| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9984301412872841, | |
| "eval_steps": 100, | |
| "global_step": 477, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.0416666666666666e-08, | |
| "logits/chosen": 0.01849743165075779, | |
| "logits/rejected": 0.013860300183296204, | |
| "logps/chosen": -318.92303466796875, | |
| "logps/rejected": -327.4117126464844, | |
| "loss": 0.0872, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": 0.0165844839066267, | |
| "logits/rejected": 0.029045505449175835, | |
| "logps/chosen": -380.119384765625, | |
| "logps/rejected": -372.70452880859375, | |
| "loss": 0.0916, | |
| "rewards/accuracies": 0.4930555522441864, | |
| "rewards/chosen": 0.00031676876824349165, | |
| "rewards/margins": 0.0008045767317526042, | |
| "rewards/rejected": -0.00048780813813209534, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -0.01443287543952465, | |
| "logits/rejected": 0.01765434443950653, | |
| "logps/chosen": -396.4976501464844, | |
| "logps/rejected": -366.0671691894531, | |
| "loss": 0.0929, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.000257034320384264, | |
| "rewards/margins": 0.0013006285298615694, | |
| "rewards/rejected": -0.0010435942094773054, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": 0.037671297788619995, | |
| "logits/rejected": 0.06698160618543625, | |
| "logps/chosen": -374.0677795410156, | |
| "logps/rejected": -360.3742370605469, | |
| "loss": 0.0849, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.0024321433156728745, | |
| "rewards/margins": 0.003862987505272031, | |
| "rewards/rejected": -0.006295130588114262, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -0.016021814197301865, | |
| "logits/rejected": 0.040130265057086945, | |
| "logps/chosen": -384.62115478515625, | |
| "logps/rejected": -369.37591552734375, | |
| "loss": 0.0899, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.005014514084905386, | |
| "rewards/margins": 0.00654798885807395, | |
| "rewards/rejected": -0.01156250387430191, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.999731868769026e-07, | |
| "logits/chosen": 0.021576542407274246, | |
| "logits/rejected": 0.04092331975698471, | |
| "logps/chosen": -395.0044860839844, | |
| "logps/rejected": -385.6026306152344, | |
| "loss": 0.0905, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.011928597465157509, | |
| "rewards/margins": 0.01728428527712822, | |
| "rewards/rejected": -0.02921288087964058, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.990353313429303e-07, | |
| "logits/chosen": 0.09396852552890778, | |
| "logits/rejected": 0.177364319562912, | |
| "logps/chosen": -373.46978759765625, | |
| "logps/rejected": -350.2561950683594, | |
| "loss": 0.0896, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.022122707217931747, | |
| "rewards/margins": 0.04510267823934555, | |
| "rewards/rejected": -0.067225381731987, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.967625656594781e-07, | |
| "logits/chosen": 0.09231746941804886, | |
| "logits/rejected": 0.10504136979579926, | |
| "logps/chosen": -380.4566955566406, | |
| "logps/rejected": -384.76495361328125, | |
| "loss": 0.0895, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.020214151591062546, | |
| "rewards/margins": 0.044125162065029144, | |
| "rewards/rejected": -0.06433931738138199, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.93167072587771e-07, | |
| "logits/chosen": 0.1812177449464798, | |
| "logits/rejected": 0.2344866693019867, | |
| "logps/chosen": -373.54779052734375, | |
| "logps/rejected": -344.9815673828125, | |
| "loss": 0.0887, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.003499386366456747, | |
| "rewards/margins": 0.11121924966573715, | |
| "rewards/rejected": -0.11471863090991974, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.882681251368548e-07, | |
| "logits/chosen": 0.23078179359436035, | |
| "logits/rejected": 0.3160688281059265, | |
| "logps/chosen": -398.22735595703125, | |
| "logps/rejected": -354.7359619140625, | |
| "loss": 0.0854, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.03982505947351456, | |
| "rewards/margins": 0.12135788053274155, | |
| "rewards/rejected": -0.1611829400062561, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.820919832540181e-07, | |
| "logits/chosen": 0.33522385358810425, | |
| "logits/rejected": 0.34693339467048645, | |
| "logps/chosen": -373.6068115234375, | |
| "logps/rejected": -393.63311767578125, | |
| "loss": 0.09, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.05229802802205086, | |
| "rewards/margins": 0.1304590255022049, | |
| "rewards/rejected": -0.18275703489780426, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_logits/chosen": 0.49261584877967834, | |
| "eval_logits/rejected": 0.5302599668502808, | |
| "eval_logps/chosen": -392.5748291015625, | |
| "eval_logps/rejected": -418.8423767089844, | |
| "eval_loss": 0.08443526923656464, | |
| "eval_rewards/accuracies": 0.69921875, | |
| "eval_rewards/chosen": -0.09445539116859436, | |
| "eval_rewards/margins": 0.20123936235904694, | |
| "eval_rewards/rejected": -0.2956947684288025, | |
| "eval_runtime": 75.5045, | |
| "eval_samples_per_second": 26.488, | |
| "eval_steps_per_second": 0.424, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7467175306295647e-07, | |
| "logits/chosen": 0.5233359336853027, | |
| "logits/rejected": 0.5924205780029297, | |
| "logps/chosen": -409.8135681152344, | |
| "logps/rejected": -400.6418151855469, | |
| "loss": 0.0775, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.17791931331157684, | |
| "rewards/margins": 0.2254853993654251, | |
| "rewards/rejected": -0.40340471267700195, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6604720940421207e-07, | |
| "logits/chosen": 0.6610409021377563, | |
| "logits/rejected": 0.8009072542190552, | |
| "logps/chosen": -459.3719787597656, | |
| "logps/rejected": -480.128662109375, | |
| "loss": 0.0697, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.2562519609928131, | |
| "rewards/margins": 0.2973101735115051, | |
| "rewards/rejected": -0.5535621643066406, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5626458262912735e-07, | |
| "logits/chosen": 0.8142817616462708, | |
| "logits/rejected": 1.0136159658432007, | |
| "logps/chosen": -453.57037353515625, | |
| "logps/rejected": -438.6094665527344, | |
| "loss": 0.0557, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.45035696029663086, | |
| "rewards/margins": 0.2075636386871338, | |
| "rewards/rejected": -0.6579206585884094, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.453763107901675e-07, | |
| "logits/chosen": 0.9267638325691223, | |
| "logits/rejected": 0.9543718099594116, | |
| "logps/chosen": -426.4134826660156, | |
| "logps/rejected": -436.49261474609375, | |
| "loss": 0.06, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.33891427516937256, | |
| "rewards/margins": 0.302972674369812, | |
| "rewards/rejected": -0.6418868899345398, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3344075855595097e-07, | |
| "logits/chosen": 0.834929347038269, | |
| "logits/rejected": 1.0096248388290405, | |
| "logps/chosen": -383.9637756347656, | |
| "logps/rejected": -392.84912109375, | |
| "loss": 0.0588, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.2390960454940796, | |
| "rewards/margins": 0.35297515988349915, | |
| "rewards/rejected": -0.5920711755752563, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.2052190435769554e-07, | |
| "logits/chosen": 1.0894076824188232, | |
| "logits/rejected": 1.2157137393951416, | |
| "logps/chosen": -429.09857177734375, | |
| "logps/rejected": -461.9745178222656, | |
| "loss": 0.0509, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.5552287101745605, | |
| "rewards/margins": 0.3786623775959015, | |
| "rewards/rejected": -0.9338911175727844, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0668899744407567e-07, | |
| "logits/chosen": 0.9078506231307983, | |
| "logits/rejected": 1.0372017621994019, | |
| "logps/chosen": -482.3373107910156, | |
| "logps/rejected": -479.88916015625, | |
| "loss": 0.0479, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.505352795124054, | |
| "rewards/margins": 0.26132458448410034, | |
| "rewards/rejected": -0.7666773796081543, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.920161866827889e-07, | |
| "logits/chosen": 0.80833500623703, | |
| "logits/rejected": 0.8488121032714844, | |
| "logps/chosen": -413.3409118652344, | |
| "logps/rejected": -438.3705139160156, | |
| "loss": 0.0476, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.4049296975135803, | |
| "rewards/margins": 0.3900560736656189, | |
| "rewards/rejected": -0.7949857115745544, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.765821230985757e-07, | |
| "logits/chosen": 0.9091412425041199, | |
| "logits/rejected": 1.0051593780517578, | |
| "logps/chosen": -395.74383544921875, | |
| "logps/rejected": -402.8367919921875, | |
| "loss": 0.0478, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.4961649775505066, | |
| "rewards/margins": 0.3637959361076355, | |
| "rewards/rejected": -0.8599609136581421, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.604695382782159e-07, | |
| "logits/chosen": 1.0421111583709717, | |
| "logits/rejected": 1.1686071157455444, | |
| "logps/chosen": -422.24224853515625, | |
| "logps/rejected": -469.1251525878906, | |
| "loss": 0.0405, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.7005800008773804, | |
| "rewards/margins": 0.46449971199035645, | |
| "rewards/rejected": -1.1650797128677368, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "eval_logits/chosen": 1.1859312057495117, | |
| "eval_logits/rejected": 1.2733540534973145, | |
| "eval_logps/chosen": -449.3788757324219, | |
| "eval_logps/rejected": -505.84661865234375, | |
| "eval_loss": 0.045209601521492004, | |
| "eval_rewards/accuracies": 0.75390625, | |
| "eval_rewards/chosen": -0.6624964475631714, | |
| "eval_rewards/margins": 0.5032405257225037, | |
| "eval_rewards/rejected": -1.1657369136810303, | |
| "eval_runtime": 75.0855, | |
| "eval_samples_per_second": 26.636, | |
| "eval_steps_per_second": 0.426, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4376480090239047e-07, | |
| "logits/chosen": 0.9289053082466125, | |
| "logits/rejected": 1.0322377681732178, | |
| "logps/chosen": -454.09521484375, | |
| "logps/rejected": -484.48956298828125, | |
| "loss": 0.0428, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.5330354571342468, | |
| "rewards/margins": 0.47441625595092773, | |
| "rewards/rejected": -1.0074517726898193, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.265574537815398e-07, | |
| "logits/chosen": 0.6325788497924805, | |
| "logits/rejected": 0.8454742431640625, | |
| "logps/chosen": -443.6888732910156, | |
| "logps/rejected": -444.2510681152344, | |
| "loss": 0.051, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.4299241006374359, | |
| "rewards/margins": 0.41193485260009766, | |
| "rewards/rejected": -0.8418590426445007, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.0893973387735683e-07, | |
| "logits/chosen": 0.8997888565063477, | |
| "logits/rejected": 0.9853512048721313, | |
| "logps/chosen": -413.89520263671875, | |
| "logps/rejected": -458.99676513671875, | |
| "loss": 0.0525, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.5308324694633484, | |
| "rewards/margins": 0.4597201943397522, | |
| "rewards/rejected": -0.9905527830123901, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.910060778827554e-07, | |
| "logits/chosen": 1.0547417402267456, | |
| "logits/rejected": 1.1306800842285156, | |
| "logps/chosen": -493.91790771484375, | |
| "logps/rejected": -539.1799926757812, | |
| "loss": 0.0471, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.6113244295120239, | |
| "rewards/margins": 0.5182110667228699, | |
| "rewards/rejected": -1.1295355558395386, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7285261601056697e-07, | |
| "logits/chosen": 1.2281643152236938, | |
| "logits/rejected": 1.359076976776123, | |
| "logps/chosen": -466.77001953125, | |
| "logps/rejected": -483.91259765625, | |
| "loss": 0.0419, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.7258759140968323, | |
| "rewards/margins": 0.42711353302001953, | |
| "rewards/rejected": -1.152989387512207, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.5457665670441937e-07, | |
| "logits/chosen": 1.2255347967147827, | |
| "logits/rejected": 1.462003469467163, | |
| "logps/chosen": -491.76190185546875, | |
| "logps/rejected": -505.47161865234375, | |
| "loss": 0.0451, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.7059242725372314, | |
| "rewards/margins": 0.6359472274780273, | |
| "rewards/rejected": -1.3418715000152588, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3627616503391812e-07, | |
| "logits/chosen": 1.3674428462982178, | |
| "logits/rejected": 1.578064203262329, | |
| "logps/chosen": -486.397216796875, | |
| "logps/rejected": -492.1827087402344, | |
| "loss": 0.0472, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.7773429155349731, | |
| "rewards/margins": 0.38945746421813965, | |
| "rewards/rejected": -1.1668003797531128, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1804923757009882e-07, | |
| "logits/chosen": 1.366081953048706, | |
| "logits/rejected": 1.5207383632659912, | |
| "logps/chosen": -477.0743103027344, | |
| "logps/rejected": -530.8953857421875, | |
| "loss": 0.0445, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.8941423296928406, | |
| "rewards/margins": 0.4790104925632477, | |
| "rewards/rejected": -1.3731528520584106, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9999357655598891e-07, | |
| "logits/chosen": 1.2689809799194336, | |
| "logits/rejected": 1.4011085033416748, | |
| "logps/chosen": -438.982421875, | |
| "logps/rejected": -469.45703125, | |
| "loss": 0.0464, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.62468022108078, | |
| "rewards/margins": 0.513271689414978, | |
| "rewards/rejected": -1.1379519701004028, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8220596619089573e-07, | |
| "logits/chosen": 1.1505718231201172, | |
| "logits/rejected": 1.4240622520446777, | |
| "logps/chosen": -458.03631591796875, | |
| "logps/rejected": -443.11712646484375, | |
| "loss": 0.0479, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.6710134148597717, | |
| "rewards/margins": 0.39567166566848755, | |
| "rewards/rejected": -1.0666849613189697, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_logits/chosen": 1.2982094287872314, | |
| "eval_logits/rejected": 1.409311056137085, | |
| "eval_logps/chosen": -435.2132568359375, | |
| "eval_logps/rejected": -501.30841064453125, | |
| "eval_loss": 0.047696553170681, | |
| "eval_rewards/accuracies": 0.73828125, | |
| "eval_rewards/chosen": -0.5208398699760437, | |
| "eval_rewards/margins": 0.5995149612426758, | |
| "eval_rewards/rejected": -1.1203548908233643, | |
| "eval_runtime": 75.296, | |
| "eval_samples_per_second": 26.562, | |
| "eval_steps_per_second": 0.425, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.647817538357072e-07, | |
| "logits/chosen": 1.2780801057815552, | |
| "logits/rejected": 1.3399560451507568, | |
| "logps/chosen": -475.42413330078125, | |
| "logps/rejected": -517.4520263671875, | |
| "loss": 0.0478, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.7529923319816589, | |
| "rewards/margins": 0.4368392825126648, | |
| "rewards/rejected": -1.1898316144943237, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.478143389201113e-07, | |
| "logits/chosen": 1.198677897453308, | |
| "logits/rejected": 1.4085700511932373, | |
| "logps/chosen": -498.35711669921875, | |
| "logps/rejected": -497.4380798339844, | |
| "loss": 0.0424, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.6401562690734863, | |
| "rewards/margins": 0.48012202978134155, | |
| "rewards/rejected": -1.1202783584594727, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3139467229135998e-07, | |
| "logits/chosen": 1.2183105945587158, | |
| "logits/rejected": 1.2747819423675537, | |
| "logps/chosen": -442.5284118652344, | |
| "logps/rejected": -533.216796875, | |
| "loss": 0.0454, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.5910875201225281, | |
| "rewards/margins": 0.5799761414527893, | |
| "rewards/rejected": -1.1710636615753174, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1561076868822755e-07, | |
| "logits/chosen": 1.203604817390442, | |
| "logits/rejected": 1.1832085847854614, | |
| "logps/chosen": -441.4521484375, | |
| "logps/rejected": -512.8982543945312, | |
| "loss": 0.0428, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7571262121200562, | |
| "rewards/margins": 0.4640630781650543, | |
| "rewards/rejected": -1.221189260482788, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0054723495346482e-07, | |
| "logits/chosen": 1.3052194118499756, | |
| "logits/rejected": 1.382683515548706, | |
| "logps/chosen": -465.3661193847656, | |
| "logps/rejected": -528.7847290039062, | |
| "loss": 0.0412, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.7757940292358398, | |
| "rewards/margins": 0.4744884967803955, | |
| "rewards/rejected": -1.2502825260162354, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.628481651367875e-08, | |
| "logits/chosen": 1.1976938247680664, | |
| "logits/rejected": 1.432969331741333, | |
| "logps/chosen": -491.15771484375, | |
| "logps/rejected": -515.0520629882812, | |
| "loss": 0.0446, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.6491819620132446, | |
| "rewards/margins": 0.6244359612464905, | |
| "rewards/rejected": -1.2736178636550903, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.289996455765748e-08, | |
| "logits/chosen": 1.192779541015625, | |
| "logits/rejected": 1.324210524559021, | |
| "logps/chosen": -504.5486755371094, | |
| "logps/rejected": -508.7030334472656, | |
| "loss": 0.0435, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7155844569206238, | |
| "rewards/margins": 0.5151349306106567, | |
| "rewards/rejected": -1.2307194471359253, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.046442623320145e-08, | |
| "logits/chosen": 1.223356008529663, | |
| "logits/rejected": 1.4434764385223389, | |
| "logps/chosen": -474.7169494628906, | |
| "logps/rejected": -518.0782470703125, | |
| "loss": 0.0476, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.6729675531387329, | |
| "rewards/margins": 0.6387326717376709, | |
| "rewards/rejected": -1.3117002248764038, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.904486005914027e-08, | |
| "logits/chosen": 1.3060978651046753, | |
| "logits/rejected": 1.4896109104156494, | |
| "logps/chosen": -470.46661376953125, | |
| "logps/rejected": -502.4981384277344, | |
| "loss": 0.0482, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.6783354878425598, | |
| "rewards/margins": 0.5045996904373169, | |
| "rewards/rejected": -1.1829349994659424, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.8702478614051345e-08, | |
| "logits/chosen": 1.3413165807724, | |
| "logits/rejected": 1.4800562858581543, | |
| "logps/chosen": -450.84844970703125, | |
| "logps/rejected": -509.7266540527344, | |
| "loss": 0.0457, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.7674819231033325, | |
| "rewards/margins": 0.4173991084098816, | |
| "rewards/rejected": -1.1848809719085693, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_logits/chosen": 1.491492748260498, | |
| "eval_logits/rejected": 1.6154029369354248, | |
| "eval_logps/chosen": -448.419677734375, | |
| "eval_logps/rejected": -518.3443603515625, | |
| "eval_loss": 0.044891636818647385, | |
| "eval_rewards/accuracies": 0.73828125, | |
| "eval_rewards/chosen": -0.6529037952423096, | |
| "eval_rewards/margins": 0.6378109455108643, | |
| "eval_rewards/rejected": -1.2907147407531738, | |
| "eval_runtime": 74.6873, | |
| "eval_samples_per_second": 26.778, | |
| "eval_steps_per_second": 0.428, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.9492720416985e-08, | |
| "logits/chosen": 1.3658090829849243, | |
| "logits/rejected": 1.523946762084961, | |
| "logps/chosen": -461.0426330566406, | |
| "logps/rejected": -491.6429138183594, | |
| "loss": 0.045, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.6857269406318665, | |
| "rewards/margins": 0.5137700438499451, | |
| "rewards/rejected": -1.1994969844818115, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.1464952759020856e-08, | |
| "logits/chosen": 1.3796783685684204, | |
| "logits/rejected": 1.5178402662277222, | |
| "logps/chosen": -454.60455322265625, | |
| "logps/rejected": -483.65704345703125, | |
| "loss": 0.0418, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.6705530285835266, | |
| "rewards/margins": 0.604373574256897, | |
| "rewards/rejected": -1.2749265432357788, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.4662207078575684e-08, | |
| "logits/chosen": 1.334680199623108, | |
| "logits/rejected": 1.4741976261138916, | |
| "logps/chosen": -504.280029296875, | |
| "logps/rejected": -529.8871459960938, | |
| "loss": 0.0453, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7305961847305298, | |
| "rewards/margins": 0.5881385207176208, | |
| "rewards/rejected": -1.3187347650527954, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.12094829893642e-09, | |
| "logits/chosen": 1.3827157020568848, | |
| "logits/rejected": 1.5478546619415283, | |
| "logps/chosen": -453.01171875, | |
| "logps/rejected": -480.3030700683594, | |
| "loss": 0.0414, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7755357027053833, | |
| "rewards/margins": 0.5378071069717407, | |
| "rewards/rejected": -1.313342809677124, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.8708793644441086e-09, | |
| "logits/chosen": 1.2280631065368652, | |
| "logits/rejected": 1.454526662826538, | |
| "logps/chosen": -487.4305114746094, | |
| "logps/rejected": -500.71087646484375, | |
| "loss": 0.0425, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.6377500295639038, | |
| "rewards/margins": 0.5590785145759583, | |
| "rewards/rejected": -1.1968284845352173, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.9347820230782295e-09, | |
| "logits/chosen": 1.336721658706665, | |
| "logits/rejected": 1.4986612796783447, | |
| "logps/chosen": -455.5997619628906, | |
| "logps/rejected": -474.46038818359375, | |
| "loss": 0.0425, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.7689257264137268, | |
| "rewards/margins": 0.4747004508972168, | |
| "rewards/rejected": -1.243626356124878, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.2839470889836627e-10, | |
| "logits/chosen": 1.2109500169754028, | |
| "logits/rejected": 1.3351854085922241, | |
| "logps/chosen": -490.6439514160156, | |
| "logps/rejected": -541.4273681640625, | |
| "loss": 0.0421, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.7248164415359497, | |
| "rewards/margins": 0.548802375793457, | |
| "rewards/rejected": -1.2736186981201172, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 477, | |
| "total_flos": 0.0, | |
| "train_loss": 0.0564979040210352, | |
| "train_runtime": 4410.0999, | |
| "train_samples_per_second": 13.862, | |
| "train_steps_per_second": 0.108 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 477, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |