| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9994666666666666, | |
| "eval_steps": 500, | |
| "global_step": 937, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.319148936170213e-08, | |
| "logits/chosen": 0.22503182291984558, | |
| "logits/rejected": 0.20745767652988434, | |
| "logps/chosen": -503.960205078125, | |
| "logps/rejected": -393.79901123046875, | |
| "loss": 0.3545, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.319148936170213e-07, | |
| "logits/chosen": 0.13639158010482788, | |
| "logits/rejected": 0.23417764902114868, | |
| "logps/chosen": -312.29608154296875, | |
| "logps/rejected": -286.0322265625, | |
| "loss": 0.3609, | |
| "rewards/accuracies": 0.4305555522441864, | |
| "rewards/chosen": 9.671921725384891e-05, | |
| "rewards/margins": 8.819119102554396e-05, | |
| "rewards/rejected": 8.528020771336742e-06, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.1057649701833725, | |
| "logits/rejected": 0.17593668401241302, | |
| "logps/chosen": -372.1605529785156, | |
| "logps/rejected": -358.22869873046875, | |
| "loss": 0.3704, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": 7.727816409897059e-05, | |
| "rewards/margins": 0.00010253030632156879, | |
| "rewards/rejected": -2.5252131308661774e-05, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.11514975130558014, | |
| "logits/rejected": 0.20654134452342987, | |
| "logps/chosen": -371.88287353515625, | |
| "logps/rejected": -313.6852722167969, | |
| "loss": 0.3724, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 5.6704669987084344e-05, | |
| "rewards/margins": 4.732383786176797e-06, | |
| "rewards/rejected": 5.197228165343404e-05, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.08795060217380524, | |
| "logits/rejected": 0.2017325907945633, | |
| "logps/chosen": -372.4765930175781, | |
| "logps/rejected": -345.6024475097656, | |
| "loss": 0.3779, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": 0.00011807264672825113, | |
| "rewards/margins": -1.2403141226968728e-05, | |
| "rewards/rejected": 0.00013047578977420926, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.12501904368400574, | |
| "logits/rejected": 0.1548817902803421, | |
| "logps/chosen": -319.83941650390625, | |
| "logps/rejected": -304.3211975097656, | |
| "loss": 0.3743, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.000266839109826833, | |
| "rewards/margins": 0.0005234330310486257, | |
| "rewards/rejected": -0.00025659389211796224, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.056398905813694, | |
| "logits/rejected": 0.16611042618751526, | |
| "logps/chosen": -342.92767333984375, | |
| "logps/rejected": -301.9825744628906, | |
| "loss": 0.3686, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.0005045770085416734, | |
| "rewards/margins": 0.0005726306699216366, | |
| "rewards/rejected": -6.805359589634463e-05, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.1134648472070694, | |
| "logits/rejected": 0.15889659523963928, | |
| "logps/chosen": -367.37451171875, | |
| "logps/rejected": -299.69818115234375, | |
| "loss": 0.3747, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": 0.0005155171384103596, | |
| "rewards/margins": 0.0006443473394028842, | |
| "rewards/rejected": -0.0001288302446482703, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.14280062913894653, | |
| "logits/rejected": 0.17763297259807587, | |
| "logps/chosen": -384.6070251464844, | |
| "logps/rejected": -323.2566223144531, | |
| "loss": 0.3719, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.0013791865203529596, | |
| "rewards/margins": 0.001650324440561235, | |
| "rewards/rejected": -0.00027113803662359715, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.11830653250217438, | |
| "logits/rejected": 0.18050488829612732, | |
| "logps/chosen": -299.129638671875, | |
| "logps/rejected": -308.74737548828125, | |
| "loss": 0.3591, | |
| "rewards/accuracies": 0.6187499761581421, | |
| "rewards/chosen": 0.0019053416326642036, | |
| "rewards/margins": 0.0032806836534291506, | |
| "rewards/rejected": -0.001375342020764947, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.099853515625, | |
| "logits/rejected": 0.20920684933662415, | |
| "logps/chosen": -351.15936279296875, | |
| "logps/rejected": -335.2391662597656, | |
| "loss": 0.3606, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": 0.0006141302874311805, | |
| "rewards/margins": 0.005217243917286396, | |
| "rewards/rejected": -0.004603113513439894, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.07743649184703827, | |
| "logits/rejected": 0.1303156465291977, | |
| "logps/chosen": -312.6484375, | |
| "logps/rejected": -312.4829406738281, | |
| "loss": 0.3599, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.004017535597085953, | |
| "rewards/margins": 0.00877195131033659, | |
| "rewards/rejected": -0.012789485976099968, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.05438951402902603, | |
| "logits/rejected": 0.15021613240242004, | |
| "logps/chosen": -383.91510009765625, | |
| "logps/rejected": -360.8976745605469, | |
| "loss": 0.3608, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.020418299362063408, | |
| "rewards/margins": 0.013604333624243736, | |
| "rewards/rejected": -0.034022632986307144, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.06449095904827118, | |
| "logits/rejected": 0.11669395118951797, | |
| "logps/chosen": -372.52386474609375, | |
| "logps/rejected": -411.42388916015625, | |
| "loss": 0.343, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.03320731967687607, | |
| "rewards/margins": 0.03409535437822342, | |
| "rewards/rejected": -0.06730266660451889, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.04302512854337692, | |
| "logits/rejected": 0.050311993807554245, | |
| "logps/chosen": -413.1893005371094, | |
| "logps/rejected": -404.35516357421875, | |
| "loss": 0.351, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.06032940745353699, | |
| "rewards/margins": 0.03157948702573776, | |
| "rewards/rejected": -0.09190888702869415, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.03025934100151062, | |
| "logits/rejected": 0.11820876598358154, | |
| "logps/chosen": -419.5621643066406, | |
| "logps/rejected": -454.58148193359375, | |
| "loss": 0.3325, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.07952941954135895, | |
| "rewards/margins": 0.04821295291185379, | |
| "rewards/rejected": -0.12774237990379333, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.09332932531833649, | |
| "logits/rejected": 0.1296716332435608, | |
| "logps/chosen": -410.50634765625, | |
| "logps/rejected": -435.3250427246094, | |
| "loss": 0.3517, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.09774389117956161, | |
| "rewards/margins": 0.045623380690813065, | |
| "rewards/rejected": -0.14336727559566498, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.035608045756816864, | |
| "logits/rejected": 0.18676114082336426, | |
| "logps/chosen": -455.2527770996094, | |
| "logps/rejected": -499.14886474609375, | |
| "loss": 0.3275, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.12179211527109146, | |
| "rewards/margins": 0.055649809539318085, | |
| "rewards/rejected": -0.17744192481040955, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.04756158962845802, | |
| "logits/rejected": 0.11594130843877792, | |
| "logps/chosen": -468.597900390625, | |
| "logps/rejected": -482.14190673828125, | |
| "loss": 0.3563, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.1254795491695404, | |
| "rewards/margins": 0.041418299078941345, | |
| "rewards/rejected": -0.16689783334732056, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.11988376080989838, | |
| "logits/rejected": 0.22085747122764587, | |
| "logps/chosen": -452.22052001953125, | |
| "logps/rejected": -387.4115905761719, | |
| "loss": 0.3779, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.07290507853031158, | |
| "rewards/margins": 0.016655491665005684, | |
| "rewards/rejected": -0.08956055343151093, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": 0.11956377327442169, | |
| "logits/rejected": 0.19818153977394104, | |
| "logps/chosen": -401.03387451171875, | |
| "logps/rejected": -415.95123291015625, | |
| "loss": 0.3403, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.06231163814663887, | |
| "rewards/margins": 0.0385347381234169, | |
| "rewards/rejected": -0.10084638744592667, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.12819179892539978, | |
| "logits/rejected": 0.11948923766613007, | |
| "logps/chosen": -411.1438903808594, | |
| "logps/rejected": -455.16455078125, | |
| "loss": 0.3232, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.08700522780418396, | |
| "rewards/margins": 0.0522175133228302, | |
| "rewards/rejected": -0.13922274112701416, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.15324261784553528, | |
| "logits/rejected": 0.20623254776000977, | |
| "logps/chosen": -480.10888671875, | |
| "logps/rejected": -499.2796936035156, | |
| "loss": 0.3304, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.10702191293239594, | |
| "rewards/margins": 0.05671936273574829, | |
| "rewards/rejected": -0.16374126076698303, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": 0.18569575250148773, | |
| "logits/rejected": 0.17661599814891815, | |
| "logps/chosen": -443.75848388671875, | |
| "logps/rejected": -451.41143798828125, | |
| "loss": 0.342, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.1080365926027298, | |
| "rewards/margins": 0.036346640437841415, | |
| "rewards/rejected": -0.14438322186470032, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": 0.08591008186340332, | |
| "logits/rejected": 0.18295328319072723, | |
| "logps/chosen": -431.0286560058594, | |
| "logps/rejected": -486.2010803222656, | |
| "loss": 0.3212, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.12310918420553207, | |
| "rewards/margins": 0.055181778967380524, | |
| "rewards/rejected": -0.1782909631729126, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": 0.11142314970493317, | |
| "logits/rejected": 0.17328810691833496, | |
| "logps/chosen": -461.910400390625, | |
| "logps/rejected": -543.1893920898438, | |
| "loss": 0.3188, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.1216365322470665, | |
| "rewards/margins": 0.08850517868995667, | |
| "rewards/rejected": -0.21014170348644257, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": 0.13197702169418335, | |
| "logits/rejected": 0.14432582259178162, | |
| "logps/chosen": -482.305908203125, | |
| "logps/rejected": -524.9083862304688, | |
| "loss": 0.3151, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.1686619073152542, | |
| "rewards/margins": 0.057187020778656006, | |
| "rewards/rejected": -0.2258489429950714, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": 0.1734444499015808, | |
| "logits/rejected": 0.21249273419380188, | |
| "logps/chosen": -479.7071838378906, | |
| "logps/rejected": -537.2325439453125, | |
| "loss": 0.327, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.16160346567630768, | |
| "rewards/margins": 0.06107072904706001, | |
| "rewards/rejected": -0.2226742058992386, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": 0.15867115557193756, | |
| "logits/rejected": 0.19647037982940674, | |
| "logps/chosen": -458.0674743652344, | |
| "logps/rejected": -501.34722900390625, | |
| "loss": 0.3286, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.1390560269355774, | |
| "rewards/margins": 0.062171000987291336, | |
| "rewards/rejected": -0.20122702419757843, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": 0.13150140643119812, | |
| "logits/rejected": 0.22199960052967072, | |
| "logps/chosen": -450.36669921875, | |
| "logps/rejected": -506.1780700683594, | |
| "loss": 0.3315, | |
| "rewards/accuracies": 0.6187499761581421, | |
| "rewards/chosen": -0.12922145426273346, | |
| "rewards/margins": 0.06099293380975723, | |
| "rewards/rejected": -0.1902143657207489, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": 0.19327369332313538, | |
| "logits/rejected": 0.24881592392921448, | |
| "logps/chosen": -497.6783142089844, | |
| "logps/rejected": -505.5125427246094, | |
| "loss": 0.3363, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.12306072562932968, | |
| "rewards/margins": 0.05413200706243515, | |
| "rewards/rejected": -0.17719274759292603, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": 0.18834055960178375, | |
| "logits/rejected": 0.23213116824626923, | |
| "logps/chosen": -515.0231323242188, | |
| "logps/rejected": -483.41217041015625, | |
| "loss": 0.3318, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.12383429706096649, | |
| "rewards/margins": 0.042060453444719315, | |
| "rewards/rejected": -0.1658947467803955, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": 0.17896682024002075, | |
| "logits/rejected": 0.2394729107618332, | |
| "logps/chosen": -463.9024353027344, | |
| "logps/rejected": -537.2124633789062, | |
| "loss": 0.3272, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.13291411101818085, | |
| "rewards/margins": 0.0873040109872818, | |
| "rewards/rejected": -0.22021810710430145, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": 0.17872880399227142, | |
| "logits/rejected": 0.24048776924610138, | |
| "logps/chosen": -512.0220947265625, | |
| "logps/rejected": -550.4924926757812, | |
| "loss": 0.3286, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.15976758301258087, | |
| "rewards/margins": 0.06323219835758209, | |
| "rewards/rejected": -0.22299976646900177, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": 0.17755231261253357, | |
| "logits/rejected": 0.21120838820934296, | |
| "logps/chosen": -548.46728515625, | |
| "logps/rejected": -568.703369140625, | |
| "loss": 0.3282, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.17922678589820862, | |
| "rewards/margins": 0.060503531247377396, | |
| "rewards/rejected": -0.23973031342029572, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": 0.16182249784469604, | |
| "logits/rejected": 0.26147884130477905, | |
| "logps/chosen": -495.8375549316406, | |
| "logps/rejected": -622.0607299804688, | |
| "loss": 0.3019, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.16881313920021057, | |
| "rewards/margins": 0.08940894901752472, | |
| "rewards/rejected": -0.2582220435142517, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": 0.18107028305530548, | |
| "logits/rejected": 0.2131391018629074, | |
| "logps/chosen": -508.89654541015625, | |
| "logps/rejected": -552.6736450195312, | |
| "loss": 0.3104, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.1358935385942459, | |
| "rewards/margins": 0.08370544016361237, | |
| "rewards/rejected": -0.21959897875785828, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": 0.1739516258239746, | |
| "logits/rejected": 0.2705302834510803, | |
| "logps/chosen": -560.09912109375, | |
| "logps/rejected": -567.2365112304688, | |
| "loss": 0.3328, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.13778193295001984, | |
| "rewards/margins": 0.07480571419000626, | |
| "rewards/rejected": -0.2125876396894455, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": 0.22400352358818054, | |
| "logits/rejected": 0.29006537795066833, | |
| "logps/chosen": -505.47930908203125, | |
| "logps/rejected": -527.4620361328125, | |
| "loss": 0.3112, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.1362600326538086, | |
| "rewards/margins": 0.07366020977497101, | |
| "rewards/rejected": -0.2099202424287796, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": 0.18031330406665802, | |
| "logits/rejected": 0.2229474037885666, | |
| "logps/chosen": -452.552734375, | |
| "logps/rejected": -508.63262939453125, | |
| "loss": 0.3029, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.14223061501979828, | |
| "rewards/margins": 0.06927666813135147, | |
| "rewards/rejected": -0.21150727570056915, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": 0.15998587012290955, | |
| "logits/rejected": 0.21808397769927979, | |
| "logps/chosen": -500.90606689453125, | |
| "logps/rejected": -549.7945556640625, | |
| "loss": 0.3176, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.1795472800731659, | |
| "rewards/margins": 0.06672011315822601, | |
| "rewards/rejected": -0.24626736342906952, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": 0.2227793037891388, | |
| "logits/rejected": 0.2791008949279785, | |
| "logps/chosen": -547.274658203125, | |
| "logps/rejected": -587.1893310546875, | |
| "loss": 0.3266, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.17155751585960388, | |
| "rewards/margins": 0.09018336981534958, | |
| "rewards/rejected": -0.26174092292785645, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": 0.1325720101594925, | |
| "logits/rejected": 0.252084881067276, | |
| "logps/chosen": -493.5597229003906, | |
| "logps/rejected": -546.8651733398438, | |
| "loss": 0.3289, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.16804993152618408, | |
| "rewards/margins": 0.0734732523560524, | |
| "rewards/rejected": -0.2415231466293335, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": 0.18316540122032166, | |
| "logits/rejected": 0.2671397030353546, | |
| "logps/chosen": -492.28607177734375, | |
| "logps/rejected": -566.4286499023438, | |
| "loss": 0.3285, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.13893112540245056, | |
| "rewards/margins": 0.09025856852531433, | |
| "rewards/rejected": -0.2291897088289261, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": 0.21058030426502228, | |
| "logits/rejected": 0.24718663096427917, | |
| "logps/chosen": -539.2806396484375, | |
| "logps/rejected": -528.8162841796875, | |
| "loss": 0.3355, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.14369945228099823, | |
| "rewards/margins": 0.05218692496418953, | |
| "rewards/rejected": -0.19588637351989746, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": 0.21172599494457245, | |
| "logits/rejected": 0.2746911644935608, | |
| "logps/chosen": -477.0757751464844, | |
| "logps/rejected": -474.86749267578125, | |
| "loss": 0.3319, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.14388036727905273, | |
| "rewards/margins": 0.03479112684726715, | |
| "rewards/rejected": -0.17867149412631989, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": 0.20078706741333008, | |
| "logits/rejected": 0.26265597343444824, | |
| "logps/chosen": -446.60693359375, | |
| "logps/rejected": -523.1649169921875, | |
| "loss": 0.3179, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.1348758488893509, | |
| "rewards/margins": 0.07731764018535614, | |
| "rewards/rejected": -0.21219351887702942, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": 0.22624699771404266, | |
| "logits/rejected": 0.278435081243515, | |
| "logps/chosen": -513.7930297851562, | |
| "logps/rejected": -577.3609619140625, | |
| "loss": 0.2867, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.15286709368228912, | |
| "rewards/margins": 0.09449265152215958, | |
| "rewards/rejected": -0.2473597526550293, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": 0.16784998774528503, | |
| "logits/rejected": 0.18377156555652618, | |
| "logps/chosen": -516.9537353515625, | |
| "logps/rejected": -570.5045776367188, | |
| "loss": 0.3264, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.18329444527626038, | |
| "rewards/margins": 0.06330955028533936, | |
| "rewards/rejected": -0.24660399556159973, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": 0.2142370045185089, | |
| "logits/rejected": 0.2654828131198883, | |
| "logps/chosen": -513.2080078125, | |
| "logps/rejected": -633.4437866210938, | |
| "loss": 0.2835, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.1864250749349594, | |
| "rewards/margins": 0.12554675340652466, | |
| "rewards/rejected": -0.3119718134403229, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": 0.2339199036359787, | |
| "logits/rejected": 0.3211553394794464, | |
| "logps/chosen": -538.3214111328125, | |
| "logps/rejected": -600.5282592773438, | |
| "loss": 0.2922, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.1798117458820343, | |
| "rewards/margins": 0.09368885308504105, | |
| "rewards/rejected": -0.27350059151649475, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": 0.20931610465049744, | |
| "logits/rejected": 0.29080599546432495, | |
| "logps/chosen": -553.5645751953125, | |
| "logps/rejected": -621.51513671875, | |
| "loss": 0.315, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.2022249698638916, | |
| "rewards/margins": 0.09137126803398132, | |
| "rewards/rejected": -0.29359620809555054, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": 0.2311556041240692, | |
| "logits/rejected": 0.2898443639278412, | |
| "logps/chosen": -564.0858764648438, | |
| "logps/rejected": -701.887939453125, | |
| "loss": 0.2773, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.22962744534015656, | |
| "rewards/margins": 0.13914015889167786, | |
| "rewards/rejected": -0.3687676191329956, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": 0.21098744869232178, | |
| "logits/rejected": 0.3181845545768738, | |
| "logps/chosen": -588.2892456054688, | |
| "logps/rejected": -676.7548828125, | |
| "loss": 0.2998, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.21903486549854279, | |
| "rewards/margins": 0.09523408114910126, | |
| "rewards/rejected": -0.31426894664764404, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": 0.2154814451932907, | |
| "logits/rejected": 0.3008938431739807, | |
| "logps/chosen": -563.6261596679688, | |
| "logps/rejected": -657.9494018554688, | |
| "loss": 0.2899, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.19673727452754974, | |
| "rewards/margins": 0.12186020612716675, | |
| "rewards/rejected": -0.3185974657535553, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": 0.22583889961242676, | |
| "logits/rejected": 0.30130234360694885, | |
| "logps/chosen": -559.0008544921875, | |
| "logps/rejected": -666.7113037109375, | |
| "loss": 0.2739, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.18335992097854614, | |
| "rewards/margins": 0.129757359623909, | |
| "rewards/rejected": -0.31311726570129395, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": 0.2066396176815033, | |
| "logits/rejected": 0.23956134915351868, | |
| "logps/chosen": -514.6549072265625, | |
| "logps/rejected": -633.8093872070312, | |
| "loss": 0.3114, | |
| "rewards/accuracies": 0.6187499761581421, | |
| "rewards/chosen": -0.21198049187660217, | |
| "rewards/margins": 0.11119239032268524, | |
| "rewards/rejected": -0.3231728672981262, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": 0.25394177436828613, | |
| "logits/rejected": 0.3233291804790497, | |
| "logps/chosen": -516.4053955078125, | |
| "logps/rejected": -631.0079345703125, | |
| "loss": 0.2863, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.18605783581733704, | |
| "rewards/margins": 0.12697571516036987, | |
| "rewards/rejected": -0.3130335807800293, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": 0.23206424713134766, | |
| "logits/rejected": 0.24580395221710205, | |
| "logps/chosen": -533.16650390625, | |
| "logps/rejected": -624.2952880859375, | |
| "loss": 0.2859, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.19318199157714844, | |
| "rewards/margins": 0.11356300115585327, | |
| "rewards/rejected": -0.3067449927330017, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": 0.2015029937028885, | |
| "logits/rejected": 0.27999237179756165, | |
| "logps/chosen": -526.4388427734375, | |
| "logps/rejected": -610.4771728515625, | |
| "loss": 0.2962, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.1759098321199417, | |
| "rewards/margins": 0.12597855925559998, | |
| "rewards/rejected": -0.3018884062767029, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": 0.22685325145721436, | |
| "logits/rejected": 0.2123553305864334, | |
| "logps/chosen": -530.7593383789062, | |
| "logps/rejected": -596.7511596679688, | |
| "loss": 0.2988, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.20077498257160187, | |
| "rewards/margins": 0.10626114904880524, | |
| "rewards/rejected": -0.3070361018180847, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": 0.2265332043170929, | |
| "logits/rejected": 0.25227993726730347, | |
| "logps/chosen": -548.7151489257812, | |
| "logps/rejected": -657.1322631835938, | |
| "loss": 0.2881, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2090146541595459, | |
| "rewards/margins": 0.12534213066101074, | |
| "rewards/rejected": -0.33435681462287903, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": 0.18398644030094147, | |
| "logits/rejected": 0.2708356976509094, | |
| "logps/chosen": -577.2225952148438, | |
| "logps/rejected": -658.6173095703125, | |
| "loss": 0.2699, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.20767459273338318, | |
| "rewards/margins": 0.11812429130077362, | |
| "rewards/rejected": -0.325798898935318, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": 0.2091066539287567, | |
| "logits/rejected": 0.273884117603302, | |
| "logps/chosen": -563.6356201171875, | |
| "logps/rejected": -645.4375, | |
| "loss": 0.3124, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.20846951007843018, | |
| "rewards/margins": 0.09923194348812103, | |
| "rewards/rejected": -0.3077014088630676, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": 0.2188616544008255, | |
| "logits/rejected": 0.2746090292930603, | |
| "logps/chosen": -567.142578125, | |
| "logps/rejected": -631.2537841796875, | |
| "loss": 0.2965, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.19390718638896942, | |
| "rewards/margins": 0.11340739578008652, | |
| "rewards/rejected": -0.30731460452079773, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": 0.242666095495224, | |
| "logits/rejected": 0.2608735263347626, | |
| "logps/chosen": -517.3770751953125, | |
| "logps/rejected": -616.4417724609375, | |
| "loss": 0.2942, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.1942964643239975, | |
| "rewards/margins": 0.11079655587673187, | |
| "rewards/rejected": -0.30509302020072937, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": 0.16175343096256256, | |
| "logits/rejected": 0.2701238989830017, | |
| "logps/chosen": -573.2815551757812, | |
| "logps/rejected": -654.2884521484375, | |
| "loss": 0.3025, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.22864040732383728, | |
| "rewards/margins": 0.119266077876091, | |
| "rewards/rejected": -0.34790652990341187, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": 0.23474529385566711, | |
| "logits/rejected": 0.29773998260498047, | |
| "logps/chosen": -530.2286376953125, | |
| "logps/rejected": -627.4801025390625, | |
| "loss": 0.3166, | |
| "rewards/accuracies": 0.6187499761581421, | |
| "rewards/chosen": -0.20864705741405487, | |
| "rewards/margins": 0.10489342361688614, | |
| "rewards/rejected": -0.3135404884815216, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": 0.23635394871234894, | |
| "logits/rejected": 0.27978676557540894, | |
| "logps/chosen": -528.3208618164062, | |
| "logps/rejected": -589.3389892578125, | |
| "loss": 0.309, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.19641873240470886, | |
| "rewards/margins": 0.07995133101940155, | |
| "rewards/rejected": -0.276370108127594, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": 0.22646358609199524, | |
| "logits/rejected": 0.31415465474128723, | |
| "logps/chosen": -522.7443237304688, | |
| "logps/rejected": -638.9615478515625, | |
| "loss": 0.2545, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.17692828178405762, | |
| "rewards/margins": 0.13757161796092987, | |
| "rewards/rejected": -0.3144999146461487, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": 0.25629907846450806, | |
| "logits/rejected": 0.32496970891952515, | |
| "logps/chosen": -573.7655029296875, | |
| "logps/rejected": -639.9420166015625, | |
| "loss": 0.3064, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": -0.22299213707447052, | |
| "rewards/margins": 0.0986771434545517, | |
| "rewards/rejected": -0.3216692805290222, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": 0.23862013220787048, | |
| "logits/rejected": 0.29511600732803345, | |
| "logps/chosen": -586.1082153320312, | |
| "logps/rejected": -664.5083618164062, | |
| "loss": 0.318, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.2209562063217163, | |
| "rewards/margins": 0.10732462257146835, | |
| "rewards/rejected": -0.32828083634376526, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": 0.20532169938087463, | |
| "logits/rejected": 0.27353784441947937, | |
| "logps/chosen": -541.2822265625, | |
| "logps/rejected": -623.6652221679688, | |
| "loss": 0.3026, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.2209356278181076, | |
| "rewards/margins": 0.08726519346237183, | |
| "rewards/rejected": -0.30820080637931824, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": 0.2527928650379181, | |
| "logits/rejected": 0.28749826550483704, | |
| "logps/chosen": -536.0897216796875, | |
| "logps/rejected": -644.0540161132812, | |
| "loss": 0.3051, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.21758422255516052, | |
| "rewards/margins": 0.10065360367298126, | |
| "rewards/rejected": -0.3182377815246582, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": 0.2695181965827942, | |
| "logits/rejected": 0.2740219235420227, | |
| "logps/chosen": -605.3865966796875, | |
| "logps/rejected": -638.5525512695312, | |
| "loss": 0.3171, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.21937242150306702, | |
| "rewards/margins": 0.09414593875408173, | |
| "rewards/rejected": -0.31351834535598755, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": 0.19430357217788696, | |
| "logits/rejected": 0.26386886835098267, | |
| "logps/chosen": -525.9102783203125, | |
| "logps/rejected": -589.8792724609375, | |
| "loss": 0.3099, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.19337861239910126, | |
| "rewards/margins": 0.10214637219905853, | |
| "rewards/rejected": -0.2955249845981598, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": 0.22553198039531708, | |
| "logits/rejected": 0.30115246772766113, | |
| "logps/chosen": -566.0553588867188, | |
| "logps/rejected": -673.5691528320312, | |
| "loss": 0.3187, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.23551130294799805, | |
| "rewards/margins": 0.09716212749481201, | |
| "rewards/rejected": -0.33267340064048767, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": 0.22058384120464325, | |
| "logits/rejected": 0.2929172217845917, | |
| "logps/chosen": -517.9133911132812, | |
| "logps/rejected": -579.1270141601562, | |
| "loss": 0.3049, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.18619771301746368, | |
| "rewards/margins": 0.10236430168151855, | |
| "rewards/rejected": -0.28856199979782104, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": 0.2658401131629944, | |
| "logits/rejected": 0.2887721657752991, | |
| "logps/chosen": -540.28125, | |
| "logps/rejected": -608.2423095703125, | |
| "loss": 0.304, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.19344370067119598, | |
| "rewards/margins": 0.10303628444671631, | |
| "rewards/rejected": -0.2964799702167511, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": 0.17139685153961182, | |
| "logits/rejected": 0.25828295946121216, | |
| "logps/chosen": -554.6468505859375, | |
| "logps/rejected": -640.1256103515625, | |
| "loss": 0.282, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.20378851890563965, | |
| "rewards/margins": 0.1270802915096283, | |
| "rewards/rejected": -0.33086878061294556, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": 0.2666166126728058, | |
| "logits/rejected": 0.34795981645584106, | |
| "logps/chosen": -539.4295654296875, | |
| "logps/rejected": -628.8125, | |
| "loss": 0.2971, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.2073674499988556, | |
| "rewards/margins": 0.1060541421175003, | |
| "rewards/rejected": -0.3134215772151947, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": 0.21775129437446594, | |
| "logits/rejected": 0.293344110250473, | |
| "logps/chosen": -570.99658203125, | |
| "logps/rejected": -677.3284301757812, | |
| "loss": 0.2876, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.20963303744792938, | |
| "rewards/margins": 0.11293938010931015, | |
| "rewards/rejected": -0.32257241010665894, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": 0.24225249886512756, | |
| "logits/rejected": 0.24781346321105957, | |
| "logps/chosen": -576.57666015625, | |
| "logps/rejected": -700.5311279296875, | |
| "loss": 0.2831, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.21174940466880798, | |
| "rewards/margins": 0.13153058290481567, | |
| "rewards/rejected": -0.34328001737594604, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": 0.24054893851280212, | |
| "logits/rejected": 0.26931822299957275, | |
| "logps/chosen": -555.5228881835938, | |
| "logps/rejected": -626.7105712890625, | |
| "loss": 0.3005, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.21045789122581482, | |
| "rewards/margins": 0.11446176469326019, | |
| "rewards/rejected": -0.3249196708202362, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": 0.2166041135787964, | |
| "logits/rejected": 0.3280956745147705, | |
| "logps/chosen": -568.7992553710938, | |
| "logps/rejected": -648.0155029296875, | |
| "loss": 0.2852, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.21820878982543945, | |
| "rewards/margins": 0.10663551092147827, | |
| "rewards/rejected": -0.32484427094459534, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": 0.239153191447258, | |
| "logits/rejected": 0.2556217610836029, | |
| "logps/chosen": -576.3333740234375, | |
| "logps/rejected": -648.7257080078125, | |
| "loss": 0.3209, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.23230747878551483, | |
| "rewards/margins": 0.08656927943229675, | |
| "rewards/rejected": -0.31887680292129517, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": 0.18877582252025604, | |
| "logits/rejected": 0.29884201288223267, | |
| "logps/chosen": -558.5150146484375, | |
| "logps/rejected": -645.3088989257812, | |
| "loss": 0.2991, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.21622328460216522, | |
| "rewards/margins": 0.11353172361850739, | |
| "rewards/rejected": -0.3297550082206726, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": 0.21539130806922913, | |
| "logits/rejected": 0.27139174938201904, | |
| "logps/chosen": -597.243408203125, | |
| "logps/rejected": -675.239501953125, | |
| "loss": 0.3016, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.2269672453403473, | |
| "rewards/margins": 0.10424689948558807, | |
| "rewards/rejected": -0.33121412992477417, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": 0.23871974647045135, | |
| "logits/rejected": 0.27142205834388733, | |
| "logps/chosen": -519.5438232421875, | |
| "logps/rejected": -582.7971801757812, | |
| "loss": 0.3158, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.19175615906715393, | |
| "rewards/margins": 0.08461669087409973, | |
| "rewards/rejected": -0.27637284994125366, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": 0.2282760590314865, | |
| "logits/rejected": 0.3373679518699646, | |
| "logps/chosen": -577.5682373046875, | |
| "logps/rejected": -626.8815307617188, | |
| "loss": 0.3167, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.22001275420188904, | |
| "rewards/margins": 0.07437258958816528, | |
| "rewards/rejected": -0.2943853735923767, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": 0.20522919297218323, | |
| "logits/rejected": 0.24701757729053497, | |
| "logps/chosen": -569.3855590820312, | |
| "logps/rejected": -645.0745849609375, | |
| "loss": 0.3067, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.22419647872447968, | |
| "rewards/margins": 0.0874943733215332, | |
| "rewards/rejected": -0.3116908669471741, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": 0.22225090861320496, | |
| "logits/rejected": 0.26438599824905396, | |
| "logps/chosen": -567.0126953125, | |
| "logps/rejected": -654.3960571289062, | |
| "loss": 0.2847, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.20670703053474426, | |
| "rewards/margins": 0.11432477086782455, | |
| "rewards/rejected": -0.32103174924850464, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": 0.24451366066932678, | |
| "logits/rejected": 0.3054306209087372, | |
| "logps/chosen": -528.2220458984375, | |
| "logps/rejected": -643.4144287109375, | |
| "loss": 0.2967, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.19906611740589142, | |
| "rewards/margins": 0.1229342445731163, | |
| "rewards/rejected": -0.32200032472610474, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": 0.2621709704399109, | |
| "logits/rejected": 0.3254636228084564, | |
| "logps/chosen": -604.9237060546875, | |
| "logps/rejected": -632.830322265625, | |
| "loss": 0.3013, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.22578959167003632, | |
| "rewards/margins": 0.09746529161930084, | |
| "rewards/rejected": -0.32325488328933716, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.24207222728871738, | |
| "train_runtime": 8784.9129, | |
| "train_samples_per_second": 3.415, | |
| "train_steps_per_second": 0.107 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 937, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |