_rate": 9.794246765879421e-06, |
| "logits/chosen": -1.041532278060913, |
| "logits/rejected": -1.0406360626220703, |
| "logps/chosen": -962.494873046875, |
| "logps/rejected": -962.85986328125, |
| "loss": 0.5723, |
| "num_input_tokens_seen": 11861824, |
| "rewards/accuracies": 0.828125, |
| "rewards/chosen": 0.08621297031641006, |
| "rewards/margins": 0.34169405698776245, |
| "rewards/rejected": -0.25548112392425537, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.1302370122097199, |
| "grad_norm": 0.9654497504234314, |
| "learning_rate": 9.776103851129706e-06, |
| "logits/chosen": -1.0220344066619873, |
| "logits/rejected": -1.015979528427124, |
| "logps/chosen": -949.3598022460938, |
| "logps/rejected": -960.7059326171875, |
| "loss": 0.5751, |
| "num_input_tokens_seen": 12229824, |
| "rewards/accuracies": 0.828125, |
| "rewards/chosen": 0.09405691176652908, |
| "rewards/margins": 0.32782119512557983, |
| "rewards/rejected": -0.23376427590847015, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.13406751256882932, |
| "grad_norm": 1.038521647453308, |
| "learning_rate": 9.757212731744973e-06, |
| "logits/chosen": -1.076894998550415, |
| "logits/rejected": -1.0720102787017822, |
| "logps/chosen": -925.4364013671875, |
| "logps/rejected": -950.203369140625, |
| "loss": 0.5121, |
| "num_input_tokens_seen": 12610880, |
| "rewards/accuracies": 0.8984375, |
| "rewards/chosen": 0.059632543474435806, |
| "rewards/margins": 0.48721498250961304, |
| "rewards/rejected": -0.42758244276046753, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.1378980129279387, |
| "grad_norm": 1.0083974599838257, |
| "learning_rate": 9.737576367129694e-06, |
| "logits/chosen": -1.034104585647583, |
| "logits/rejected": -1.0210139751434326, |
| "logps/chosen": -871.239013671875, |
| "logps/rejected": -879.6558837890625, |
| "loss": 0.536, |
| "num_input_tokens_seen": 12958592, |
| "rewards/accuracies": 0.796875, |
| "rewards/chosen": 0.1341094672679901, |
| "rewards/margins": 0.45023247599601746, |
| "rewards/rejected": -0.31612300872802734, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.1417285132870481, |
| "grad_norm": 1.0032159090042114, |
| "learning_rate": 9.717197833435367e-06, |
| "logits/chosen": -1.0570846796035767, |
| "logits/rejected": -1.049383521080017, |
| "logps/chosen": -875.153076171875, |
| "logps/rejected": -868.0505981445312, |
| "loss": 0.5294, |
| "num_input_tokens_seen": 13310336, |
| "rewards/accuracies": 0.8203125, |
| "rewards/chosen": 0.11251354217529297, |
| "rewards/margins": 0.4998120069503784, |
| "rewards/rejected": -0.38729846477508545, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.14555901364615753, |
| "grad_norm": 0.973970353603363, |
| "learning_rate": 9.696080323078621e-06, |
| "logits/chosen": -1.0454506874084473, |
| "logits/rejected": -1.0396976470947266, |
| "logps/chosen": -945.59521484375, |
| "logps/rejected": -946.391845703125, |
| "loss": 0.5191, |
| "num_input_tokens_seen": 13681920, |
| "rewards/accuracies": 0.8046875, |
| "rewards/chosen": 0.1408233493566513, |
| "rewards/margins": 0.5279620885848999, |
| "rewards/rejected": -0.3871387839317322, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.14938951400526693, |
| "grad_norm": 0.9903539419174194, |
| "learning_rate": 9.67422714424111e-06, |
| "logits/chosen": -1.0471606254577637, |
| "logits/rejected": -1.029019832611084, |
| "logps/chosen": -898.5994262695312, |
| "logps/rejected": -893.9972534179688, |
| "loss": 0.5371, |
| "num_input_tokens_seen": 14033792, |
| "rewards/accuracies": 0.8515625, |
| "rewards/chosen": 0.05680917203426361, |
| "rewards/margins": 0.4695061147212982, |
| "rewards/rejected": -0.4126969575881958, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.15322001436437635, |
| "grad_norm": 0.9689764976501465, |
| "learning_rate": 9.651641720351262e-06, |
| "logits/chosen": -1.0580934286117554, |
| "logits/rejected": -1.065984845161438, |
| "logps/chosen": -908.4832763671875, |
| "logps/rejected": -895.956298828125, |
| "loss": 0.49, |
| "num_input_tokens_seen": 14392064, |
| "rewards/accuracies": 0.8828125, |
| "rewards/chosen": 0.17145198583602905, |
| "rewards/margins": 0.5947294235229492, |
| "rewards/rejected": -0.42327749729156494, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.15705051472348575, |
| "grad_norm": 0.9584202170372009, |
| "learning_rate": 9.628327589547977e-06, |
| "logits/chosen": -1.0466926097869873, |
| "logits/rejected": -1.0385541915893555, |
| "logps/chosen": -913.1895141601562, |
| "logps/rejected": -910.8951416015625, |
| "loss": 0.4843, |
| "num_input_tokens_seen": 14750144, |
| "rewards/accuracies": 0.8828125, |
| "rewards/chosen": 0.12090866267681122, |
| "rewards/margins": 0.5805420875549316, |
| "rewards/rejected": -0.45963340997695923, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.16088101508259517, |
| "grad_norm": 0.9737739562988281, |
| "learning_rate": 9.604288404126362e-06, |
| "logits/chosen": -1.0408737659454346, |
| "logits/rejected": -1.0337262153625488, |
| "logps/chosen": -947.9314575195312, |
| "logps/rejected": -940.6116943359375, |
| "loss": 0.4864, |
| "num_input_tokens_seen": 15115648, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": 0.10129190981388092, |
| "rewards/margins": 0.609031081199646, |
| "rewards/rejected": -0.5077391862869263, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.16471151544170456, |
| "grad_norm": 1.0550665855407715, |
| "learning_rate": 9.579527929965581e-06, |
| "logits/chosen": -1.0519779920578003, |
| "logits/rejected": -1.0420559644699097, |
| "logps/chosen": -938.4093017578125, |
| "logps/rejected": -936.85498046875, |
| "loss": 0.4906, |
| "num_input_tokens_seen": 15473984, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.12262112647294998, |
| "rewards/margins": 0.6415243148803711, |
| "rewards/rejected": -0.5189031958580017, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.168542015800814, |
| "grad_norm": 0.951257050037384, |
| "learning_rate": 9.554050045938893e-06, |
| "logits/chosen": -1.0564297437667847, |
| "logits/rejected": -1.036995530128479, |
| "logps/chosen": -922.6273803710938, |
| "logps/rejected": -950.3002319335938, |
| "loss": 0.4451, |
| "num_input_tokens_seen": 15818304, |
| "rewards/accuracies": 0.8671875, |
| "rewards/chosen": 0.0952005386352539, |
| "rewards/margins": 0.8160958886146545, |
| "rewards/rejected": -0.7208954095840454, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.17237251615992338, |
| "grad_norm": 0.8960257172584534, |
| "learning_rate": 9.52785874330602e-06, |
| "logits/chosen": -1.0616629123687744, |
| "logits/rejected": -1.059366226196289, |
| "logps/chosen": -919.501220703125, |
| "logps/rejected": -901.478759765625, |
| "loss": 0.448, |
| "num_input_tokens_seen": 16173312, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": 0.227450430393219, |
| "rewards/margins": 0.8265661001205444, |
| "rewards/rejected": -0.5991157293319702, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.1762030165190328, |
| "grad_norm": 0.8991801738739014, |
| "learning_rate": 9.500958125087882e-06, |
| "logits/chosen": -1.0398144721984863, |
| "logits/rejected": -1.0337408781051636, |
| "logps/chosen": -921.608154296875, |
| "logps/rejected": -926.8184814453125, |
| "loss": 0.4443, |
| "num_input_tokens_seen": 16542400, |
| "rewards/accuracies": 0.8671875, |
| "rewards/chosen": 0.1689867377281189, |
| "rewards/margins": 0.7900327444076538, |
| "rewards/rejected": -0.6210460066795349, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.1800335168781422, |
| "grad_norm": 0.9368561506271362, |
| "learning_rate": 9.473352405423845e-06, |
| "logits/chosen": -1.0370045900344849, |
| "logits/rejected": -1.034510612487793, |
| "logps/chosen": -927.1217041015625, |
| "logps/rejected": -940.380615234375, |
| "loss": 0.4219, |
| "num_input_tokens_seen": 16906368, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": 0.12228920310735703, |
| "rewards/margins": 0.91481614112854, |
| "rewards/rejected": -0.7925269603729248, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.18386401723725163, |
| "grad_norm": 0.8737442493438721, |
| "learning_rate": 9.445045908911536e-06, |
| "logits/chosen": -1.0180749893188477, |
| "logits/rejected": -1.0195181369781494, |
| "logps/chosen": -914.4266357421875, |
| "logps/rejected": -919.2938842773438, |
| "loss": 0.4229, |
| "num_input_tokens_seen": 17244160, |
| "rewards/accuracies": 0.8828125, |
| "rewards/chosen": 0.22764116525650024, |
| "rewards/margins": 0.9235501289367676, |
| "rewards/rejected": -0.6959089040756226, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.18769451759636102, |
| "grad_norm": 0.9246760606765747, |
| "learning_rate": 9.416043069929389e-06, |
| "logits/chosen": -1.0696072578430176, |
| "logits/rejected": -1.0749962329864502, |
| "logps/chosen": -915.3636474609375, |
| "logps/rejected": -915.0205688476562, |
| "loss": 0.421, |
| "num_input_tokens_seen": 17605632, |
| "rewards/accuracies": 0.8828125, |
| "rewards/chosen": 0.03486424311995506, |
| "rewards/margins": 0.9285318851470947, |
| "rewards/rejected": -0.8936675786972046, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.19152501795547044, |
| "grad_norm": 0.8932781219482422, |
| "learning_rate": 9.386348431941953e-06, |
| "logits/chosen": -1.030189871788025, |
| "logits/rejected": -1.0274410247802734, |
| "logps/chosen": -927.2352905273438, |
| "logps/rejected": -913.6710205078125, |
| "loss": 0.4248, |
| "num_input_tokens_seen": 17966784, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.15712372958660126, |
| "rewards/margins": 0.8464722037315369, |
| "rewards/rejected": -0.6893484592437744, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.19535551831457984, |
| "grad_norm": 0.8208088278770447, |
| "learning_rate": 9.355966646788152e-06, |
| "logits/chosen": -1.0399882793426514, |
| "logits/rejected": -1.049922227859497, |
| "logps/chosen": -922.302490234375, |
| "logps/rejected": -926.0870971679688, |
| "loss": 0.3664, |
| "num_input_tokens_seen": 18328768, |
| "rewards/accuracies": 0.8984375, |
| "rewards/chosen": 0.2352968007326126, |
| "rewards/margins": 1.2056996822357178, |
| "rewards/rejected": -0.9704028367996216, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.19918601867368926, |
| "grad_norm": 0.9681217670440674, |
| "learning_rate": 9.324902473952529e-06, |
| "logits/chosen": -1.0520392656326294, |
| "logits/rejected": -1.0435035228729248, |
| "logps/chosen": -912.5722045898438, |
| "logps/rejected": -910.1470947265625, |
| "loss": 0.427, |
| "num_input_tokens_seen": 18682304, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": 0.007225107401609421, |
| "rewards/margins": 0.9377115368843079, |
| "rewards/rejected": -0.9304863810539246, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.20301651903279866, |
| "grad_norm": 0.8848323225975037, |
| "learning_rate": 9.293160779819658e-06, |
| "logits/chosen": -1.0435223579406738, |
| "logits/rejected": -1.0460057258605957, |
| "logps/chosen": -913.4359741210938, |
| "logps/rejected": -926.9989624023438, |
| "loss": 0.3614, |
| "num_input_tokens_seen": 19037760, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.2528911828994751, |
| "rewards/margins": 1.2186728715896606, |
| "rewards/rejected": -0.9657817482948303, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.20684701939190805, |
| "grad_norm": 0.8267402052879333, |
| "learning_rate": 9.260746536911792e-06, |
| "logits/chosen": -1.0281553268432617, |
| "logits/rejected": -1.0281143188476562, |
| "logps/chosen": -927.7498779296875, |
| "logps/rejected": -912.3658447265625, |
| "loss": 0.3549, |
| "num_input_tokens_seen": 19394688, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": 0.14229370653629303, |
| "rewards/margins": 1.2576112747192383, |
| "rewards/rejected": -1.1153175830841064, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.21067751975101748, |
| "grad_norm": 0.8928641080856323, |
| "learning_rate": 9.227664823109884e-06, |
| "logits/chosen": -1.0537580251693726, |
| "logits/rejected": -1.0493597984313965, |
| "logps/chosen": -946.2432250976562, |
| "logps/rejected": -960.2678833007812, |
| "loss": 0.3531, |
| "num_input_tokens_seen": 19760704, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.15718036890029907, |
| "rewards/margins": 1.3443400859832764, |
| "rewards/rejected": -1.187159776687622, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.21450802011012687, |
| "grad_norm": 0.8100174069404602, |
| "learning_rate": 9.193920820858113e-06, |
| "logits/chosen": -1.0525925159454346, |
| "logits/rejected": -1.0595781803131104, |
| "logps/chosen": -925.456298828125, |
| "logps/rejected": -904.4075927734375, |
| "loss": 0.3455, |
| "num_input_tokens_seen": 20121856, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": 0.1917523741722107, |
| "rewards/margins": 1.3067724704742432, |
| "rewards/rejected": -1.1150201559066772, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.2183385204692363, |
| "grad_norm": 0.7760931253433228, |
| "learning_rate": 9.159519816352021e-06, |
| "logits/chosen": -1.0448260307312012, |
| "logits/rejected": -1.0453829765319824, |
| "logps/chosen": -930.4881591796875, |
| "logps/rejected": -940.2239990234375, |
| "loss": 0.3191, |
| "num_input_tokens_seen": 20482304, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": 0.07732844352722168, |
| "rewards/margins": 1.3429336547851562, |
| "rewards/rejected": -1.2656052112579346, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.2221690208283457, |
| "grad_norm": 0.7795583605766296, |
| "learning_rate": 9.124467198710401e-06, |
| "logits/chosen": -1.0545520782470703, |
| "logits/rejected": -1.0434613227844238, |
| "logps/chosen": -903.4138793945312, |
| "logps/rejected": -914.3572998046875, |
| "loss": 0.3171, |
| "num_input_tokens_seen": 20822656, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.00938415713608265, |
| "rewards/margins": 1.46409273147583, |
| "rewards/rejected": -1.4734768867492676, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.22599952118745512, |
| "grad_norm": 0.7819092273712158, |
| "learning_rate": 9.08876845913106e-06, |
| "logits/chosen": -1.065345287322998, |
| "logits/rejected": -1.071925163269043, |
| "logps/chosen": -912.1345825195312, |
| "logps/rejected": -898.795166015625, |
| "loss": 0.3078, |
| "num_input_tokens_seen": 21187392, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": 0.2630062997341156, |
| "rewards/margins": 1.5684618949890137, |
| "rewards/rejected": -1.3054554462432861, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.2298300215465645, |
| "grad_norm": 0.810053288936615, |
| "learning_rate": 9.052429190030589e-06, |
| "logits/chosen": -1.0523853302001953, |
| "logits/rejected": -1.0394606590270996, |
| "logps/chosen": -940.7644653320312, |
| "logps/rejected": -968.588134765625, |
| "loss": 0.3253, |
| "num_input_tokens_seen": 21545600, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.11970774829387665, |
| "rewards/margins": 1.4579362869262695, |
| "rewards/rejected": -1.3382285833358765, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.23366052190567393, |
| "grad_norm": 0.8415746092796326, |
| "learning_rate": 9.015455084168279e-06, |
| "logits/chosen": -1.0666999816894531, |
| "logits/rejected": -1.0684211254119873, |
| "logps/chosen": -869.7279052734375, |
| "logps/rejected": -885.7716064453125, |
| "loss": 0.3212, |
| "num_input_tokens_seen": 21887936, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": 0.0877489447593689, |
| "rewards/margins": 1.599064588546753, |
| "rewards/rejected": -1.5113155841827393, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.23749102226478333, |
| "grad_norm": 0.8196543455123901, |
| "learning_rate": 8.977851933754317e-06, |
| "logits/chosen": -1.0270411968231201, |
| "logits/rejected": -1.0330016613006592, |
| "logps/chosen": -931.991455078125, |
| "logps/rejected": -925.2451171875, |
| "loss": 0.3188, |
| "num_input_tokens_seen": 22233024, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": 0.1586177945137024, |
| "rewards/margins": 1.4711058139801025, |
| "rewards/rejected": -1.312488079071045, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.24132152262389275, |
| "grad_norm": 0.7564411759376526, |
| "learning_rate": 8.939625629542401e-06, |
| "logits/chosen": -1.0505956411361694, |
| "logits/rejected": -1.0580633878707886, |
| "logps/chosen": -902.9009399414062, |
| "logps/rejected": -906.3607177734375, |
| "loss": 0.2957, |
| "num_input_tokens_seen": 22585472, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": 0.10947957634925842, |
| "rewards/margins": 1.675947666168213, |
| "rewards/rejected": -1.5664680004119873, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.24515202298300215, |
| "grad_norm": 0.805608868598938, |
| "learning_rate": 8.900782159906927e-06, |
| "logits/chosen": -1.0741575956344604, |
| "logits/rejected": -1.0680464506149292, |
| "logps/chosen": -903.8414306640625, |
| "logps/rejected": -929.549560546875, |
| "loss": 0.2977, |
| "num_input_tokens_seen": 22946880, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.1019095927476883, |
| "rewards/margins": 1.7661054134368896, |
| "rewards/rejected": -1.6641957759857178, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.24898252334211157, |
| "grad_norm": 0.733465313911438, |
| "learning_rate": 8.861327609904859e-06, |
| "logits/chosen": -1.0529414415359497, |
| "logits/rejected": -1.048985481262207, |
| "logps/chosen": -935.5690307617188, |
| "logps/rejected": -958.1524658203125, |
| "loss": 0.2786, |
| "num_input_tokens_seen": 23304320, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": 0.12524667382240295, |
| "rewards/margins": 1.8969340324401855, |
| "rewards/rejected": -1.7716875076293945, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.252813023701221, |
| "grad_norm": 0.8141180872917175, |
| "learning_rate": 8.821268160322482e-06, |
| "logits/chosen": -1.0694968700408936, |
| "logits/rejected": -1.0694319009780884, |
| "logps/chosen": -927.3983154296875, |
| "logps/rejected": -939.122314453125, |
| "loss": 0.2977, |
| "num_input_tokens_seen": 23665536, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": 0.02337331511080265, |
| "rewards/margins": 1.768088936805725, |
| "rewards/rejected": -1.7447155714035034, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.25664352406033036, |
| "grad_norm": 0.7073566317558289, |
| "learning_rate": 8.780610086707149e-06, |
| "logits/chosen": -1.0607355833053589, |
| "logits/rejected": -1.0568872690200806, |
| "logps/chosen": -948.0847778320312, |
| "logps/rejected": -965.6436767578125, |
| "loss": 0.2613, |
| "num_input_tokens_seen": 24021376, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.10131228715181351, |
| "rewards/margins": 1.7949880361557007, |
| "rewards/rejected": -1.8963003158569336, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.2604740244194398, |
| "grad_norm": 0.7342793941497803, |
| "learning_rate": 8.739359758384162e-06, |
| "logits/chosen": -1.0782382488250732, |
| "logits/rejected": -1.077387809753418, |
| "logps/chosen": -932.740478515625, |
| "logps/rejected": -940.6168823242188, |
| "loss": 0.2722, |
| "num_input_tokens_seen": 24385216, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.0452754907310009, |
| "rewards/margins": 1.871610164642334, |
| "rewards/rejected": -1.916885495185852, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.2643045247785492, |
| "grad_norm": 0.812954306602478, |
| "learning_rate": 8.697523637458997e-06, |
| "logits/chosen": -1.063349962234497, |
| "logits/rejected": -1.0582802295684814, |
| "logps/chosen": -923.201171875, |
| "logps/rejected": -935.93603515625, |
| "loss": 0.2872, |
| "num_input_tokens_seen": 24750080, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.011177586391568184, |
| "rewards/margins": 1.758663296699524, |
| "rewards/rejected": -1.747485637664795, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.26813502513765863, |
| "grad_norm": 0.7448921203613281, |
| "learning_rate": 8.655108277804975e-06, |
| "logits/chosen": -1.0851333141326904, |
| "logits/rejected": -1.079848289489746, |
| "logps/chosen": -936.8907470703125, |
| "logps/rejected": -967.2071533203125, |
| "loss": 0.272, |
| "num_input_tokens_seen": 25116480, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": -0.16820263862609863, |
| "rewards/margins": 1.8837871551513672, |
| "rewards/rejected": -2.051989793777466, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.271965525496768, |
| "grad_norm": 0.6441535353660583, |
| "learning_rate": 8.612120324036548e-06, |
| "logits/chosen": -1.0761862993240356, |
| "logits/rejected": -1.0810930728912354, |
| "logps/chosen": -910.4297485351562, |
| "logps/rejected": -922.0003662109375, |
| "loss": 0.2075, |
| "num_input_tokens_seen": 25465536, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.0027351118624210358, |
| "rewards/margins": 2.2052128314971924, |
| "rewards/rejected": -2.2079479694366455, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.2757960258558774, |
| "grad_norm": 0.7206379771232605, |
| "learning_rate": 8.568566510468392e-06, |
| "logits/chosen": -1.0868092775344849, |
| "logits/rejected": -1.0811575651168823, |
| "logps/chosen": -941.665771484375, |
| "logps/rejected": -973.3070068359375, |
| "loss": 0.2464, |
| "num_input_tokens_seen": 25833344, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.2859920263290405, |
| "rewards/margins": 2.0870699882507324, |
| "rewards/rejected": -2.3730621337890625, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.27962652621498685, |
| "grad_norm": 0.8260279297828674, |
| "learning_rate": 8.524453660060434e-06, |
| "logits/chosen": -1.0668997764587402, |
| "logits/rejected": -1.06461763381958, |
| "logps/chosen": -894.1995849609375, |
| "logps/rejected": -909.6748657226562, |
| "loss": 0.2724, |
| "num_input_tokens_seen": 26183872, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.35972511768341064, |
| "rewards/margins": 1.8394525051116943, |
| "rewards/rejected": -2.1991777420043945, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.2834570265740962, |
| "grad_norm": 0.8776343464851379, |
| "learning_rate": 8.479788683348996e-06, |
| "logits/chosen": -1.0618414878845215, |
| "logits/rejected": -1.0610969066619873, |
| "logps/chosen": -920.0272216796875, |
| "logps/rejected": -902.9821166992188, |
| "loss": 0.2695, |
| "num_input_tokens_seen": 26539904, |
| "rewards/accuracies": 0.8984375, |
| "rewards/chosen": -0.3319213390350342, |
| "rewards/margins": 1.9274811744689941, |
| "rewards/rejected": -2.259402275085449, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.28728752693320564, |
| "grad_norm": 0.7757173180580139, |
| "learning_rate": 8.434578577364218e-06, |
| "logits/chosen": -1.0758130550384521, |
| "logits/rejected": -1.0636892318725586, |
| "logps/chosen": -931.8839111328125, |
| "logps/rejected": -947.8310546875, |
| "loss": 0.266, |
| "num_input_tokens_seen": 26902464, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": -0.3294548988342285, |
| "rewards/margins": 2.0559473037719727, |
| "rewards/rejected": -2.385402202606201, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.29111802729231506, |
| "grad_norm": 0.7610207796096802, |
| "learning_rate": 8.388830424533935e-06, |
| "logits/chosen": -1.078747272491455, |
| "logits/rejected": -1.0726932287216187, |
| "logps/chosen": -928.5173950195312, |
| "logps/rejected": -959.848876953125, |
| "loss": 0.2346, |
| "num_input_tokens_seen": 27269504, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.27425873279571533, |
| "rewards/margins": 2.1390914916992188, |
| "rewards/rejected": -2.4133503437042236, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.2949485276514245, |
| "grad_norm": 0.7980623841285706, |
| "learning_rate": 8.342551391574165e-06, |
| "logits/chosen": -1.0808157920837402, |
| "logits/rejected": -1.0889337062835693, |
| "logps/chosen": -942.5945434570312, |
| "logps/rejected": -927.933349609375, |
| "loss": 0.263, |
| "num_input_tokens_seen": 27629248, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.3413712978363037, |
| "rewards/margins": 2.096808433532715, |
| "rewards/rejected": -2.4381794929504395, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.29877902801053385, |
| "grad_norm": 0.7272222638130188, |
| "learning_rate": 8.295748728366414e-06, |
| "logits/chosen": -1.0566456317901611, |
| "logits/rejected": -1.051896333694458, |
| "logps/chosen": -941.3970947265625, |
| "logps/rejected": -983.9376220703125, |
| "loss": 0.2319, |
| "num_input_tokens_seen": 27988096, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.41651400923728943, |
| "rewards/margins": 2.1791977882385254, |
| "rewards/rejected": -2.5957119464874268, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.3026095283696433, |
| "grad_norm": 0.7870813608169556, |
| "learning_rate": 8.248429766821925e-06, |
| "logits/chosen": -1.0612484216690063, |
| "logits/rejected": -1.0511322021484375, |
| "logps/chosen": -965.112548828125, |
| "logps/rejected": -980.8787231445312, |
| "loss": 0.2636, |
| "num_input_tokens_seen": 28346432, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.5220478177070618, |
| "rewards/margins": 2.1551966667175293, |
| "rewards/rejected": -2.6772446632385254, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.3064400287287527, |
| "grad_norm": 0.8286476731300354, |
| "learning_rate": 8.200601919733106e-06, |
| "logits/chosen": -1.084940791130066, |
| "logits/rejected": -1.0818970203399658, |
| "logps/chosen": -959.9788818359375, |
| "logps/rejected": -981.2055053710938, |
| "loss": 0.2616, |
| "num_input_tokens_seen": 28715200, |
| "rewards/accuracies": 0.8984375, |
| "rewards/chosen": -0.49848473072052, |
| "rewards/margins": 2.191138744354248, |
| "rewards/rejected": -2.6896233558654785, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.3102705290878621, |
| "grad_norm": 0.7395047545433044, |
| "learning_rate": 8.15227267961226e-06, |
| "logits/chosen": -1.071365237236023, |
| "logits/rejected": -1.0694466829299927, |
| "logps/chosen": -950.6376342773438, |
| "logps/rejected": -960.358154296875, |
| "loss": 0.2285, |
| "num_input_tokens_seen": 29068800, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.37492138147354126, |
| "rewards/margins": 2.5624656677246094, |
| "rewards/rejected": -2.937386989593506, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.3141010294469715, |
| "grad_norm": 0.8469851613044739, |
| "learning_rate": 8.10344961751785e-06, |
| "logits/chosen": -1.0703625679016113, |
| "logits/rejected": -1.0763499736785889, |
| "logps/chosen": -921.09326171875, |
| "logps/rejected": -924.1490478515625, |
| "loss": 0.257, |
| "num_input_tokens_seen": 29416000, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": -0.47378402948379517, |
| "rewards/margins": 2.3074848651885986, |
| "rewards/rejected": -2.781269073486328, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.3179315298060809, |
| "grad_norm": 0.7990094423294067, |
| "learning_rate": 8.054140381868435e-06, |
| "logits/chosen": -1.1002495288848877, |
| "logits/rejected": -1.086364984512329, |
| "logps/chosen": -932.9931640625, |
| "logps/rejected": -955.93017578125, |
| "loss": 0.2235, |
| "num_input_tokens_seen": 29768704, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.4593415856361389, |
| "rewards/margins": 2.4236037731170654, |
| "rewards/rejected": -2.8829455375671387, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.32176203016519034, |
| "grad_norm": 0.8141796588897705, |
| "learning_rate": 8.004352697244516e-06, |
| "logits/chosen": -1.0791590213775635, |
| "logits/rejected": -1.07072114944458, |
| "logps/chosen": -971.5861206054688, |
| "logps/rejected": -981.2337646484375, |
| "loss": 0.2401, |
| "num_input_tokens_seen": 30143360, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.5980307459831238, |
| "rewards/margins": 2.2348849773406982, |
| "rewards/rejected": -2.8329155445098877, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.32559253052429976, |
| "grad_norm": 0.6741338968276978, |
| "learning_rate": 7.954094363178421e-06, |
| "logits/chosen": -1.0647037029266357, |
| "logits/rejected": -1.053682804107666, |
| "logps/chosen": -924.1261596679688, |
| "logps/rejected": -958.7186889648438, |
| "loss": 0.2107, |
| "num_input_tokens_seen": 30496832, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.48919373750686646, |
| "rewards/margins": 2.4839816093444824, |
| "rewards/rejected": -2.973175525665283, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.32942303088340913, |
| "grad_norm": 0.7606660723686218, |
| "learning_rate": 7.903373252932474e-06, |
| "logits/chosen": -1.090849757194519, |
| "logits/rejected": -1.0853286981582642, |
| "logps/chosen": -919.6220092773438, |
| "logps/rejected": -942.2991943359375, |
| "loss": 0.2237, |
| "num_input_tokens_seen": 30854400, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.6518514156341553, |
| "rewards/margins": 2.4913079738616943, |
| "rewards/rejected": -3.1431593894958496, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.33325353124251855, |
| "grad_norm": 0.7608671188354492, |
| "learning_rate": 7.852197312265592e-06, |
| "logits/chosen": -1.0895733833312988, |
| "logits/rejected": -1.0881158113479614, |
| "logps/chosen": -909.59716796875, |
| "logps/rejected": -909.6011962890625, |
| "loss": 0.2435, |
| "num_input_tokens_seen": 31212288, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": -0.43335431814193726, |
| "rewards/margins": 2.310985565185547, |
| "rewards/rejected": -2.744339942932129, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.337084031601628, |
| "grad_norm": 0.8353620171546936, |
| "learning_rate": 7.800574558188548e-06, |
| "logits/chosen": -1.090893030166626, |
| "logits/rejected": -1.0922658443450928, |
| "logps/chosen": -939.388427734375, |
| "logps/rejected": -945.8134765625, |
| "loss": 0.2273, |
| "num_input_tokens_seen": 31576000, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.6613324284553528, |
| "rewards/margins": 2.4176292419433594, |
| "rewards/rejected": -3.0789618492126465, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.34091453196073734, |
| "grad_norm": 0.8224096298217773, |
| "learning_rate": 7.748513077708044e-06, |
| "logits/chosen": -1.0749074220657349, |
| "logits/rejected": -1.0710772275924683, |
| "logps/chosen": -918.1564331054688, |
| "logps/rejected": -931.9786376953125, |
| "loss": 0.2568, |
| "num_input_tokens_seen": 31930560, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.7582589387893677, |
| "rewards/margins": 2.1596531867980957, |
| "rewards/rejected": -2.917912006378174, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.34474503231984677, |
| "grad_norm": 0.6878222227096558, |
| "learning_rate": 7.69602102655985e-06, |
| "logits/chosen": -1.1127848625183105, |
| "logits/rejected": -1.113338589668274, |
| "logps/chosen": -934.9884643554688, |
| "logps/rejected": -962.1505126953125, |
| "loss": 0.1811, |
| "num_input_tokens_seen": 32295488, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.605439305305481, |
| "rewards/margins": 2.958923816680908, |
| "rewards/rejected": -3.5643632411956787, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.3485755326789562, |
| "grad_norm": 0.6399231553077698, |
| "learning_rate": 7.643106627931148e-06, |
| "logits/chosen": -1.1128592491149902, |
| "logits/rejected": -1.107898473739624, |
| "logps/chosen": -984.710205078125, |
| "logps/rejected": -1000.18701171875, |
| "loss": 0.1776, |
| "num_input_tokens_seen": 32673984, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.5592378377914429, |
| "rewards/margins": 2.7699828147888184, |
| "rewards/rejected": -3.3292205333709717, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.3524060330380656, |
| "grad_norm": 0.8195556998252869, |
| "learning_rate": 7.5897781711723215e-06, |
| "logits/chosen": -1.097868800163269, |
| "logits/rejected": -1.0952892303466797, |
| "logps/chosen": -920.0769653320312, |
| "logps/rejected": -936.2384033203125, |
| "loss": 0.2421, |
| "num_input_tokens_seen": 33035968, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": -0.6490311622619629, |
| "rewards/margins": 2.5610485076904297, |
| "rewards/rejected": -3.2100796699523926, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.356236533397175, |
| "grad_norm": 0.7686013579368591, |
| "learning_rate": 7.536044010498396e-06, |
| "logits/chosen": -1.095731258392334, |
| "logits/rejected": -1.1059951782226562, |
| "logps/chosen": -896.857421875, |
| "logps/rejected": -911.7064208984375, |
| "loss": 0.2218, |
| "num_input_tokens_seen": 33392192, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.6852568984031677, |
| "rewards/margins": 2.4506115913391113, |
| "rewards/rejected": -3.1358680725097656, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.3600670337562844, |
| "grad_norm": 0.6556101441383362, |
| "learning_rate": 7.48191256368028e-06, |
| "logits/chosen": -1.1224572658538818, |
| "logits/rejected": -1.114153265953064, |
| "logps/chosen": -952.65966796875, |
| "logps/rejected": -986.4073486328125, |
| "loss": 0.1707, |
| "num_input_tokens_seen": 33758976, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.6452797651290894, |
| "rewards/margins": 3.0503456592559814, |
| "rewards/rejected": -3.6956255435943604, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.36389753411539383, |
| "grad_norm": 0.7410134673118591, |
| "learning_rate": 7.427392310726088e-06, |
| "logits/chosen": -1.1127829551696777, |
| "logits/rejected": -1.098555326461792, |
| "logps/chosen": -939.870849609375, |
| "logps/rejected": -954.7718505859375, |
| "loss": 0.2029, |
| "num_input_tokens_seen": 34123840, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": -0.7152698040008545, |
| "rewards/margins": 2.7910680770874023, |
| "rewards/rejected": -3.5063376426696777, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.36772803447450325, |
| "grad_norm": 0.6477031707763672, |
| "learning_rate": 7.372491792552694e-06, |
| "logits/chosen": -1.1184864044189453, |
| "logits/rejected": -1.110473394393921, |
| "logps/chosen": -943.9148559570312, |
| "logps/rejected": -957.20068359375, |
| "loss": 0.1596, |
| "num_input_tokens_seen": 34479552, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.680740237236023, |
| "rewards/margins": 2.777618885040283, |
| "rewards/rejected": -3.4583587646484375, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.3715585348336126, |
| "grad_norm": 0.6712774634361267, |
| "learning_rate": 7.31721960964774e-06, |
| "logits/chosen": -1.0910961627960205, |
| "logits/rejected": -1.0890021324157715, |
| "logps/chosen": -918.433349609375, |
| "logps/rejected": -942.1868896484375, |
| "loss": 0.1881, |
| "num_input_tokens_seen": 34835840, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.40874814987182617, |
| "rewards/margins": 2.9342246055603027, |
| "rewards/rejected": -3.342972755432129, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.37538903519272204, |
| "grad_norm": 0.8931744694709778, |
| "learning_rate": 7.261584420722328e-06, |
| "logits/chosen": -1.111809253692627, |
| "logits/rejected": -1.117790699005127, |
| "logps/chosen": -947.8694458007812, |
| "logps/rejected": -973.1971435546875, |
| "loss": 0.2405, |
| "num_input_tokens_seen": 35210176, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.7285137176513672, |
| "rewards/margins": 2.551473617553711, |
| "rewards/rejected": -3.279987335205078, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.37921953555183147, |
| "grad_norm": 0.722402811050415, |
| "learning_rate": 7.20559494135458e-06, |
| "logits/chosen": -1.0994579792022705, |
| "logits/rejected": -1.1112189292907715, |
| "logps/chosen": -933.018798828125, |
| "logps/rejected": -954.1309814453125, |
| "loss": 0.195, |
| "num_input_tokens_seen": 35568576, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.6893086433410645, |
| "rewards/margins": 2.9537758827209473, |
| "rewards/rejected": -3.643084764480591, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.3830500359109409, |
| "grad_norm": 0.7066813707351685, |
| "learning_rate": 7.149259942624287e-06, |
| "logits/chosen": -1.118371605873108, |
| "logits/rejected": -1.1083605289459229, |
| "logps/chosen": -933.5869750976562, |
| "logps/rejected": -955.2203369140625, |
| "loss": 0.186, |
| "num_input_tokens_seen": 35929024, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.6603685617446899, |
| "rewards/margins": 2.6682770252227783, |
| "rewards/rejected": -3.328645706176758, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.38688053627005026, |
| "grad_norm": 0.7495232224464417, |
| "learning_rate": 7.092588249738871e-06, |
| "logits/chosen": -1.1121894121170044, |
| "logits/rejected": -1.1247451305389404, |
| "logps/chosen": -929.92431640625, |
| "logps/rejected": -921.3439331054688, |
| "loss": 0.1837, |
| "num_input_tokens_seen": 36292032, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.5730222463607788, |
| "rewards/margins": 2.735011100769043, |
| "rewards/rejected": -3.3080334663391113, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.3907110366291597, |
| "grad_norm": 0.7650550007820129, |
| "learning_rate": 7.03558874065087e-06, |
| "logits/chosen": -1.1010173559188843, |
| "logits/rejected": -1.1006340980529785, |
| "logps/chosen": -895.2073974609375, |
| "logps/rejected": -915.1663208007812, |
| "loss": 0.181, |
| "num_input_tokens_seen": 36647104, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.6195077896118164, |
| "rewards/margins": 2.826021432876587, |
| "rewards/rejected": -3.4455294609069824, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.3945415369882691, |
| "grad_norm": 0.6909847259521484, |
| "learning_rate": 6.978270344667143e-06, |
| "logits/chosen": -1.1063586473464966, |
| "logits/rejected": -1.105713129043579, |
| "logps/chosen": -921.406005859375, |
| "logps/rejected": -936.5244140625, |
| "loss": 0.1815, |
| "num_input_tokens_seen": 36996736, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.5552771687507629, |
| "rewards/margins": 2.764892578125, |
| "rewards/rejected": -3.320169448852539, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.3983720373473785, |
| "grad_norm": 0.9186336398124695, |
| "learning_rate": 6.920642041050055e-06, |
| "logits/chosen": -1.0915818214416504, |
| "logits/rejected": -1.094931960105896, |
| "logps/chosen": -947.0763549804688, |
| "logps/rejected": -944.1019897460938, |
| "loss": 0.2464, |
| "num_input_tokens_seen": 37351296, |
| "rewards/accuracies": 0.8828125, |
| "rewards/chosen": -0.7677267789840698, |
| "rewards/margins": 2.4657530784606934, |
| "rewards/rejected": -3.2334797382354736, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.4022025377064879, |
| "grad_norm": 0.6091768741607666, |
| "learning_rate": 6.862712857610812e-06, |
| "logits/chosen": -1.100638508796692, |
| "logits/rejected": -1.102898120880127, |
| "logps/chosen": -956.29833984375, |
| "logps/rejected": -961.6322021484375, |
| "loss": 0.128, |
| "num_input_tokens_seen": 37710464, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.5590561032295227, |
| "rewards/margins": 3.1229941844940186, |
| "rewards/rejected": -3.6820502281188965, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.4060330380655973, |
| "grad_norm": 0.7067527770996094, |
| "learning_rate": 6.804491869295207e-06, |
| "logits/chosen": -1.1310827732086182, |
| "logits/rejected": -1.1312439441680908, |
| "logps/chosen": -952.262451171875, |
| "logps/rejected": -952.248291015625, |
| "loss": 0.1922, |
| "num_input_tokens_seen": 38075136, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.574321448802948, |
| "rewards/margins": 2.7026026248931885, |
| "rewards/rejected": -3.2769241333007812, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.40986353842470674, |
| "grad_norm": 0.6532292366027832, |
| "learning_rate": 6.745988196761976e-06, |
| "logits/chosen": -1.0836384296417236, |
| "logits/rejected": -1.082061529159546, |
| "logps/chosen": -958.8799438476562, |
| "logps/rejected": -977.3197021484375, |
| "loss": 0.1652, |
| "num_input_tokens_seen": 38438464, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.4467155337333679, |
| "rewards/margins": 3.1544971466064453, |
| "rewards/rejected": -3.601212739944458, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.4136940387838161, |
| "grad_norm": 0.593960165977478, |
| "learning_rate": 6.687211004953992e-06, |
| "logits/chosen": -1.1067237854003906, |
| "logits/rejected": -1.0953024625778198, |
| "logps/chosen": -911.4755249023438, |
| "logps/rejected": -923.3955078125, |
| "loss": 0.1384, |
| "num_input_tokens_seen": 38786112, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.5207453370094299, |
| "rewards/margins": 3.25771427154541, |
| "rewards/rejected": -3.7784600257873535, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.41752453914292553, |
| "grad_norm": 0.6579384803771973, |
| "learning_rate": 6.628169501662527e-06, |
| "logits/chosen": -1.1251742839813232, |
| "logits/rejected": -1.1185029745101929, |
| "logps/chosen": -912.025390625, |
| "logps/rejected": -927.5040283203125, |
| "loss": 0.1425, |
| "num_input_tokens_seen": 39137920, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.6473366618156433, |
| "rewards/margins": 3.030540943145752, |
| "rewards/rejected": -3.67787766456604, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.42135503950203496, |
| "grad_norm": 0.8059095144271851, |
| "learning_rate": 6.568872936084789e-06, |
| "logits/chosen": -1.120704174041748, |
| "logits/rejected": -1.1155285835266113, |
| "logps/chosen": -918.2825317382812, |
| "logps/rejected": -955.4237060546875, |
| "loss": 0.2094, |
| "num_input_tokens_seen": 39496896, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.7922378778457642, |
| "rewards/margins": 2.839639902114868, |
| "rewards/rejected": -3.6318776607513428, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.4251855398611444, |
| "grad_norm": 0.7774308323860168, |
| "learning_rate": 6.509330597374993e-06, |
| "logits/chosen": -1.1296982765197754, |
| "logits/rejected": -1.12125825881958, |
| "logps/chosen": -955.875, |
| "logps/rejected": -982.0862426757812, |
| "loss": 0.1983, |
| "num_input_tokens_seen": 39862784, |
| "rewards/accuracies": 0.9140625, |
| "rewards/chosen": -0.7407379150390625, |
| "rewards/margins": 2.9459829330444336, |
| "rewards/rejected": -3.686720848083496, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.42901604022025375, |
| "grad_norm": 0.8787282705307007, |
| "learning_rate": 6.44955181318915e-06, |
| "logits/chosen": -1.130476951599121, |
| "logits/rejected": -1.124819040298462, |
| "logps/chosen": -951.8047485351562, |
| "logps/rejected": -974.5203857421875, |
| "loss": 0.2063, |
| "num_input_tokens_seen": 40229504, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.7864696383476257, |
| "rewards/margins": 3.234025478363037, |
| "rewards/rejected": -4.020495414733887, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.43284654057936317, |
| "grad_norm": 0.8070006966590881, |
| "learning_rate": 6.389545948223841e-06, |
| "logits/chosen": -1.1072964668273926, |
| "logits/rejected": -1.0939278602600098, |
| "logps/chosen": -906.2698974609375, |
| "logps/rejected": -934.371337890625, |
| "loss": 0.2079, |
| "num_input_tokens_seen": 40578304, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.6210307478904724, |
| "rewards/margins": 2.9884490966796875, |
| "rewards/rejected": -3.6094799041748047, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.4366770409384726, |
| "grad_norm": 0.6278772354125977, |
| "learning_rate": 6.329322402749181e-06, |
| "logits/chosen": -1.1068530082702637, |
| "logits/rejected": -1.0973081588745117, |
| "logps/chosen": -923.3222045898438, |
| "logps/rejected": -973.787109375, |
| "loss": 0.1346, |
| "num_input_tokens_seen": 40926400, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.6330747604370117, |
| "rewards/margins": 3.4524285793304443, |
| "rewards/rejected": -4.085503578186035, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.440507541297582, |
| "grad_norm": 0.7459933161735535, |
| "learning_rate": 6.2688906111362115e-06, |
| "logits/chosen": -1.1147935390472412, |
| "logits/rejected": -1.1260608434677124, |
| "logps/chosen": -942.9789428710938, |
| "logps/rejected": -942.0307006835938, |
| "loss": 0.1798, |
| "num_input_tokens_seen": 41283008, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.8582703471183777, |
| "rewards/margins": 3.0465266704559326, |
| "rewards/rejected": -3.904797077178955, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.4443380416566914, |
| "grad_norm": 0.7103105187416077, |
| "learning_rate": 6.208260040378946e-06, |
| "logits/chosen": -1.1097609996795654, |
| "logits/rejected": -1.09922194480896, |
| "logps/chosen": -877.43603515625, |
| "logps/rejected": -892.90771484375, |
| "loss": 0.1762, |
| "num_input_tokens_seen": 41619520, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.48140305280685425, |
| "rewards/margins": 3.118536949157715, |
| "rewards/rejected": -3.5999398231506348, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.4481685420158008, |
| "grad_norm": 0.7059844136238098, |
| "learning_rate": 6.147440188611324e-06, |
| "logits/chosen": -1.1247339248657227, |
| "logits/rejected": -1.1204876899719238, |
| "logps/chosen": -969.3895263671875, |
| "logps/rejected": -1005.1538696289062, |
| "loss": 0.1582, |
| "num_input_tokens_seen": 41986368, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.6705014109611511, |
| "rewards/margins": 3.128121852874756, |
| "rewards/rejected": -3.7986230850219727, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.45199904237491023, |
| "grad_norm": 0.683424711227417, |
| "learning_rate": 6.0864405836192575e-06, |
| "logits/chosen": -1.1361314058303833, |
| "logits/rejected": -1.1225978136062622, |
| "logps/chosen": -896.8311157226562, |
| "logps/rejected": -937.1458129882812, |
| "loss": 0.1576, |
| "num_input_tokens_seen": 42347008, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.5880584716796875, |
| "rewards/margins": 3.3888418674468994, |
| "rewards/rejected": -3.976900339126587, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.45582954273401965, |
| "grad_norm": 0.6680045127868652, |
| "learning_rate": 6.025270781348055e-06, |
| "logits/chosen": -1.114820957183838, |
| "logits/rejected": -1.1184210777282715, |
| "logps/chosen": -940.8297119140625, |
| "logps/rejected": -953.9190673828125, |
| "loss": 0.1499, |
| "num_input_tokens_seen": 42698240, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.6581835746765137, |
| "rewards/margins": 3.291698932647705, |
| "rewards/rejected": -3.9498825073242188, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.459660043093129, |
| "grad_norm": 0.7042851448059082, |
| "learning_rate": 5.963940364405425e-06, |
| "logits/chosen": -1.1057615280151367, |
| "logits/rejected": -1.1188658475875854, |
| "logps/chosen": -935.0254516601562, |
| "logps/rejected": -929.1004028320312, |
| "loss": 0.1678, |
| "num_input_tokens_seen": 43050688, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.7741080522537231, |
| "rewards/margins": 3.301640033721924, |
| "rewards/rejected": -4.075747966766357, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.46349054345223845, |
| "grad_norm": 0.773540198802948, |
| "learning_rate": 5.902458940560304e-06, |
| "logits/chosen": -1.1311380863189697, |
| "logits/rejected": -1.1221275329589844, |
| "logps/chosen": -964.8699951171875, |
| "logps/rejected": -970.51171875, |
| "loss": 0.1805, |
| "num_input_tokens_seen": 43415744, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.7004190683364868, |
| "rewards/margins": 3.115402936935425, |
| "rewards/rejected": -3.815822124481201, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.46732104381134787, |
| "grad_norm": 0.6558127999305725, |
| "learning_rate": 5.8408361412377475e-06, |
| "logits/chosen": -1.1317683458328247, |
| "logits/rejected": -1.1370477676391602, |
| "logps/chosen": -915.1818237304688, |
| "logps/rejected": -931.9642333984375, |
| "loss": 0.1412, |
| "num_input_tokens_seen": 43769984, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.46988123655319214, |
| "rewards/margins": 3.3140010833740234, |
| "rewards/rejected": -3.7838823795318604, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.4711515441704573, |
| "grad_norm": 0.608486533164978, |
| "learning_rate": 5.779081620010104e-06, |
| "logits/chosen": -1.1101739406585693, |
| "logits/rejected": -1.1085362434387207, |
| "logps/chosen": -950.239501953125, |
| "logps/rejected": -966.5133056640625, |
| "loss": 0.1359, |
| "num_input_tokens_seen": 44125376, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.5609409213066101, |
| "rewards/margins": 3.2361843585968018, |
| "rewards/rejected": -3.7971253395080566, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.47498204452956666, |
| "grad_norm": 0.6579767465591431, |
| "learning_rate": 5.717205051084731e-06, |
| "logits/chosen": -1.12186861038208, |
| "logits/rejected": -1.119635820388794, |
| "logps/chosen": -970.2237548828125, |
| "logps/rejected": -1001.27880859375, |
| "loss": 0.159, |
| "num_input_tokens_seen": 44496896, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.5873010754585266, |
| "rewards/margins": 3.099958896636963, |
| "rewards/rejected": -3.687260150909424, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.4788125448886761, |
| "grad_norm": 0.6986194849014282, |
| "learning_rate": 5.655216127788472e-06, |
| "logits/chosen": -1.1113070249557495, |
| "logits/rejected": -1.1141784191131592, |
| "logps/chosen": -978.4093017578125, |
| "logps/rejected": -995.634765625, |
| "loss": 0.1569, |
| "num_input_tokens_seen": 44858496, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.5494914650917053, |
| "rewards/margins": 3.107781171798706, |
| "rewards/rejected": -3.6572725772857666, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.4826430452477855, |
| "grad_norm": 0.727834939956665, |
| "learning_rate": 5.593124561049141e-06, |
| "logits/chosen": -1.1294147968292236, |
| "logits/rejected": -1.1388496160507202, |
| "logps/chosen": -930.8519287109375, |
| "logps/rejected": -939.8653564453125, |
| "loss": 0.1527, |
| "num_input_tokens_seen": 45215680, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.6654321551322937, |
| "rewards/margins": 3.4994919300079346, |
| "rewards/rejected": -4.164923667907715, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.4864735456068949, |
| "grad_norm": 0.8437095880508423, |
| "learning_rate": 5.530940077874248e-06, |
| "logits/chosen": -1.1250782012939453, |
| "logits/rejected": -1.1162904500961304, |
| "logps/chosen": -915.8817138671875, |
| "logps/rejected": -932.8961181640625, |
| "loss": 0.1848, |
| "num_input_tokens_seen": 45575872, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.8156818151473999, |
| "rewards/margins": 2.9767351150512695, |
| "rewards/rejected": -3.792417049407959, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.4903040459660043, |
| "grad_norm": 0.6814424991607666, |
| "learning_rate": 5.468672419827208e-06, |
| "logits/chosen": -1.1594264507293701, |
| "logits/rejected": -1.150148868560791, |
| "logps/chosen": -945.5958862304688, |
| "logps/rejected": -989.4842529296875, |
| "loss": 0.1428, |
| "num_input_tokens_seen": 45951488, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.5141391754150391, |
| "rewards/margins": 3.470471143722534, |
| "rewards/rejected": -3.984610080718994, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.4941345463251137, |
| "grad_norm": 0.8450974822044373, |
| "learning_rate": 5.406331341501264e-06, |
| "logits/chosen": -1.1080952882766724, |
| "logits/rejected": -1.1018576622009277, |
| "logps/chosen": -936.5384521484375, |
| "logps/rejected": -965.4146728515625, |
| "loss": 0.1905, |
| "num_input_tokens_seen": 46313472, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.655163049697876, |
| "rewards/margins": 3.0373594760894775, |
| "rewards/rejected": -3.6925225257873535, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.49796504668422314, |
| "grad_norm": 0.7701146006584167, |
| "learning_rate": 5.34392660899138e-06, |
| "logits/chosen": -1.1277241706848145, |
| "logits/rejected": -1.1296637058258057, |
| "logps/chosen": -920.0867919921875, |
| "logps/rejected": -939.5274658203125, |
| "loss": 0.1683, |
| "num_input_tokens_seen": 46683136, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.764032244682312, |
| "rewards/margins": 3.104192018508911, |
| "rewards/rejected": -3.8682241439819336, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5017955470433325, |
| "grad_norm": 0.8093336820602417, |
| "learning_rate": 5.281467998364314e-06, |
| "logits/chosen": -1.1174012422561646, |
| "logits/rejected": -1.1064536571502686, |
| "logps/chosen": -930.655517578125, |
| "logps/rejected": -972.0757446289062, |
| "loss": 0.1818, |
| "num_input_tokens_seen": 47033152, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.6495652198791504, |
| "rewards/margins": 3.3375585079193115, |
| "rewards/rejected": -3.987123966217041, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.505626047402442, |
| "grad_norm": 0.6671730875968933, |
| "learning_rate": 5.218965294127155e-06, |
| "logits/chosen": -1.1446601152420044, |
| "logits/rejected": -1.1331120729446411, |
| "logps/chosen": -955.3788452148438, |
| "logps/rejected": -990.6216430664062, |
| "loss": 0.1355, |
| "num_input_tokens_seen": 47407552, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.708296537399292, |
| "rewards/margins": 3.5515570640563965, |
| "rewards/rejected": -4.259853363037109, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.5094565477615514, |
| "grad_norm": 0.7050759196281433, |
| "learning_rate": 5.156428287694508e-06, |
| "logits/chosen": -1.1224825382232666, |
| "logits/rejected": -1.1169312000274658, |
| "logps/chosen": -975.88671875, |
| "logps/rejected": -1000.4472045898438, |
| "loss": 0.1517, |
| "num_input_tokens_seen": 47777088, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.9042822122573853, |
| "rewards/margins": 3.1744394302368164, |
| "rewards/rejected": -4.078721523284912, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.5132870481206607, |
| "grad_norm": 0.6690219640731812, |
| "learning_rate": 5.093866775854618e-06, |
| "logits/chosen": -1.1281201839447021, |
| "logits/rejected": -1.118111252784729, |
| "logps/chosen": -950.8082885742188, |
| "logps/rejected": -967.8411865234375, |
| "loss": 0.1486, |
| "num_input_tokens_seen": 48146176, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.4145764708518982, |
| "rewards/margins": 3.500164031982422, |
| "rewards/rejected": -3.914740562438965, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.5171175484797702, |
| "grad_norm": 0.6712304353713989, |
| "learning_rate": 5.03129055923465e-06, |
| "logits/chosen": -1.152693510055542, |
| "logits/rejected": -1.1446518898010254, |
| "logps/chosen": -931.6079711914062, |
| "logps/rejected": -960.1605834960938, |
| "loss": 0.1411, |
| "num_input_tokens_seen": 48511488, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.683695912361145, |
| "rewards/margins": 3.693258285522461, |
| "rewards/rejected": -4.376954078674316, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.5209480488388796, |
| "grad_norm": 0.6588895320892334, |
| "learning_rate": 4.968709440765352e-06, |
| "logits/chosen": -1.1270694732666016, |
| "logits/rejected": -1.144219160079956, |
| "logps/chosen": -921.58251953125, |
| "logps/rejected": -936.573486328125, |
| "loss": 0.1288, |
| "num_input_tokens_seen": 48862336, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.6117515563964844, |
| "rewards/margins": 3.4521889686584473, |
| "rewards/rejected": -4.063940525054932, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.5247785491979889, |
| "grad_norm": 0.7437635660171509, |
| "learning_rate": 4.906133224145384e-06, |
| "logits/chosen": -1.1290967464447021, |
| "logits/rejected": -1.1110491752624512, |
| "logps/chosen": -939.525634765625, |
| "logps/rejected": -999.9234008789062, |
| "loss": 0.1461, |
| "num_input_tokens_seen": 49226048, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.5727760791778564, |
| "rewards/margins": 3.2480969429016113, |
| "rewards/rejected": -3.820873260498047, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.5286090495570984, |
| "grad_norm": 0.7835173010826111, |
| "learning_rate": 4.843571712305493e-06, |
| "logits/chosen": -1.116170883178711, |
| "logits/rejected": -1.123787760734558, |
| "logps/chosen": -927.2295532226562, |
| "logps/rejected": -945.8253173828125, |
| "loss": 0.1788, |
| "num_input_tokens_seen": 49581888, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.6745204925537109, |
| "rewards/margins": 3.3101115226745605, |
| "rewards/rejected": -3.9846320152282715, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.5324395499162078, |
| "grad_norm": 0.5597019791603088, |
| "learning_rate": 4.781034705872846e-06, |
| "logits/chosen": -1.1176984310150146, |
| "logits/rejected": -1.117140531539917, |
| "logps/chosen": -933.116455078125, |
| "logps/rejected": -970.36669921875, |
| "loss": 0.1061, |
| "num_input_tokens_seen": 49936512, |
| "rewards/accuracies": 0.9921875, |
| "rewards/chosen": -0.6796407103538513, |
| "rewards/margins": 3.7050676345825195, |
| "rewards/rejected": -4.384708404541016, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.5362700502753173, |
| "grad_norm": 0.615868091583252, |
| "learning_rate": 4.7185320016356865e-06, |
| "logits/chosen": -1.1357049942016602, |
| "logits/rejected": -1.1324001550674438, |
| "logps/chosen": -943.2921142578125, |
| "logps/rejected": -975.155517578125, |
| "loss": 0.1376, |
| "num_input_tokens_seen": 50309824, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.8460093140602112, |
| "rewards/margins": 3.435059070587158, |
| "rewards/rejected": -4.281068801879883, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.5401005506344266, |
| "grad_norm": 0.6606157422065735, |
| "learning_rate": 4.656073391008622e-06, |
| "logits/chosen": -1.1140908002853394, |
| "logits/rejected": -1.108850121498108, |
| "logps/chosen": -895.8781127929688, |
| "logps/rejected": -936.751220703125, |
| "loss": 0.1519, |
| "num_input_tokens_seen": 50659712, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.7830679416656494, |
| "rewards/margins": 3.3882458209991455, |
| "rewards/rejected": -4.171314239501953, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.543931050993536, |
| "grad_norm": 0.7001786231994629, |
| "learning_rate": 4.593668658498737e-06, |
| "logits/chosen": -1.149137020111084, |
| "logits/rejected": -1.1395349502563477, |
| "logps/chosen": -941.0504150390625, |
| "logps/rejected": -963.2518310546875, |
| "loss": 0.1491, |
| "num_input_tokens_seen": 51020992, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.6404644846916199, |
| "rewards/margins": 3.440929412841797, |
| "rewards/rejected": -4.081393718719482, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.5477615513526455, |
| "grad_norm": 0.6836583614349365, |
| "learning_rate": 4.531327580172794e-06, |
| "logits/chosen": -1.1339120864868164, |
| "logits/rejected": -1.1352245807647705, |
| "logps/chosen": -964.3081665039062, |
| "logps/rejected": -978.4052734375, |
| "loss": 0.1244, |
| "num_input_tokens_seen": 51378624, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.8491271734237671, |
| "rewards/margins": 3.6870431900024414, |
| "rewards/rejected": -4.536170482635498, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.5515920517117548, |
| "grad_norm": 0.7100102305412292, |
| "learning_rate": 4.469059922125753e-06, |
| "logits/chosen": -1.148242712020874, |
| "logits/rejected": -1.1385154724121094, |
| "logps/chosen": -998.3412475585938, |
| "logps/rejected": -1044.08935546875, |
| "loss": 0.1593, |
| "num_input_tokens_seen": 51760384, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.0328764915466309, |
| "rewards/margins": 3.6171908378601074, |
| "rewards/rejected": -4.650067329406738, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.5554225520708642, |
| "grad_norm": 0.7961605191230774, |
| "learning_rate": 4.4068754389508616e-06, |
| "logits/chosen": -1.1213231086730957, |
| "logits/rejected": -1.1205980777740479, |
| "logps/chosen": -946.205810546875, |
| "logps/rejected": -974.6113891601562, |
| "loss": 0.1374, |
| "num_input_tokens_seen": 52114304, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.7596480250358582, |
| "rewards/margins": 3.813307762145996, |
| "rewards/rejected": -4.572955131530762, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.5592530524299737, |
| "grad_norm": 0.7323804497718811, |
| "learning_rate": 4.34478387221153e-06, |
| "logits/chosen": -1.115981101989746, |
| "logits/rejected": -1.1200889348983765, |
| "logps/chosen": -942.8148193359375, |
| "logps/rejected": -965.3985595703125, |
| "loss": 0.1643, |
| "num_input_tokens_seen": 52486784, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -0.8299667239189148, |
| "rewards/margins": 3.4589433670043945, |
| "rewards/rejected": -4.288909912109375, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.5630835527890831, |
| "grad_norm": 0.5200232267379761, |
| "learning_rate": 4.282794948915271e-06, |
| "logits/chosen": -1.120424747467041, |
| "logits/rejected": -1.1166388988494873, |
| "logps/chosen": -906.1971435546875, |
| "logps/rejected": -953.6253662109375, |
| "loss": 0.0999, |
| "num_input_tokens_seen": 52831680, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.5943716764450073, |
| "rewards/margins": 3.9285550117492676, |
| "rewards/rejected": -4.522926330566406, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.5669140531481924, |
| "grad_norm": 0.5851389765739441, |
| "learning_rate": 4.220918379989898e-06, |
| "logits/chosen": -1.1159402132034302, |
| "logits/rejected": -1.1205713748931885, |
| "logps/chosen": -923.9879760742188, |
| "logps/rejected": -925.5596923828125, |
| "loss": 0.1095, |
| "num_input_tokens_seen": 53175680, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.8688675165176392, |
| "rewards/margins": 3.639319896697998, |
| "rewards/rejected": -4.508187770843506, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.5707445535073019, |
| "grad_norm": 0.6805673241615295, |
| "learning_rate": 4.159163858762255e-06, |
| "logits/chosen": -1.108860731124878, |
| "logits/rejected": -1.1102187633514404, |
| "logps/chosen": -937.15478515625, |
| "logps/rejected": -952.2811279296875, |
| "loss": 0.1384, |
| "num_input_tokens_seen": 53537024, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.9211418628692627, |
| "rewards/margins": 3.680717706680298, |
| "rewards/rejected": -4.601859092712402, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.5745750538664113, |
| "grad_norm": 0.631395161151886, |
| "learning_rate": 4.097541059439698e-06, |
| "logits/chosen": -1.1475733518600464, |
| "logits/rejected": -1.1469578742980957, |
| "logps/chosen": -928.3204956054688, |
| "logps/rejected": -947.0354614257812, |
| "loss": 0.1174, |
| "num_input_tokens_seen": 53894528, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.0530691146850586, |
| "rewards/margins": 3.8212642669677734, |
| "rewards/rejected": -4.874333381652832, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.5784055542255208, |
| "grad_norm": 0.7877935171127319, |
| "learning_rate": 4.036059635594578e-06, |
| "logits/chosen": -1.1463123559951782, |
| "logits/rejected": -1.1365106105804443, |
| "logps/chosen": -949.924072265625, |
| "logps/rejected": -962.1156005859375, |
| "loss": 0.1603, |
| "num_input_tokens_seen": 54262848, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.9331613779067993, |
| "rewards/margins": 3.1643664836883545, |
| "rewards/rejected": -4.097527980804443, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.5822360545846301, |
| "grad_norm": 0.5540483593940735, |
| "learning_rate": 3.974729218651946e-06, |
| "logits/chosen": -1.166491150856018, |
| "logits/rejected": -1.1675529479980469, |
| "logps/chosen": -940.0262451171875, |
| "logps/rejected": -979.6585083007812, |
| "loss": 0.1102, |
| "num_input_tokens_seen": 54635776, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.0611553192138672, |
| "rewards/margins": 3.9694066047668457, |
| "rewards/rejected": -5.030562400817871, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.5860665549437395, |
| "grad_norm": 0.6366962790489197, |
| "learning_rate": 3.913559416380743e-06, |
| "logits/chosen": -1.135746955871582, |
| "logits/rejected": -1.1493985652923584, |
| "logps/chosen": -921.271484375, |
| "logps/rejected": -946.207763671875, |
| "loss": 0.1277, |
| "num_input_tokens_seen": 54999744, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.9022237658500671, |
| "rewards/margins": 3.6466073989868164, |
| "rewards/rejected": -4.548830986022949, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.589897055302849, |
| "grad_norm": 0.4817214012145996, |
| "learning_rate": 3.852559811388676e-06, |
| "logits/chosen": -1.1166200637817383, |
| "logits/rejected": -1.1204478740692139, |
| "logps/chosen": -932.076416015625, |
| "logps/rejected": -951.6412353515625, |
| "loss": 0.0814, |
| "num_input_tokens_seen": 55337280, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.8710342049598694, |
| "rewards/margins": 4.16388463973999, |
| "rewards/rejected": -5.034918785095215, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.5937275556619583, |
| "grad_norm": 0.7406041026115417, |
| "learning_rate": 3.791739959621054e-06, |
| "logits/chosen": -1.1301652193069458, |
| "logits/rejected": -1.1277787685394287, |
| "logps/chosen": -939.067626953125, |
| "logps/rejected": -976.2159423828125, |
| "loss": 0.1583, |
| "num_input_tokens_seen": 55702336, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.148180603981018, |
| "rewards/margins": 3.349360704421997, |
| "rewards/rejected": -4.497541427612305, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.5975580560210677, |
| "grad_norm": 0.6545127034187317, |
| "learning_rate": 3.7311093888637906e-06, |
| "logits/chosen": -1.1367216110229492, |
| "logits/rejected": -1.135441541671753, |
| "logps/chosen": -947.7169799804688, |
| "logps/rejected": -957.501953125, |
| "loss": 0.1345, |
| "num_input_tokens_seen": 56070336, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.793830394744873, |
| "rewards/margins": 3.778103828430176, |
| "rewards/rejected": -4.571934223175049, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.6013885563801772, |
| "grad_norm": 0.6249314546585083, |
| "learning_rate": 3.670677597250819e-06, |
| "logits/chosen": -1.1367524862289429, |
| "logits/rejected": -1.1322413682937622, |
| "logps/chosen": -919.898193359375, |
| "logps/rejected": -953.7801513671875, |
| "loss": 0.1229, |
| "num_input_tokens_seen": 56428800, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.0889718532562256, |
| "rewards/margins": 3.635096788406372, |
| "rewards/rejected": -4.724068641662598, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.6052190567392866, |
| "grad_norm": 0.6867994070053101, |
| "learning_rate": 3.6104540517761594e-06, |
| "logits/chosen": -1.1208783388137817, |
| "logits/rejected": -1.1281899213790894, |
| "logps/chosen": -942.1203002929688, |
| "logps/rejected": -941.297607421875, |
| "loss": 0.1324, |
| "num_input_tokens_seen": 56783040, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.8960638046264648, |
| "rewards/margins": 3.8797450065612793, |
| "rewards/rejected": -4.775808811187744, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.609049557098396, |
| "grad_norm": 0.5042504668235779, |
| "learning_rate": 3.55044818681085e-06, |
| "logits/chosen": -1.1335363388061523, |
| "logits/rejected": -1.1184139251708984, |
| "logps/chosen": -954.19921875, |
| "logps/rejected": -998.2352905273438, |
| "loss": 0.0872, |
| "num_input_tokens_seen": 57145920, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.7512755990028381, |
| "rewards/margins": 4.050450325012207, |
| "rewards/rejected": -4.801725387573242, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.6128800574575054, |
| "grad_norm": 0.6282093524932861, |
| "learning_rate": 3.4906694026250075e-06, |
| "logits/chosen": -1.1352434158325195, |
| "logits/rejected": -1.1456270217895508, |
| "logps/chosen": -935.1751708984375, |
| "logps/rejected": -959.360107421875, |
| "loss": 0.1244, |
| "num_input_tokens_seen": 57502528, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.9671937227249146, |
| "rewards/margins": 3.746589422225952, |
| "rewards/rejected": -4.713783264160156, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6167105578166148, |
| "grad_norm": 0.7335207462310791, |
| "learning_rate": 3.431127063915213e-06, |
| "logits/chosen": -1.1380847692489624, |
| "logits/rejected": -1.1418529748916626, |
| "logps/chosen": -939.71142578125, |
| "logps/rejected": -956.5693359375, |
| "loss": 0.1404, |
| "num_input_tokens_seen": 57864896, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.9701119065284729, |
| "rewards/margins": 3.5857200622558594, |
| "rewards/rejected": -4.5558319091796875, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.6205410581757242, |
| "grad_norm": 0.525834321975708, |
| "learning_rate": 3.371830498337475e-06, |
| "logits/chosen": -1.1229634284973145, |
| "logits/rejected": -1.1113239526748657, |
| "logps/chosen": -906.3638916015625, |
| "logps/rejected": -968.2791748046875, |
| "loss": 0.0902, |
| "num_input_tokens_seen": 58218496, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.853689432144165, |
| "rewards/margins": 4.11014986038208, |
| "rewards/rejected": -4.963839054107666, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.6243715585348336, |
| "grad_norm": 0.597028911113739, |
| "learning_rate": 3.3127889950460094e-06, |
| "logits/chosen": -1.115204095840454, |
| "logits/rejected": -1.1108593940734863, |
| "logps/chosen": -898.75390625, |
| "logps/rejected": -959.1620483398438, |
| "loss": 0.1118, |
| "num_input_tokens_seen": 58561152, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.8724937438964844, |
| "rewards/margins": 3.783143997192383, |
| "rewards/rejected": -4.655637741088867, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.628202058893943, |
| "grad_norm": 0.773404598236084, |
| "learning_rate": 3.254011803238026e-06, |
| "logits/chosen": -1.1355056762695312, |
| "logits/rejected": -1.1327035427093506, |
| "logps/chosen": -964.1258544921875, |
| "logps/rejected": -1013.5973510742188, |
| "loss": 0.154, |
| "num_input_tokens_seen": 58925504, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.280595302581787, |
| "rewards/margins": 3.811772346496582, |
| "rewards/rejected": -5.092367649078369, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.6320325592530525, |
| "grad_norm": 0.6448091268539429, |
| "learning_rate": 3.195508130704795e-06, |
| "logits/chosen": -1.1301665306091309, |
| "logits/rejected": -1.1308307647705078, |
| "logps/chosen": -919.0938720703125, |
| "logps/rejected": -935.1650390625, |
| "loss": 0.1232, |
| "num_input_tokens_seen": 59269120, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.9420739412307739, |
| "rewards/margins": 3.73829984664917, |
| "rewards/rejected": -4.680373668670654, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.6358630596121618, |
| "grad_norm": 0.717473566532135, |
| "learning_rate": 3.1372871423891894e-06, |
| "logits/chosen": -1.1233844757080078, |
| "logits/rejected": -1.1225996017456055, |
| "logps/chosen": -926.3175659179688, |
| "logps/rejected": -951.1182861328125, |
| "loss": 0.1412, |
| "num_input_tokens_seen": 59620928, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -0.8571826219558716, |
| "rewards/margins": 3.7891440391540527, |
| "rewards/rejected": -4.646326065063477, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.6396935599712712, |
| "grad_norm": 0.5268615484237671, |
| "learning_rate": 3.079357958949946e-06, |
| "logits/chosen": -1.1460094451904297, |
| "logits/rejected": -1.1440532207489014, |
| "logps/chosen": -923.2747802734375, |
| "logps/rejected": -965.303955078125, |
| "loss": 0.0887, |
| "num_input_tokens_seen": 59974656, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.8316411972045898, |
| "rewards/margins": 4.237669944763184, |
| "rewards/rejected": -5.069311618804932, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.6435240603303807, |
| "grad_norm": 0.7504249811172485, |
| "learning_rate": 3.021729655332858e-06, |
| "logits/chosen": -1.1353933811187744, |
| "logits/rejected": -1.126386046409607, |
| "logps/chosen": -941.4724731445312, |
| "logps/rejected": -971.0452880859375, |
| "loss": 0.1555, |
| "num_input_tokens_seen": 60329216, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -0.8113993406295776, |
| "rewards/margins": 3.449709892272949, |
| "rewards/rejected": -4.261109352111816, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.64735456068949, |
| "grad_norm": 0.5045068264007568, |
| "learning_rate": 2.9644112593491315e-06, |
| "logits/chosen": -1.113010048866272, |
| "logits/rejected": -1.1124825477600098, |
| "logps/chosen": -913.4754638671875, |
| "logps/rejected": -939.3057861328125, |
| "loss": 0.086, |
| "num_input_tokens_seen": 60686464, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.6832062005996704, |
| "rewards/margins": 4.223135948181152, |
| "rewards/rejected": -4.906341552734375, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.6511850610485995, |
| "grad_norm": 0.7200180888175964, |
| "learning_rate": 2.90741175026113e-06, |
| "logits/chosen": -1.143190622329712, |
| "logits/rejected": -1.1372324228286743, |
| "logps/chosen": -933.2850952148438, |
| "logps/rejected": -970.13232421875, |
| "loss": 0.134, |
| "num_input_tokens_seen": 61046144, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.9384927153587341, |
| "rewards/margins": 4.193242073059082, |
| "rewards/rejected": -5.131734848022461, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.6550155614077089, |
| "grad_norm": 0.770698070526123, |
| "learning_rate": 2.850740057375716e-06, |
| "logits/chosen": -1.1460022926330566, |
| "logits/rejected": -1.1353541612625122, |
| "logps/chosen": -906.2474365234375, |
| "logps/rejected": -944.6781005859375, |
| "loss": 0.1397, |
| "num_input_tokens_seen": 61395776, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.2490543127059937, |
| "rewards/margins": 3.7003769874572754, |
| "rewards/rejected": -4.949431419372559, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.6588460617668183, |
| "grad_norm": 0.7854964137077332, |
| "learning_rate": 2.7944050586454215e-06, |
| "logits/chosen": -1.1241507530212402, |
| "logits/rejected": -1.1199979782104492, |
| "logps/chosen": -881.9456787109375, |
| "logps/rejected": -925.2673950195312, |
| "loss": 0.1393, |
| "num_input_tokens_seen": 61739072, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -1.119002342224121, |
| "rewards/margins": 4.081979274749756, |
| "rewards/rejected": -5.200981616973877, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.6626765621259277, |
| "grad_norm": 0.614799976348877, |
| "learning_rate": 2.7384155792776724e-06, |
| "logits/chosen": -1.1314318180084229, |
| "logits/rejected": -1.1282424926757812, |
| "logps/chosen": -937.385498046875, |
| "logps/rejected": -974.9067993164062, |
| "loss": 0.107, |
| "num_input_tokens_seen": 62091840, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.2679835557937622, |
| "rewards/margins": 4.420315265655518, |
| "rewards/rejected": -5.688299179077148, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.6665070624850371, |
| "grad_norm": 0.7183716297149658, |
| "learning_rate": 2.682780390352262e-06, |
| "logits/chosen": -1.147839903831482, |
| "logits/rejected": -1.1522760391235352, |
| "logps/chosen": -915.7791748046875, |
| "logps/rejected": -955.96240234375, |
| "loss": 0.1178, |
| "num_input_tokens_seen": 62452416, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.1561710834503174, |
| "rewards/margins": 4.244651794433594, |
| "rewards/rejected": -5.40082311630249, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.6703375628441465, |
| "grad_norm": 0.7553685903549194, |
| "learning_rate": 2.627508207447308e-06, |
| "logits/chosen": -1.1484794616699219, |
| "logits/rejected": -1.1376214027404785, |
| "logps/chosen": -959.5919799804688, |
| "logps/rejected": -1002.1358642578125, |
| "loss": 0.1382, |
| "num_input_tokens_seen": 62826432, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.2018121480941772, |
| "rewards/margins": 4.0351972579956055, |
| "rewards/rejected": -5.237009048461914, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.674168063203256, |
| "grad_norm": 0.7371348738670349, |
| "learning_rate": 2.5726076892739127e-06, |
| "logits/chosen": -1.133622646331787, |
| "logits/rejected": -1.1346787214279175, |
| "logps/chosen": -981.44091796875, |
| "logps/rejected": -1008.0592041015625, |
| "loss": 0.1451, |
| "num_input_tokens_seen": 63188736, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -0.9973090887069702, |
| "rewards/margins": 3.770475387573242, |
| "rewards/rejected": -4.767784118652344, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.6779985635623653, |
| "grad_norm": 0.6027622222900391, |
| "learning_rate": 2.5180874363197217e-06, |
| "logits/chosen": -1.1473917961120605, |
| "logits/rejected": -1.136796474456787, |
| "logps/chosen": -966.965087890625, |
| "logps/rejected": -992.8527221679688, |
| "loss": 0.1113, |
| "num_input_tokens_seen": 63559040, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.037949562072754, |
| "rewards/margins": 3.698408365249634, |
| "rewards/rejected": -4.736357688903809, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.6818290639214747, |
| "grad_norm": 0.6934034824371338, |
| "learning_rate": 2.463955989501607e-06, |
| "logits/chosen": -1.1472060680389404, |
| "logits/rejected": -1.142443299293518, |
| "logps/chosen": -948.6500244140625, |
| "logps/rejected": -976.03955078125, |
| "loss": 0.1304, |
| "num_input_tokens_seen": 63918080, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.2286975383758545, |
| "rewards/margins": 4.057483673095703, |
| "rewards/rejected": -5.286181449890137, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.6856595642805842, |
| "grad_norm": 0.574459969997406, |
| "learning_rate": 2.41022182882768e-06, |
| "logits/chosen": -1.1256879568099976, |
| "logits/rejected": -1.1263935565948486, |
| "logps/chosen": -920.3141479492188, |
| "logps/rejected": -937.0470581054688, |
| "loss": 0.1064, |
| "num_input_tokens_seen": 64264384, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.0772805213928223, |
| "rewards/margins": 3.8612425327301025, |
| "rewards/rejected": -4.938523292541504, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.6894900646396935, |
| "grad_norm": 0.6900708079338074, |
| "learning_rate": 2.356893372068855e-06, |
| "logits/chosen": -1.146414041519165, |
| "logits/rejected": -1.1333107948303223, |
| "logps/chosen": -920.6194458007812, |
| "logps/rejected": -948.978515625, |
| "loss": 0.1286, |
| "num_input_tokens_seen": 64627520, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.1199493408203125, |
| "rewards/margins": 4.144634246826172, |
| "rewards/rejected": -5.264584064483643, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.693320564998803, |
| "grad_norm": 0.6781615018844604, |
| "learning_rate": 2.3039789734401524e-06, |
| "logits/chosen": -1.1684128046035767, |
| "logits/rejected": -1.156698226928711, |
| "logps/chosen": -886.8367919921875, |
| "logps/rejected": -941.868896484375, |
| "loss": 0.1146, |
| "num_input_tokens_seen": 64985472, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.0930626392364502, |
| "rewards/margins": 4.096789836883545, |
| "rewards/rejected": -5.189852714538574, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.6971510653579124, |
| "grad_norm": 0.6015796661376953, |
| "learning_rate": 2.251486922291957e-06, |
| "logits/chosen": -1.1457183361053467, |
| "logits/rejected": -1.1400107145309448, |
| "logps/chosen": -878.5657958984375, |
| "logps/rejected": -924.116943359375, |
| "loss": 0.1146, |
| "num_input_tokens_seen": 65333056, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.3387953042984009, |
| "rewards/margins": 3.838289976119995, |
| "rewards/rejected": -5.177084922790527, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.7009815657170217, |
| "grad_norm": 0.7410080432891846, |
| "learning_rate": 2.1994254418114524e-06, |
| "logits/chosen": -1.1427900791168213, |
| "logits/rejected": -1.1322269439697266, |
| "logps/chosen": -970.937255859375, |
| "logps/rejected": -1017.3334350585938, |
| "loss": 0.1325, |
| "num_input_tokens_seen": 65701376, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.2159329652786255, |
| "rewards/margins": 3.69895601272583, |
| "rewards/rejected": -4.914888858795166, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.7048120660761312, |
| "grad_norm": 0.6679302453994751, |
| "learning_rate": 2.147802687734409e-06, |
| "logits/chosen": -1.124457597732544, |
| "logits/rejected": -1.1152386665344238, |
| "logps/chosen": -939.2366943359375, |
| "logps/rejected": -980.7069702148438, |
| "loss": 0.1278, |
| "num_input_tokens_seen": 66057152, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.2831361293792725, |
| "rewards/margins": 3.8988122940063477, |
| "rewards/rejected": -5.181948661804199, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.7086425664352406, |
| "grad_norm": 0.6172199249267578, |
| "learning_rate": 2.0966267470675273e-06, |
| "logits/chosen": -1.1368470191955566, |
| "logits/rejected": -1.1430565118789673, |
| "logps/chosen": -935.0543212890625, |
| "logps/rejected": -954.5626220703125, |
| "loss": 0.1137, |
| "num_input_tokens_seen": 66413248, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.1827101707458496, |
| "rewards/margins": 4.264597415924072, |
| "rewards/rejected": -5.447307586669922, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.71247306679435, |
| "grad_norm": 0.7237995862960815, |
| "learning_rate": 2.0459056368215786e-06, |
| "logits/chosen": -1.142282247543335, |
| "logits/rejected": -1.1388722658157349, |
| "logps/chosen": -920.2362060546875, |
| "logps/rejected": -950.0621337890625, |
| "loss": 0.1468, |
| "num_input_tokens_seen": 66777600, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -1.1261134147644043, |
| "rewards/margins": 3.742077350616455, |
| "rewards/rejected": -4.868190765380859, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.7163035671534594, |
| "grad_norm": 0.8806514739990234, |
| "learning_rate": 1.9956473027554846e-06, |
| "logits/chosen": -1.1200953722000122, |
| "logits/rejected": -1.12364661693573, |
| "logps/chosen": -913.1375732421875, |
| "logps/rejected": -942.7782592773438, |
| "loss": 0.1509, |
| "num_input_tokens_seen": 67126784, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.018437385559082, |
| "rewards/margins": 3.621182918548584, |
| "rewards/rejected": -4.639620304107666, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.7201340675125688, |
| "grad_norm": 0.6537996530532837, |
| "learning_rate": 1.9458596181315643e-06, |
| "logits/chosen": -1.1244375705718994, |
| "logits/rejected": -1.1293549537658691, |
| "logps/chosen": -950.2227783203125, |
| "logps/rejected": -959.3658447265625, |
| "loss": 0.1212, |
| "num_input_tokens_seen": 67486848, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.301206111907959, |
| "rewards/margins": 3.810688018798828, |
| "rewards/rejected": -5.111893653869629, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.7239645678716783, |
| "grad_norm": 0.5897997617721558, |
| "learning_rate": 1.8965503824821496e-06, |
| "logits/chosen": -1.115061640739441, |
| "logits/rejected": -1.113053321838379, |
| "logps/chosen": -918.2599487304688, |
| "logps/rejected": -964.2479248046875, |
| "loss": 0.1096, |
| "num_input_tokens_seen": 67830400, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.3825678825378418, |
| "rewards/margins": 4.108716011047363, |
| "rewards/rejected": -5.491284370422363, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.7277950682307877, |
| "grad_norm": 0.7189119458198547, |
| "learning_rate": 1.84772732038774e-06, |
| "logits/chosen": -1.1144726276397705, |
| "logits/rejected": -1.1201146841049194, |
| "logps/chosen": -930.0636596679688, |
| "logps/rejected": -959.1786499023438, |
| "loss": 0.156, |
| "num_input_tokens_seen": 68184448, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.131813406944275, |
| "rewards/margins": 3.7312142848968506, |
| "rewards/rejected": -4.863027572631836, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.731625568589897, |
| "grad_norm": 0.5770811438560486, |
| "learning_rate": 1.7993980802668947e-06, |
| "logits/chosen": -1.1150379180908203, |
| "logits/rejected": -1.1045951843261719, |
| "logps/chosen": -927.9925537109375, |
| "logps/rejected": -981.8740234375, |
| "loss": 0.105, |
| "num_input_tokens_seen": 68536640, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.8180875778198242, |
| "rewards/margins": 4.335508823394775, |
| "rewards/rejected": -5.153596878051758, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.7354560689490065, |
| "grad_norm": 0.6197463870048523, |
| "learning_rate": 1.7515702331780753e-06, |
| "logits/chosen": -1.1397392749786377, |
| "logits/rejected": -1.1332669258117676, |
| "logps/chosen": -888.0272216796875, |
| "logps/rejected": -920.6478271484375, |
| "loss": 0.124, |
| "num_input_tokens_seen": 68881216, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.044264793395996, |
| "rewards/margins": 3.917402744293213, |
| "rewards/rejected": -4.961667060852051, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.7392865693081159, |
| "grad_norm": 0.6910933256149292, |
| "learning_rate": 1.7042512716335873e-06, |
| "logits/chosen": -1.144860863685608, |
| "logits/rejected": -1.1406577825546265, |
| "logps/chosen": -968.2261962890625, |
| "logps/rejected": -1015.7372436523438, |
| "loss": 0.1209, |
| "num_input_tokens_seen": 69258368, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.4211889505386353, |
| "rewards/margins": 3.7208251953125, |
| "rewards/rejected": -5.142014026641846, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.7431170696672252, |
| "grad_norm": 0.591352105140686, |
| "learning_rate": 1.6574486084258369e-06, |
| "logits/chosen": -1.1517504453659058, |
| "logits/rejected": -1.1535696983337402, |
| "logps/chosen": -941.3185424804688, |
| "logps/rejected": -941.9823608398438, |
| "loss": 0.1065, |
| "num_input_tokens_seen": 69617152, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.0724523067474365, |
| "rewards/margins": 4.041698932647705, |
| "rewards/rejected": -5.114151477813721, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.7469475700263347, |
| "grad_norm": 0.6189924478530884, |
| "learning_rate": 1.6111695754660667e-06, |
| "logits/chosen": -1.1325167417526245, |
| "logits/rejected": -1.1246585845947266, |
| "logps/chosen": -957.8402099609375, |
| "logps/rejected": -987.0758056640625, |
| "loss": 0.1183, |
| "num_input_tokens_seen": 69979072, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.395351767539978, |
| "rewards/margins": 4.022106170654297, |
| "rewards/rejected": -5.4174580574035645, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.7507780703854441, |
| "grad_norm": 0.48755350708961487, |
| "learning_rate": 1.5654214226357822e-06, |
| "logits/chosen": -1.1281023025512695, |
| "logits/rejected": -1.1127030849456787, |
| "logps/chosen": -953.8341064453125, |
| "logps/rejected": -1023.0802001953125, |
| "loss": 0.0792, |
| "num_input_tokens_seen": 70344768, |
| "rewards/accuracies": 0.9921875, |
| "rewards/chosen": -1.2984752655029297, |
| "rewards/margins": 3.9046237468719482, |
| "rewards/rejected": -5.203099250793457, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.7546085707445535, |
| "grad_norm": 0.7763097882270813, |
| "learning_rate": 1.5202113166510058e-06, |
| "logits/chosen": -1.1633872985839844, |
| "logits/rejected": -1.1649155616760254, |
| "logps/chosen": -940.2249755859375, |
| "logps/rejected": -974.9447021484375, |
| "loss": 0.1402, |
| "num_input_tokens_seen": 70721984, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.1531100273132324, |
| "rewards/margins": 4.258032321929932, |
| "rewards/rejected": -5.411142349243164, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.7584390711036629, |
| "grad_norm": 0.5113739371299744, |
| "learning_rate": 1.475546339939568e-06, |
| "logits/chosen": -1.1158440113067627, |
| "logits/rejected": -1.1148247718811035, |
| "logps/chosen": -949.8948974609375, |
| "logps/rejected": -1000.0040283203125, |
| "loss": 0.0939, |
| "num_input_tokens_seen": 71082368, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.1646056175231934, |
| "rewards/margins": 3.9723587036132812, |
| "rewards/rejected": -5.136964797973633, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.7622695714627723, |
| "grad_norm": 0.655764102935791, |
| "learning_rate": 1.4314334895316095e-06, |
| "logits/chosen": -1.1430144309997559, |
| "logits/rejected": -1.1407995223999023, |
| "logps/chosen": -922.4736328125, |
| "logps/rejected": -955.0093383789062, |
| "loss": 0.1234, |
| "num_input_tokens_seen": 71436928, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1026368141174316, |
| "rewards/margins": 4.110426425933838, |
| "rewards/rejected": -5.2130632400512695, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.7661000718218818, |
| "grad_norm": 0.5658289790153503, |
| "learning_rate": 1.3878796759634544e-06, |
| "logits/chosen": -1.1523005962371826, |
| "logits/rejected": -1.1472749710083008, |
| "logps/chosen": -942.7454833984375, |
| "logps/rejected": -991.8628540039062, |
| "loss": 0.1007, |
| "num_input_tokens_seen": 71801024, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.088900089263916, |
| "rewards/margins": 4.152194976806641, |
| "rewards/rejected": -5.241095542907715, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.7699305721809911, |
| "grad_norm": 0.6669675707817078, |
| "learning_rate": 1.3448917221950264e-06, |
| "logits/chosen": -1.1476324796676636, |
| "logits/rejected": -1.1498050689697266, |
| "logps/chosen": -956.4420166015625, |
| "logps/rejected": -968.7861328125, |
| "loss": 0.1182, |
| "num_input_tokens_seen": 72172800, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1832457780838013, |
| "rewards/margins": 3.7966558933258057, |
| "rewards/rejected": -4.9799017906188965, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.7737610725401005, |
| "grad_norm": 0.5420801639556885, |
| "learning_rate": 1.3024763625410025e-06, |
| "logits/chosen": -1.133588194847107, |
| "logits/rejected": -1.1364024877548218, |
| "logps/chosen": -943.519775390625, |
| "logps/rejected": -960.7150268554688, |
| "loss": 0.0918, |
| "num_input_tokens_seen": 72536000, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.925828218460083, |
| "rewards/margins": 4.087068557739258, |
| "rewards/rejected": -5.012896537780762, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.77759157289921, |
| "grad_norm": 0.6389860510826111, |
| "learning_rate": 1.2606402416158391e-06, |
| "logits/chosen": -1.1481473445892334, |
| "logits/rejected": -1.1378352642059326, |
| "logps/chosen": -939.3446655273438, |
| "logps/rejected": -977.73876953125, |
| "loss": 0.1069, |
| "num_input_tokens_seen": 72900672, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.294952154159546, |
| "rewards/margins": 4.067165374755859, |
| "rewards/rejected": -5.362117767333984, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.7814220732583194, |
| "grad_norm": 0.6677370667457581, |
| "learning_rate": 1.2193899132928539e-06, |
| "logits/chosen": -1.1154191493988037, |
| "logits/rejected": -1.1181483268737793, |
| "logps/chosen": -905.6656494140625, |
| "logps/rejected": -916.131103515625, |
| "loss": 0.1164, |
| "num_input_tokens_seen": 73242240, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1728456020355225, |
| "rewards/margins": 4.02401065826416, |
| "rewards/rejected": -5.1968560218811035, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.7852525736174287, |
| "grad_norm": 0.6846520900726318, |
| "learning_rate": 1.1787318396775188e-06, |
| "logits/chosen": -1.1279726028442383, |
| "logits/rejected": -1.1289410591125488, |
| "logps/chosen": -945.6762084960938, |
| "logps/rejected": -985.7901611328125, |
| "loss": 0.1268, |
| "num_input_tokens_seen": 73604288, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.2765809297561646, |
| "rewards/margins": 4.0222930908203125, |
| "rewards/rejected": -5.2988739013671875, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.7890830739765382, |
| "grad_norm": 0.7878302335739136, |
| "learning_rate": 1.138672390095143e-06, |
| "logits/chosen": -1.1450525522232056, |
| "logits/rejected": -1.1443026065826416, |
| "logps/chosen": -924.56103515625, |
| "logps/rejected": -933.9475708007812, |
| "loss": 0.1272, |
| "num_input_tokens_seen": 73953984, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.205392837524414, |
| "rewards/margins": 4.218493938446045, |
| "rewards/rejected": -5.423886775970459, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.7929135743356476, |
| "grad_norm": 0.7352907061576843, |
| "learning_rate": 1.0992178400930753e-06, |
| "logits/chosen": -1.1588523387908936, |
| "logits/rejected": -1.1516337394714355, |
| "logps/chosen": -935.2378540039062, |
| "logps/rejected": -979.25244140625, |
| "loss": 0.1464, |
| "num_input_tokens_seen": 74322752, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.1177600622177124, |
| "rewards/margins": 4.066137313842773, |
| "rewards/rejected": -5.183897018432617, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.796744074694757, |
| "grad_norm": 0.5740320682525635, |
| "learning_rate": 1.0603743704575992e-06, |
| "logits/chosen": -1.1424620151519775, |
| "logits/rejected": -1.1347261667251587, |
| "logps/chosen": -982.45751953125, |
| "logps/rejected": -1003.0057373046875, |
| "loss": 0.0975, |
| "num_input_tokens_seen": 74692672, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.184654712677002, |
| "rewards/margins": 4.110013484954834, |
| "rewards/rejected": -5.294668197631836, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.8005745750538664, |
| "grad_norm": 0.7261523008346558, |
| "learning_rate": 1.0221480662456845e-06, |
| "logits/chosen": -1.1375901699066162, |
| "logits/rejected": -1.1277306079864502, |
| "logps/chosen": -949.3741455078125, |
| "logps/rejected": -985.18896484375, |
| "loss": 0.1429, |
| "num_input_tokens_seen": 75043648, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.4194461107254028, |
| "rewards/margins": 3.990281105041504, |
| "rewards/rejected": -5.409727096557617, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.8044050754129758, |
| "grad_norm": 0.5886514782905579, |
| "learning_rate": 9.845449158317216e-07, |
| "logits/chosen": -1.1304702758789062, |
| "logits/rejected": -1.1260240077972412, |
| "logps/chosen": -939.6072998046875, |
| "logps/rejected": -961.4407348632812, |
| "loss": 0.1222, |
| "num_input_tokens_seen": 75408832, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1726312637329102, |
| "rewards/margins": 3.7976341247558594, |
| "rewards/rejected": -4.9702653884887695, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.8082355757720853, |
| "grad_norm": 0.5627762079238892, |
| "learning_rate": 9.475708099694125e-07, |
| "logits/chosen": -1.1464478969573975, |
| "logits/rejected": -1.1441315412521362, |
| "logps/chosen": -963.1434936523438, |
| "logps/rejected": -995.056884765625, |
| "loss": 0.09, |
| "num_input_tokens_seen": 75767744, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.3474948406219482, |
| "rewards/margins": 4.602136135101318, |
| "rewards/rejected": -5.949631214141846, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.8120660761311946, |
| "grad_norm": 0.8463510274887085, |
| "learning_rate": 9.112315408689415e-07, |
| "logits/chosen": -1.1504496335983276, |
| "logits/rejected": -1.145203709602356, |
| "logps/chosen": -953.384521484375, |
| "logps/rejected": -979.9056396484375, |
| "loss": 0.151, |
| "num_input_tokens_seen": 76143744, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.280893087387085, |
| "rewards/margins": 3.798523426055908, |
| "rewards/rejected": -5.079416275024414, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.815896576490304, |
| "grad_norm": 0.6893050670623779, |
| "learning_rate": 8.755328012896002e-07, |
| "logits/chosen": -1.1258955001831055, |
| "logits/rejected": -1.1326565742492676, |
| "logps/chosen": -915.5339965820312, |
| "logps/rejected": -946.0889892578125, |
| "loss": 0.1176, |
| "num_input_tokens_seen": 76490112, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.3296691179275513, |
| "rewards/margins": 4.028443336486816, |
| "rewards/rejected": -5.358112335205078, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.8197270768494135, |
| "grad_norm": 0.7727059125900269, |
| "learning_rate": 8.404801836479809e-07, |
| "logits/chosen": -1.1532236337661743, |
| "logits/rejected": -1.1491987705230713, |
| "logps/chosen": -971.9912109375, |
| "logps/rejected": -1012.7716064453125, |
| "loss": 0.1373, |
| "num_input_tokens_seen": 76866048, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.4460994005203247, |
| "rewards/margins": 4.311342716217041, |
| "rewards/rejected": -5.757441997528076, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.8235575772085229, |
| "grad_norm": 0.6790155172348022, |
| "learning_rate": 8.060791791418887e-07, |
| "logits/chosen": -1.1398448944091797, |
| "logits/rejected": -1.1423518657684326, |
| "logps/chosen": -952.3850708007812, |
| "logps/rejected": -974.2009887695312, |
| "loss": 0.1295, |
| "num_input_tokens_seen": 77235840, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.5031237602233887, |
| "rewards/margins": 4.125448703765869, |
| "rewards/rejected": -5.628572463989258, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.8273880775676322, |
| "grad_norm": 0.6497118473052979, |
| "learning_rate": 7.723351768901172e-07, |
| "logits/chosen": -1.125891923904419, |
| "logits/rejected": -1.1162381172180176, |
| "logps/chosen": -928.42919921875, |
| "logps/rejected": -950.609375, |
| "loss": 0.1176, |
| "num_input_tokens_seen": 77590592, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.1432021856307983, |
| "rewards/margins": 4.01569938659668, |
| "rewards/rejected": -5.158901214599609, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.8312185779267417, |
| "grad_norm": 0.6960030198097229, |
| "learning_rate": 7.392534630882092e-07, |
| "logits/chosen": -1.1320910453796387, |
| "logits/rejected": -1.1252555847167969, |
| "logps/chosen": -906.5370483398438, |
| "logps/rejected": -943.4966430664062, |
| "loss": 0.1177, |
| "num_input_tokens_seen": 77947072, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.3496477603912354, |
| "rewards/margins": 3.8906993865966797, |
| "rewards/rejected": -5.240346908569336, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.8350490782858511, |
| "grad_norm": 0.5776065587997437, |
| "learning_rate": 7.06839220180342e-07, |
| "logits/chosen": -1.124521017074585, |
| "logits/rejected": -1.1157386302947998, |
| "logps/chosen": -957.765380859375, |
| "logps/rejected": -1006.8399658203125, |
| "loss": 0.1055, |
| "num_input_tokens_seen": 78299264, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.0568280220031738, |
| "rewards/margins": 4.409466743469238, |
| "rewards/rejected": -5.46629524230957, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.8388795786449605, |
| "grad_norm": 0.7293028235435486, |
| "learning_rate": 6.750975260474718e-07, |
| "logits/chosen": -1.125768780708313, |
| "logits/rejected": -1.1296476125717163, |
| "logps/chosen": -926.0774536132812, |
| "logps/rejected": -962.322998046875, |
| "loss": 0.1392, |
| "num_input_tokens_seen": 78658112, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.2768231630325317, |
| "rewards/margins": 3.9475696086883545, |
| "rewards/rejected": -5.224392890930176, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.8427100790040699, |
| "grad_norm": 0.5765385627746582, |
| "learning_rate": 6.440333532118503e-07, |
| "logits/chosen": -1.127885103225708, |
| "logits/rejected": -1.141706943511963, |
| "logps/chosen": -926.9733276367188, |
| "logps/rejected": -950.8040771484375, |
| "loss": 0.1061, |
| "num_input_tokens_seen": 79021824, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -0.9148169159889221, |
| "rewards/margins": 4.253884792327881, |
| "rewards/rejected": -5.168701648712158, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.8465405793631793, |
| "grad_norm": 0.4936072826385498, |
| "learning_rate": 6.136515680580479e-07, |
| "logits/chosen": -1.1449276208877563, |
| "logits/rejected": -1.144270420074463, |
| "logps/chosen": -951.5328979492188, |
| "logps/rejected": -974.8268432617188, |
| "loss": 0.0769, |
| "num_input_tokens_seen": 79377856, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.3788504600524902, |
| "rewards/margins": 4.367871284484863, |
| "rewards/rejected": -5.746721267700195, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.8503710797222888, |
| "grad_norm": 0.7145355343818665, |
| "learning_rate": 5.839569300706127e-07, |
| "logits/chosen": -1.1404030323028564, |
| "logits/rejected": -1.1440632343292236, |
| "logps/chosen": -955.458984375, |
| "logps/rejected": -985.0751342773438, |
| "loss": 0.0993, |
| "num_input_tokens_seen": 79735616, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.1539299488067627, |
| "rewards/margins": 4.464937210083008, |
| "rewards/rejected": -5.618866920471191, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.8542015800813981, |
| "grad_norm": 0.728387713432312, |
| "learning_rate": 5.549540910884649e-07, |
| "logits/chosen": -1.149441123008728, |
| "logits/rejected": -1.142416000366211, |
| "logps/chosen": -958.7587890625, |
| "logps/rejected": -992.5391845703125, |
| "loss": 0.0954, |
| "num_input_tokens_seen": 80104704, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1392351388931274, |
| "rewards/margins": 4.274951934814453, |
| "rewards/rejected": -5.414187431335449, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.8580320804405075, |
| "grad_norm": 0.44472503662109375, |
| "learning_rate": 5.266475945761562e-07, |
| "logits/chosen": -1.1332907676696777, |
| "logits/rejected": -1.1400041580200195, |
| "logps/chosen": -939.7073364257812, |
| "logps/rejected": -959.638916015625, |
| "loss": 0.0731, |
| "num_input_tokens_seen": 80450752, |
| "rewards/accuracies": 0.9921875, |
| "rewards/chosen": -1.034857153892517, |
| "rewards/margins": 4.585856914520264, |
| "rewards/rejected": -5.62071418762207, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.861862580799617, |
| "grad_norm": 0.6553906202316284, |
| "learning_rate": 4.990418749121179e-07, |
| "logits/chosen": -1.157515525817871, |
| "logits/rejected": -1.1576440334320068, |
| "logps/chosen": -981.6168212890625, |
| "logps/rejected": -996.4295043945312, |
| "loss": 0.1237, |
| "num_input_tokens_seen": 80818432, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.253407597541809, |
| "rewards/margins": 3.791903257369995, |
| "rewards/rejected": -5.045310974121094, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.8656930811587263, |
| "grad_norm": 0.5193009972572327, |
| "learning_rate": 4.721412566939804e-07, |
| "logits/chosen": -1.1454250812530518, |
| "logits/rejected": -1.1320409774780273, |
| "logps/chosen": -903.3690185546875, |
| "logps/rejected": -966.2847900390625, |
| "loss": 0.0816, |
| "num_input_tokens_seen": 81175296, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.2726759910583496, |
| "rewards/margins": 4.386380195617676, |
| "rewards/rejected": -5.659055709838867, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.8695235815178358, |
| "grad_norm": 0.558116614818573, |
| "learning_rate": 4.4594995406110785e-07, |
| "logits/chosen": -1.1296035051345825, |
| "logits/rejected": -1.126936912536621, |
| "logps/chosen": -904.9466552734375, |
| "logps/rejected": -930.12255859375, |
| "loss": 0.1063, |
| "num_input_tokens_seen": 81528576, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.2420897483825684, |
| "rewards/margins": 4.35035514831543, |
| "rewards/rejected": -5.59244441986084, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.8733540818769452, |
| "grad_norm": 0.7199344038963318, |
| "learning_rate": 4.2047207003442003e-07, |
| "logits/chosen": -1.1620718240737915, |
| "logits/rejected": -1.1633644104003906, |
| "logps/chosen": -944.5686645507812, |
| "logps/rejected": -986.4454345703125, |
| "loss": 0.1303, |
| "num_input_tokens_seen": 81893888, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.367161750793457, |
| "rewards/margins": 3.982255220413208, |
| "rewards/rejected": -5.349416732788086, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.8771845822360546, |
| "grad_norm": 0.770560085773468, |
| "learning_rate": 3.957115958736374e-07, |
| "logits/chosen": -1.1352251768112183, |
| "logits/rejected": -1.1215541362762451, |
| "logps/chosen": -881.5318603515625, |
| "logps/rejected": -917.5421142578125, |
| "loss": 0.1263, |
| "num_input_tokens_seen": 82220416, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.1326651573181152, |
| "rewards/margins": 4.0631184577941895, |
| "rewards/rejected": -5.195783615112305, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.881015082595164, |
| "grad_norm": 0.6659558415412903, |
| "learning_rate": 3.7167241045202474e-07, |
| "logits/chosen": -1.143093228340149, |
| "logits/rejected": -1.1408135890960693, |
| "logps/chosen": -967.25830078125, |
| "logps/rejected": -995.8428955078125, |
| "loss": 0.1151, |
| "num_input_tokens_seen": 82582784, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.2594966888427734, |
| "rewards/margins": 4.3952860832214355, |
| "rewards/rejected": -5.654782772064209, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.8848455829542734, |
| "grad_norm": 0.5715085864067078, |
| "learning_rate": 3.483582796487395e-07, |
| "logits/chosen": -1.1340765953063965, |
| "logits/rejected": -1.1376596689224243, |
| "logps/chosen": -950.77294921875, |
| "logps/rejected": -955.8455200195312, |
| "loss": 0.0953, |
| "num_input_tokens_seen": 82945856, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.2390849590301514, |
| "rewards/margins": 4.258786201477051, |
| "rewards/rejected": -5.497871398925781, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.8886760833133828, |
| "grad_norm": 0.6102214455604553, |
| "learning_rate": 3.257728557588902e-07, |
| "logits/chosen": -1.151562213897705, |
| "logits/rejected": -1.1459643840789795, |
| "logps/chosen": -951.6468505859375, |
| "logps/rejected": -975.72314453125, |
| "loss": 0.1001, |
| "num_input_tokens_seen": 83325760, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.3845793008804321, |
| "rewards/margins": 4.108867645263672, |
| "rewards/rejected": -5.493447303771973, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.8925065836724922, |
| "grad_norm": 0.5366233587265015, |
| "learning_rate": 3.039196769213787e-07, |
| "logits/chosen": -1.140844464302063, |
| "logits/rejected": -1.1291863918304443, |
| "logps/chosen": -923.5592041015625, |
| "logps/rejected": -968.033203125, |
| "loss": 0.0911, |
| "num_input_tokens_seen": 83684032, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.2117785215377808, |
| "rewards/margins": 4.085620880126953, |
| "rewards/rejected": -5.297399520874023, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.8963370840316016, |
| "grad_norm": 0.6304879784584045, |
| "learning_rate": 2.828021665646341e-07, |
| "logits/chosen": -1.1628239154815674, |
| "logits/rejected": -1.1564610004425049, |
| "logps/chosen": -984.918701171875, |
| "logps/rejected": -1041.814453125, |
| "loss": 0.1165, |
| "num_input_tokens_seen": 84058624, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.4553375244140625, |
| "rewards/margins": 4.140695571899414, |
| "rewards/rejected": -5.596033096313477, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.900167584390711, |
| "grad_norm": 0.8794420957565308, |
| "learning_rate": 2.6242363287030617e-07, |
| "logits/chosen": -1.153059959411621, |
| "logits/rejected": -1.152005910873413, |
| "logps/chosen": -919.1626586914062, |
| "logps/rejected": -951.9708251953125, |
| "loss": 0.1757, |
| "num_input_tokens_seen": 84418816, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -1.485947608947754, |
| "rewards/margins": 3.8721084594726562, |
| "rewards/rejected": -5.35805606842041, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.9039980847498205, |
| "grad_norm": 0.8899986743927002, |
| "learning_rate": 2.4278726825502696e-07, |
| "logits/chosen": -1.172439694404602, |
| "logits/rejected": -1.1706933975219727, |
| "logps/chosen": -941.1187744140625, |
| "logps/rejected": -960.712158203125, |
| "loss": 0.1672, |
| "num_input_tokens_seen": 84791232, |
| "rewards/accuracies": 0.9296875, |
| "rewards/chosen": -1.478002905845642, |
| "rewards/margins": 4.020918846130371, |
| "rewards/rejected": -5.4989213943481445, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.9078285851089298, |
| "grad_norm": 0.7009980082511902, |
| "learning_rate": 2.2389614887029564e-07, |
| "logits/chosen": -1.158771276473999, |
| "logits/rejected": -1.1504409313201904, |
| "logps/chosen": -931.8646240234375, |
| "logps/rejected": -975.0748901367188, |
| "loss": 0.1262, |
| "num_input_tokens_seen": 85148544, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1419326066970825, |
| "rewards/margins": 4.0716447830200195, |
| "rewards/rejected": -5.2135772705078125, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.9116590854680393, |
| "grad_norm": 0.8356382846832275, |
| "learning_rate": 2.0575323412058036e-07, |
| "logits/chosen": -1.1342267990112305, |
| "logits/rejected": -1.1222342252731323, |
| "logps/chosen": -956.0548095703125, |
| "logps/rejected": -1007.9989013671875, |
| "loss": 0.1671, |
| "num_input_tokens_seen": 85514624, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.1702017784118652, |
| "rewards/margins": 3.732792377471924, |
| "rewards/rejected": -4.902994155883789, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.9154895858271487, |
| "grad_norm": 0.5944716930389404, |
| "learning_rate": 1.8836136619971468e-07, |
| "logits/chosen": -1.1205250024795532, |
| "logits/rejected": -1.1232792139053345, |
| "logps/chosen": -959.98681640625, |
| "logps/rejected": -1005.0180053710938, |
| "loss": 0.1018, |
| "num_input_tokens_seen": 85866560, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.3545148372650146, |
| "rewards/margins": 4.15582275390625, |
| "rewards/rejected": -5.510337829589844, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.919320086186258, |
| "grad_norm": 0.6897103190422058, |
| "learning_rate": 1.7172326964564777e-07, |
| "logits/chosen": -1.1565879583358765, |
| "logits/rejected": -1.1416594982147217, |
| "logps/chosen": -908.073974609375, |
| "logps/rejected": -963.4183349609375, |
| "loss": 0.1222, |
| "num_input_tokens_seen": 86227584, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.198516607284546, |
| "rewards/margins": 4.089000225067139, |
| "rewards/rejected": -5.2875165939331055, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.9231505865453675, |
| "grad_norm": 0.7914372682571411, |
| "learning_rate": 1.5584155091362907e-07, |
| "logits/chosen": -1.161057949066162, |
| "logits/rejected": -1.1563315391540527, |
| "logps/chosen": -974.1826171875, |
| "logps/rejected": -986.8939208984375, |
| "loss": 0.1439, |
| "num_input_tokens_seen": 86599360, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.2627431154251099, |
| "rewards/margins": 3.8127615451812744, |
| "rewards/rejected": -5.075504302978516, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.9269810869044769, |
| "grad_norm": 0.622022807598114, |
| "learning_rate": 1.4071869796789427e-07, |
| "logits/chosen": -1.1303247213363647, |
| "logits/rejected": -1.122231125831604, |
| "logps/chosen": -922.9158935546875, |
| "logps/rejected": -966.453125, |
| "loss": 0.1047, |
| "num_input_tokens_seen": 86962112, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.1618643999099731, |
| "rewards/margins": 4.100822448730469, |
| "rewards/rejected": -5.262686729431152, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.9308115872635863, |
| "grad_norm": 0.6456294059753418, |
| "learning_rate": 1.263570798919106e-07, |
| "logits/chosen": -1.153860330581665, |
| "logits/rejected": -1.1510493755340576, |
| "logps/chosen": -967.277587890625, |
| "logps/rejected": -1002.4358520507812, |
| "loss": 0.0951, |
| "num_input_tokens_seen": 87320064, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.1851540803909302, |
| "rewards/margins": 4.55943489074707, |
| "rewards/rejected": -5.744588851928711, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.9346420876226957, |
| "grad_norm": 0.6258938908576965, |
| "learning_rate": 1.1275894651724517e-07, |
| "logits/chosen": -1.1263501644134521, |
| "logits/rejected": -1.1272790431976318, |
| "logps/chosen": -916.5689086914062, |
| "logps/rejected": -943.59814453125, |
| "loss": 0.111, |
| "num_input_tokens_seen": 87670976, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.3743226528167725, |
| "rewards/margins": 4.170286178588867, |
| "rewards/rejected": -5.5446085929870605, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.9384725879818051, |
| "grad_norm": 0.6042916178703308, |
| "learning_rate": 9.992642807111486e-08, |
| "logits/chosen": -1.1254693269729614, |
| "logits/rejected": -1.1179457902908325, |
| "logps/chosen": -866.78955078125, |
| "logps/rejected": -914.8248901367188, |
| "loss": 0.0918, |
| "num_input_tokens_seen": 88002176, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.1663532257080078, |
| "rewards/margins": 4.326272010803223, |
| "rewards/rejected": -5.4926252365112305, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.9423030883409146, |
| "grad_norm": 0.4170897603034973, |
| "learning_rate": 8.78615348426759e-08, |
| "logits/chosen": -1.1225149631500244, |
| "logits/rejected": -1.121935486793518, |
| "logps/chosen": -905.6043701171875, |
| "logps/rejected": -942.7872314453125, |
| "loss": 0.0682, |
| "num_input_tokens_seen": 88350976, |
| "rewards/accuracies": 0.9921875, |
| "rewards/chosen": -0.9678192734718323, |
| "rewards/margins": 4.1514763832092285, |
| "rewards/rejected": -5.119295597076416, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.946133588700024, |
| "grad_norm": 0.5161164999008179, |
| "learning_rate": 7.656615686809976e-08, |
| "logits/chosen": -1.1459851264953613, |
| "logits/rejected": -1.1481428146362305, |
| "logps/chosen": -931.0963134765625, |
| "logps/rejected": -971.57080078125, |
| "loss": 0.0867, |
| "num_input_tokens_seen": 88715456, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.2464854717254639, |
| "rewards/margins": 4.32740592956543, |
| "rewards/rejected": -5.5738911628723145, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.9499640890591333, |
| "grad_norm": 0.5249192118644714, |
| "learning_rate": 6.604206363448662e-08, |
| "logits/chosen": -1.1482356786727905, |
| "logits/rejected": -1.1453156471252441, |
| "logps/chosen": -953.91796875, |
| "logps/rejected": -984.6043701171875, |
| "loss": 0.0874, |
| "num_input_tokens_seen": 89086656, |
| "rewards/accuracies": 0.984375, |
| "rewards/chosen": -1.0581538677215576, |
| "rewards/margins": 4.51418399810791, |
| "rewards/rejected": -5.572338104248047, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.9537945894182428, |
| "grad_norm": 0.5978475213050842, |
| "learning_rate": 5.6290903802665444e-08, |
| "logits/chosen": -1.1580530405044556, |
| "logits/rejected": -1.1452796459197998, |
| "logps/chosen": -985.2161865234375, |
| "logps/rejected": -1021.5433959960938, |
| "loss": 0.0937, |
| "num_input_tokens_seen": 89469120, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.0257568359375, |
| "rewards/margins": 4.238351821899414, |
| "rewards/rejected": -5.264108657836914, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.9576250897773522, |
| "grad_norm": 0.726586103439331, |
| "learning_rate": 4.7314204948923356e-08, |
| "logits/chosen": -1.126878261566162, |
| "logits/rejected": -1.1194556951522827, |
| "logps/chosen": -931.3563842773438, |
| "logps/rejected": -967.1663818359375, |
| "loss": 0.1485, |
| "num_input_tokens_seen": 89832832, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.1478817462921143, |
| "rewards/margins": 3.753807544708252, |
| "rewards/rejected": -4.901689529418945, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.9614555901364615, |
| "grad_norm": 0.705651581287384, |
| "learning_rate": 3.911337332569876e-08, |
| "logits/chosen": -1.1198232173919678, |
| "logits/rejected": -1.1242961883544922, |
| "logps/chosen": -925.5709228515625, |
| "logps/rejected": -947.575927734375, |
| "loss": 0.143, |
| "num_input_tokens_seen": 90184960, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": -1.1102354526519775, |
| "rewards/margins": 4.033443450927734, |
| "rewards/rejected": -5.143678665161133, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.965286090495571, |
| "grad_norm": 0.619276762008667, |
| "learning_rate": 3.168969364128527e-08, |
| "logits/chosen": -1.1234852075576782, |
| "logits/rejected": -1.1284223794937134, |
| "logps/chosen": -901.2196044921875, |
| "logps/rejected": -934.9398193359375, |
| "loss": 0.1005, |
| "num_input_tokens_seen": 90540608, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.3671045303344727, |
| "rewards/margins": 4.210000991821289, |
| "rewards/rejected": -5.577105522155762, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.9691165908546804, |
| "grad_norm": 0.6134979128837585, |
| "learning_rate": 2.5044328858576105e-08, |
| "logits/chosen": -1.1416277885437012, |
| "logits/rejected": -1.1393930912017822, |
| "logps/chosen": -944.0597534179688, |
| "logps/rejected": -983.457275390625, |
| "loss": 0.1082, |
| "num_input_tokens_seen": 90903552, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.03385329246521, |
| "rewards/margins": 4.257451057434082, |
| "rewards/rejected": -5.291304111480713, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.9729470912137897, |
| "grad_norm": 0.6683414578437805, |
| "learning_rate": 1.917832001287645e-08, |
| "logits/chosen": -1.1274126768112183, |
| "logits/rejected": -1.1147723197937012, |
| "logps/chosen": -930.3221435546875, |
| "logps/rejected": -962.2817993164062, |
| "loss": 0.1193, |
| "num_input_tokens_seen": 91277824, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.2914443016052246, |
| "rewards/margins": 4.013041019439697, |
| "rewards/rejected": -5.304485321044922, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.9767775915728992, |
| "grad_norm": 0.6220396757125854, |
| "learning_rate": 1.4092586048820578e-08, |
| "logits/chosen": -1.156324863433838, |
| "logits/rejected": -1.1507492065429688, |
| "logps/chosen": -936.469482421875, |
| "logps/rejected": -965.8038330078125, |
| "loss": 0.1073, |
| "num_input_tokens_seen": 91637568, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1197013854980469, |
| "rewards/margins": 4.163107872009277, |
| "rewards/rejected": -5.282809257507324, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.9806080919320086, |
| "grad_norm": 0.6376728415489197, |
| "learning_rate": 9.787923676414235e-09, |
| "logits/chosen": -1.131098985671997, |
| "logits/rejected": -1.1358366012573242, |
| "logps/chosen": -904.1517944335938, |
| "logps/rejected": -921.351806640625, |
| "loss": 0.1165, |
| "num_input_tokens_seen": 91986624, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.3765130043029785, |
| "rewards/margins": 3.9553894996643066, |
| "rewards/rejected": -5.331902503967285, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.9844385922911181, |
| "grad_norm": 0.6460384726524353, |
| "learning_rate": 6.265007246223365e-09, |
| "logits/chosen": -1.1434887647628784, |
| "logits/rejected": -1.1374716758728027, |
| "logps/chosen": -952.5870361328125, |
| "logps/rejected": -980.957275390625, |
| "loss": 0.1232, |
| "num_input_tokens_seen": 92359232, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.1729092597961426, |
| "rewards/margins": 3.816133499145508, |
| "rewards/rejected": -4.98904275894165, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.9882690926502274, |
| "grad_norm": 0.7076448202133179, |
| "learning_rate": 3.524388643736387e-09, |
| "logits/chosen": -1.1206657886505127, |
| "logits/rejected": -1.1284191608428955, |
| "logps/chosen": -945.7650146484375, |
| "logps/rejected": -951.6602783203125, |
| "loss": 0.1287, |
| "num_input_tokens_seen": 92725888, |
| "rewards/accuracies": 0.9609375, |
| "rewards/chosen": -1.2057055234909058, |
| "rewards/margins": 3.72121000289917, |
| "rewards/rejected": -4.926915645599365, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.9920995930093368, |
| "grad_norm": 0.8474130034446716, |
| "learning_rate": 1.566497202904471e-09, |
| "logits/chosen": -1.1205601692199707, |
| "logits/rejected": -1.1278849840164185, |
| "logps/chosen": -967.0809326171875, |
| "logps/rejected": -996.765380859375, |
| "loss": 0.1515, |
| "num_input_tokens_seen": 93092224, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.9331145882606506, |
| "rewards/margins": 3.8447437286376953, |
| "rewards/rejected": -4.777858734130859, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.9959300933684463, |
| "grad_norm": 0.6645534634590149, |
| "learning_rate": 3.916396388869981e-10, |
| "logits/chosen": -1.1607704162597656, |
| "logits/rejected": -1.154091238975525, |
| "logps/chosen": -940.1096801757812, |
| "logps/rejected": -995.0182495117188, |
| "loss": 0.1015, |
| "num_input_tokens_seen": 93450688, |
| "rewards/accuracies": 0.9765625, |
| "rewards/chosen": -1.526294231414795, |
| "rewards/margins": 4.032198429107666, |
| "rewards/rejected": -5.558492660522461, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.9997605937275557, |
| "grad_norm": 0.7298867106437683, |
| "learning_rate": 0.0, |
| "logits/chosen": -1.1293647289276123, |
| "logits/rejected": -1.123779296875, |
| "logps/chosen": -938.12451171875, |
| "logps/rejected": -974.994140625, |
| "loss": 0.1225, |
| "num_input_tokens_seen": 93801152, |
| "rewards/accuracies": 0.9453125, |
| "rewards/chosen": -1.6063673496246338, |
| "rewards/margins": 4.141149044036865, |
| "rewards/rejected": -5.747516632080078, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.9997605937275557, |
| "num_input_tokens_seen": 93801152, |
| "step": 261, |
| "total_flos": 4.318246812239528e+18, |
| "train_loss": 0.2454147177009747, |
| "train_runtime": 22841.5151, |
| "train_samples_per_second": 1.463, |
| "train_steps_per_second": 0.011 |
| } |
| ], |
| "logging_steps": 1.0, |
| "max_steps": 261, |
| "num_input_tokens_seen": 93801152, |
| "num_train_epochs": 1, |
| "save_steps": 1000, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 4.318246812239528e+18, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
