| { | |
| "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": 0.07539638876914978, | |
| "logits/rejected": 0.3617660701274872, | |
| "logps/chosen": -391.36358642578125, | |
| "logps/rejected": -351.79266357421875, | |
| "loss": 0.3823, | |
| "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": 0.1823267638683319, | |
| "logits/rejected": 0.30340075492858887, | |
| "logps/chosen": -419.4432373046875, | |
| "logps/rejected": -388.3770446777344, | |
| "loss": 0.377, | |
| "rewards/accuracies": 0.3680555522441864, | |
| "rewards/chosen": -0.0008056727820076048, | |
| "rewards/margins": 0.00010184818529523909, | |
| "rewards/rejected": -0.0009075209964066744, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.20788808166980743, | |
| "logits/rejected": 0.23254188895225525, | |
| "logps/chosen": -416.3082580566406, | |
| "logps/rejected": -387.38531494140625, | |
| "loss": 0.3736, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.0008876107749529183, | |
| "rewards/margins": -1.8838401956600137e-05, | |
| "rewards/rejected": -0.0008687723311595619, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.30031031370162964, | |
| "logits/rejected": 0.15295112133026123, | |
| "logps/chosen": -425.69256591796875, | |
| "logps/rejected": -429.0267639160156, | |
| "loss": 0.3744, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": -0.000916177115868777, | |
| "rewards/margins": 9.849786147242412e-05, | |
| "rewards/rejected": -0.0010146748973056674, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.208384707570076, | |
| "logits/rejected": 0.18005970120429993, | |
| "logps/chosen": -401.88677978515625, | |
| "logps/rejected": -385.52313232421875, | |
| "loss": 0.3671, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.000506656477227807, | |
| "rewards/margins": 2.4630819098092616e-05, | |
| "rewards/rejected": -0.0005312872817739844, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.20258402824401855, | |
| "logits/rejected": 0.2793141305446625, | |
| "logps/chosen": -387.0594177246094, | |
| "logps/rejected": -397.8907165527344, | |
| "loss": 0.3671, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.00127916329074651, | |
| "rewards/margins": 5.479707760969177e-05, | |
| "rewards/rejected": -0.0013339603319764137, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.15620803833007812, | |
| "logits/rejected": 0.1983640044927597, | |
| "logps/chosen": -393.00323486328125, | |
| "logps/rejected": -389.00091552734375, | |
| "loss": 0.376, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": -0.0012187453685328364, | |
| "rewards/margins": 0.0002520512498449534, | |
| "rewards/rejected": -0.0014707966474816203, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.2471141517162323, | |
| "logits/rejected": 0.23372165858745575, | |
| "logps/chosen": -422.782470703125, | |
| "logps/rejected": -409.0695495605469, | |
| "loss": 0.361, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -6.791514897486195e-05, | |
| "rewards/margins": 0.00030927141779102385, | |
| "rewards/rejected": -0.00037718654493801296, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.3184092342853546, | |
| "logits/rejected": 0.2963259518146515, | |
| "logps/chosen": -430.31878662109375, | |
| "logps/rejected": -441.485107421875, | |
| "loss": 0.3777, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.0043855770491063595, | |
| "rewards/margins": 0.0006927828071638942, | |
| "rewards/rejected": 0.0036927941255271435, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.21129055321216583, | |
| "logits/rejected": 0.1998635232448578, | |
| "logps/chosen": -372.2428894042969, | |
| "logps/rejected": -381.28765869140625, | |
| "loss": 0.3832, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.009740473702549934, | |
| "rewards/margins": 0.0007645908044651151, | |
| "rewards/rejected": 0.008975883945822716, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.21418185532093048, | |
| "logits/rejected": 0.31756535172462463, | |
| "logps/chosen": -408.43646240234375, | |
| "logps/rejected": -402.8905944824219, | |
| "loss": 0.3762, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": 0.008459924720227718, | |
| "rewards/margins": 0.0009970309911295772, | |
| "rewards/rejected": 0.007462893612682819, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.1467427909374237, | |
| "logits/rejected": 0.12404219806194305, | |
| "logps/chosen": -399.4310302734375, | |
| "logps/rejected": -400.8306884765625, | |
| "loss": 0.392, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.002305335830897093, | |
| "rewards/margins": 0.0017863574903458357, | |
| "rewards/rejected": 0.0005189782241359353, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.1542399823665619, | |
| "logits/rejected": 0.148942768573761, | |
| "logps/chosen": -427.81634521484375, | |
| "logps/rejected": -432.287109375, | |
| "loss": 0.373, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.015028266236186028, | |
| "rewards/margins": 0.003260324476286769, | |
| "rewards/rejected": -0.018288591876626015, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.11779679358005524, | |
| "logits/rejected": 0.21365532279014587, | |
| "logps/chosen": -380.7659912109375, | |
| "logps/rejected": -397.8762512207031, | |
| "loss": 0.3522, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.029504213482141495, | |
| "rewards/margins": 0.004709848202764988, | |
| "rewards/rejected": -0.03421406075358391, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.2473243921995163, | |
| "logits/rejected": 0.16422489285469055, | |
| "logps/chosen": -465.10711669921875, | |
| "logps/rejected": -491.26904296875, | |
| "loss": 0.3624, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.0538320355117321, | |
| "rewards/margins": 0.007020815275609493, | |
| "rewards/rejected": -0.06085285544395447, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.16355708241462708, | |
| "logits/rejected": 0.12124369293451309, | |
| "logps/chosen": -456.09307861328125, | |
| "logps/rejected": -461.9251403808594, | |
| "loss": 0.3809, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.07473380863666534, | |
| "rewards/margins": 0.00868835486471653, | |
| "rewards/rejected": -0.08342215418815613, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.11441226303577423, | |
| "logits/rejected": 0.2628568708896637, | |
| "logps/chosen": -494.88946533203125, | |
| "logps/rejected": -473.18304443359375, | |
| "loss": 0.3558, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.09776302427053452, | |
| "rewards/margins": 0.013915225863456726, | |
| "rewards/rejected": -0.11167825758457184, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.07239022850990295, | |
| "logits/rejected": 0.14864209294319153, | |
| "logps/chosen": -531.4957275390625, | |
| "logps/rejected": -559.1382446289062, | |
| "loss": 0.3526, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.12973885238170624, | |
| "rewards/margins": 0.02028977870941162, | |
| "rewards/rejected": -0.15002863109111786, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.026783421635627747, | |
| "logits/rejected": 0.09836456179618835, | |
| "logps/chosen": -559.0204467773438, | |
| "logps/rejected": -592.0581665039062, | |
| "loss": 0.3391, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.14703384041786194, | |
| "rewards/margins": 0.03492132946848869, | |
| "rewards/rejected": -0.18195518851280212, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.037959806621074677, | |
| "logits/rejected": 0.03822736442089081, | |
| "logps/chosen": -631.74365234375, | |
| "logps/rejected": -699.8177490234375, | |
| "loss": 0.3379, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.20971231162548065, | |
| "rewards/margins": 0.06991457939147949, | |
| "rewards/rejected": -0.27962690591812134, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": -0.024742988869547844, | |
| "logits/rejected": 0.0659240335226059, | |
| "logps/chosen": -594.5228271484375, | |
| "logps/rejected": -611.7971801757812, | |
| "loss": 0.3779, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.21358820796012878, | |
| "rewards/margins": 0.021843884140253067, | |
| "rewards/rejected": -0.23543210327625275, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.08687545359134674, | |
| "logits/rejected": 0.03665763512253761, | |
| "logps/chosen": -591.6993408203125, | |
| "logps/rejected": -627.2652587890625, | |
| "loss": 0.3521, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.16842979192733765, | |
| "rewards/margins": 0.035929836332798004, | |
| "rewards/rejected": -0.20435962080955505, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.03541503846645355, | |
| "logits/rejected": 0.1400221884250641, | |
| "logps/chosen": -533.7828369140625, | |
| "logps/rejected": -550.1221313476562, | |
| "loss": 0.3479, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.1404583603143692, | |
| "rewards/margins": 0.026776695623993874, | |
| "rewards/rejected": -0.16723506152629852, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": -0.02703988552093506, | |
| "logits/rejected": 0.0387953519821167, | |
| "logps/chosen": -544.995361328125, | |
| "logps/rejected": -571.0706176757812, | |
| "loss": 0.3344, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.1598329395055771, | |
| "rewards/margins": 0.03278448060154915, | |
| "rewards/rejected": -0.19261741638183594, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": 0.032285176217556, | |
| "logits/rejected": 0.03509441763162613, | |
| "logps/chosen": -533.4124755859375, | |
| "logps/rejected": -583.5164794921875, | |
| "loss": 0.3273, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.18457840383052826, | |
| "rewards/margins": 0.05398773029446602, | |
| "rewards/rejected": -0.23856613039970398, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": 0.041622038930654526, | |
| "logits/rejected": 0.14562161266803741, | |
| "logps/chosen": -629.537353515625, | |
| "logps/rejected": -698.6555786132812, | |
| "loss": 0.3356, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.20703263580799103, | |
| "rewards/margins": 0.07567483186721802, | |
| "rewards/rejected": -0.28270745277404785, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": -0.05856188014149666, | |
| "logits/rejected": 0.12126834690570831, | |
| "logps/chosen": -572.82763671875, | |
| "logps/rejected": -595.30810546875, | |
| "loss": 0.3371, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.1882476508617401, | |
| "rewards/margins": 0.049884133040905, | |
| "rewards/rejected": -0.2381317913532257, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": 0.04912557825446129, | |
| "logits/rejected": 0.0359189435839653, | |
| "logps/chosen": -605.077392578125, | |
| "logps/rejected": -658.1011352539062, | |
| "loss": 0.3329, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.19004479050636292, | |
| "rewards/margins": 0.06682942062616348, | |
| "rewards/rejected": -0.2568742036819458, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": -0.04056745022535324, | |
| "logits/rejected": 0.09117879718542099, | |
| "logps/chosen": -607.6412353515625, | |
| "logps/rejected": -669.0006103515625, | |
| "loss": 0.337, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -0.20891769230365753, | |
| "rewards/margins": 0.07138092070817947, | |
| "rewards/rejected": -0.2802986204624176, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": 0.10313913971185684, | |
| "logits/rejected": -0.032890208065509796, | |
| "logps/chosen": -645.5613403320312, | |
| "logps/rejected": -766.77685546875, | |
| "loss": 0.3408, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.24055080115795135, | |
| "rewards/margins": 0.07783609628677368, | |
| "rewards/rejected": -0.31838691234588623, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": -0.10391390323638916, | |
| "logits/rejected": 0.03353533148765564, | |
| "logps/chosen": -579.8109130859375, | |
| "logps/rejected": -659.9757080078125, | |
| "loss": 0.3052, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.2175387442111969, | |
| "rewards/margins": 0.08425301313400269, | |
| "rewards/rejected": -0.3017917573451996, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": -0.04866858571767807, | |
| "logits/rejected": 0.06607881933450699, | |
| "logps/chosen": -654.9437255859375, | |
| "logps/rejected": -746.3504638671875, | |
| "loss": 0.3188, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.21447813510894775, | |
| "rewards/margins": 0.0885075181722641, | |
| "rewards/rejected": -0.30298566818237305, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": -0.08321225643157959, | |
| "logits/rejected": 0.02133125625550747, | |
| "logps/chosen": -561.8787231445312, | |
| "logps/rejected": -612.1022338867188, | |
| "loss": 0.3395, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.1877143234014511, | |
| "rewards/margins": 0.0630793645977974, | |
| "rewards/rejected": -0.2507936656475067, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": -0.008984372019767761, | |
| "logits/rejected": 0.020171385258436203, | |
| "logps/chosen": -611.2442626953125, | |
| "logps/rejected": -706.269775390625, | |
| "loss": 0.3215, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.20123758912086487, | |
| "rewards/margins": 0.08567605912685394, | |
| "rewards/rejected": -0.28691360354423523, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": 0.025400999933481216, | |
| "logits/rejected": 0.12927094101905823, | |
| "logps/chosen": -630.3967895507812, | |
| "logps/rejected": -654.9688110351562, | |
| "loss": 0.334, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.22138580679893494, | |
| "rewards/margins": 0.044631440192461014, | |
| "rewards/rejected": -0.26601722836494446, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": 0.06408616155385971, | |
| "logits/rejected": 0.09784664213657379, | |
| "logps/chosen": -646.9287719726562, | |
| "logps/rejected": -651.2354736328125, | |
| "loss": 0.3722, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.22586758434772491, | |
| "rewards/margins": 0.029935162514448166, | |
| "rewards/rejected": -0.255802720785141, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": -0.011558137834072113, | |
| "logits/rejected": 0.08119530230760574, | |
| "logps/chosen": -505.4185485839844, | |
| "logps/rejected": -552.3299560546875, | |
| "loss": 0.3726, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.16497987508773804, | |
| "rewards/margins": 0.024407176300883293, | |
| "rewards/rejected": -0.18938703835010529, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": 0.03600141033530235, | |
| "logits/rejected": 0.11767586320638657, | |
| "logps/chosen": -537.4278564453125, | |
| "logps/rejected": -573.6280517578125, | |
| "loss": 0.3597, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.15761788189411163, | |
| "rewards/margins": 0.04068927466869354, | |
| "rewards/rejected": -0.19830715656280518, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": 0.0605524480342865, | |
| "logits/rejected": 0.04419669136404991, | |
| "logps/chosen": -567.4146118164062, | |
| "logps/rejected": -611.6965942382812, | |
| "loss": 0.3484, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.16456562280654907, | |
| "rewards/margins": 0.045978985726833344, | |
| "rewards/rejected": -0.210544615983963, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": 0.06785359233617783, | |
| "logits/rejected": 0.029774094000458717, | |
| "logps/chosen": -584.1076049804688, | |
| "logps/rejected": -663.6577758789062, | |
| "loss": 0.3269, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.18006856739521027, | |
| "rewards/margins": 0.05508347600698471, | |
| "rewards/rejected": -0.23515205085277557, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": 0.004598482511937618, | |
| "logits/rejected": 0.034767135977745056, | |
| "logps/chosen": -539.9484252929688, | |
| "logps/rejected": -576.4744262695312, | |
| "loss": 0.3491, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.17368122935295105, | |
| "rewards/margins": 0.03443225473165512, | |
| "rewards/rejected": -0.20811347663402557, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": 0.09188822656869888, | |
| "logits/rejected": 0.043555885553359985, | |
| "logps/chosen": -514.0177612304688, | |
| "logps/rejected": -582.0616455078125, | |
| "loss": 0.298, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.15144076943397522, | |
| "rewards/margins": 0.07400556653738022, | |
| "rewards/rejected": -0.22544634342193604, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": -0.004905997309833765, | |
| "logits/rejected": 0.040687330067157745, | |
| "logps/chosen": -578.5809326171875, | |
| "logps/rejected": -645.4890747070312, | |
| "loss": 0.3125, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.18565499782562256, | |
| "rewards/margins": 0.07045260071754456, | |
| "rewards/rejected": -0.2561075985431671, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": 0.030105963349342346, | |
| "logits/rejected": 0.14350736141204834, | |
| "logps/chosen": -614.5733642578125, | |
| "logps/rejected": -665.1195678710938, | |
| "loss": 0.3496, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.21485750377178192, | |
| "rewards/margins": 0.06542513519525528, | |
| "rewards/rejected": -0.2802826464176178, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": -0.033211253583431244, | |
| "logits/rejected": -0.030144518241286278, | |
| "logps/chosen": -654.5762939453125, | |
| "logps/rejected": -726.3638916015625, | |
| "loss": 0.3242, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.21793492138385773, | |
| "rewards/margins": 0.08130789548158646, | |
| "rewards/rejected": -0.2992427945137024, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": -0.011334149166941643, | |
| "logits/rejected": 0.09229062497615814, | |
| "logps/chosen": -581.5757446289062, | |
| "logps/rejected": -688.3638916015625, | |
| "loss": 0.3043, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.1874237060546875, | |
| "rewards/margins": 0.09659047424793243, | |
| "rewards/rejected": -0.28401416540145874, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": 0.033252302557229996, | |
| "logits/rejected": -0.009059171192348003, | |
| "logps/chosen": -611.2565307617188, | |
| "logps/rejected": -677.837646484375, | |
| "loss": 0.318, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.19349732995033264, | |
| "rewards/margins": 0.07655458152294159, | |
| "rewards/rejected": -0.27005189657211304, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": -0.06340669840574265, | |
| "logits/rejected": -0.0008907020092010498, | |
| "logps/chosen": -603.5202026367188, | |
| "logps/rejected": -673.9171752929688, | |
| "loss": 0.3375, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.21294096112251282, | |
| "rewards/margins": 0.07124762237071991, | |
| "rewards/rejected": -0.28418856859207153, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": 0.025766396895051003, | |
| "logits/rejected": 0.0163635965436697, | |
| "logps/chosen": -653.7705078125, | |
| "logps/rejected": -762.9187622070312, | |
| "loss": 0.3379, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.23698878288269043, | |
| "rewards/margins": 0.08854912966489792, | |
| "rewards/rejected": -0.32553791999816895, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": 0.07303385436534882, | |
| "logits/rejected": -0.03683042526245117, | |
| "logps/chosen": -620.9476318359375, | |
| "logps/rejected": -691.1075439453125, | |
| "loss": 0.2908, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.18487033247947693, | |
| "rewards/margins": 0.09076735377311707, | |
| "rewards/rejected": -0.275637686252594, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": -0.07815182954072952, | |
| "logits/rejected": -0.010281940922141075, | |
| "logps/chosen": -641.4134521484375, | |
| "logps/rejected": -706.6598510742188, | |
| "loss": 0.3186, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.2273898422718048, | |
| "rewards/margins": 0.08002471923828125, | |
| "rewards/rejected": -0.30741453170776367, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": -0.1134188324213028, | |
| "logits/rejected": -0.011527213267982006, | |
| "logps/chosen": -614.1680908203125, | |
| "logps/rejected": -699.6660766601562, | |
| "loss": 0.326, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.2197011411190033, | |
| "rewards/margins": 0.07829214632511139, | |
| "rewards/rejected": -0.2979933023452759, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": 0.04103394225239754, | |
| "logits/rejected": 0.009344205260276794, | |
| "logps/chosen": -588.2103271484375, | |
| "logps/rejected": -654.2349853515625, | |
| "loss": 0.3201, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.1895955502986908, | |
| "rewards/margins": 0.07577956467866898, | |
| "rewards/rejected": -0.2653751075267792, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": 0.026180285960435867, | |
| "logits/rejected": -0.057329438626766205, | |
| "logps/chosen": -636.9844970703125, | |
| "logps/rejected": -699.1533813476562, | |
| "loss": 0.3382, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.21527251601219177, | |
| "rewards/margins": 0.06197941303253174, | |
| "rewards/rejected": -0.2772519588470459, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.09959807246923447, | |
| "logits/rejected": 0.034060269594192505, | |
| "logps/chosen": -618.979248046875, | |
| "logps/rejected": -660.626220703125, | |
| "loss": 0.3315, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.2118925154209137, | |
| "rewards/margins": 0.08266349881887436, | |
| "rewards/rejected": -0.29455602169036865, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": -0.0629156231880188, | |
| "logits/rejected": 0.06692133843898773, | |
| "logps/chosen": -657.5345458984375, | |
| "logps/rejected": -740.254638671875, | |
| "loss": 0.312, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.24066373705863953, | |
| "rewards/margins": 0.09562061727046967, | |
| "rewards/rejected": -0.3362843990325928, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": -0.019260389730334282, | |
| "logits/rejected": 0.044143788516521454, | |
| "logps/chosen": -611.283447265625, | |
| "logps/rejected": -677.8404541015625, | |
| "loss": 0.3327, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.2026386559009552, | |
| "rewards/margins": 0.06741929799318314, | |
| "rewards/rejected": -0.27005794644355774, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": -0.10834735631942749, | |
| "logits/rejected": 0.014343018643558025, | |
| "logps/chosen": -584.4176635742188, | |
| "logps/rejected": -631.9229736328125, | |
| "loss": 0.3439, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.20305252075195312, | |
| "rewards/margins": 0.061229951679706573, | |
| "rewards/rejected": -0.2642824649810791, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": -0.059888023883104324, | |
| "logits/rejected": 0.10185569524765015, | |
| "logps/chosen": -684.3859252929688, | |
| "logps/rejected": -750.8058471679688, | |
| "loss": 0.3211, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.22337186336517334, | |
| "rewards/margins": 0.0742068737745285, | |
| "rewards/rejected": -0.29757875204086304, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": -0.08253158628940582, | |
| "logits/rejected": 0.15623542666435242, | |
| "logps/chosen": -641.6725463867188, | |
| "logps/rejected": -700.8177490234375, | |
| "loss": 0.3242, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.18204794824123383, | |
| "rewards/margins": 0.0828886330127716, | |
| "rewards/rejected": -0.2649365961551666, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": -0.07821720838546753, | |
| "logits/rejected": 0.09279079735279083, | |
| "logps/chosen": -574.6522216796875, | |
| "logps/rejected": -609.2244262695312, | |
| "loss": 0.3267, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -0.17055127024650574, | |
| "rewards/margins": 0.056847114115953445, | |
| "rewards/rejected": -0.2273983657360077, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": -0.14354786276817322, | |
| "logits/rejected": 0.06310894340276718, | |
| "logps/chosen": -606.2435913085938, | |
| "logps/rejected": -621.083251953125, | |
| "loss": 0.32, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.189885213971138, | |
| "rewards/margins": 0.05728505924344063, | |
| "rewards/rejected": -0.24717025458812714, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": -0.002320159925147891, | |
| "logits/rejected": -0.11932463943958282, | |
| "logps/chosen": -649.9357299804688, | |
| "logps/rejected": -733.55908203125, | |
| "loss": 0.3075, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.2011071890592575, | |
| "rewards/margins": 0.08436000347137451, | |
| "rewards/rejected": -0.2854672074317932, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": -0.029952585697174072, | |
| "logits/rejected": -0.007862815633416176, | |
| "logps/chosen": -579.582275390625, | |
| "logps/rejected": -615.4111328125, | |
| "loss": 0.3215, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -0.18586835265159607, | |
| "rewards/margins": 0.06111454218626022, | |
| "rewards/rejected": -0.2469829022884369, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": 0.006623442284762859, | |
| "logits/rejected": -0.05229584500193596, | |
| "logps/chosen": -651.7686767578125, | |
| "logps/rejected": -718.4263916015625, | |
| "loss": 0.349, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.22425830364227295, | |
| "rewards/margins": 0.07099372893571854, | |
| "rewards/rejected": -0.2952520251274109, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.057037353515625, | |
| "logits/rejected": -0.0304800383746624, | |
| "logps/chosen": -614.2849731445312, | |
| "logps/rejected": -700.621826171875, | |
| "loss": 0.3187, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.21285679936408997, | |
| "rewards/margins": 0.07572495192289352, | |
| "rewards/rejected": -0.2885817587375641, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": -0.06914228200912476, | |
| "logits/rejected": 0.01761237345635891, | |
| "logps/chosen": -562.9390869140625, | |
| "logps/rejected": -623.0194091796875, | |
| "loss": 0.3213, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.18713270127773285, | |
| "rewards/margins": 0.07669297605752945, | |
| "rewards/rejected": -0.2638256847858429, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": -0.02265956625342369, | |
| "logits/rejected": 0.01389719732105732, | |
| "logps/chosen": -605.0291748046875, | |
| "logps/rejected": -664.8007202148438, | |
| "loss": 0.3292, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.19236616790294647, | |
| "rewards/margins": 0.0655684620141983, | |
| "rewards/rejected": -0.2579346299171448, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": -0.05592130497097969, | |
| "logits/rejected": 0.02719860151410103, | |
| "logps/chosen": -636.6947021484375, | |
| "logps/rejected": -714.4791870117188, | |
| "loss": 0.3358, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.22510233521461487, | |
| "rewards/margins": 0.07263530045747757, | |
| "rewards/rejected": -0.29773762822151184, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": -0.03522191569209099, | |
| "logits/rejected": -0.05070105940103531, | |
| "logps/chosen": -558.2520751953125, | |
| "logps/rejected": -610.8953857421875, | |
| "loss": 0.3335, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.17959460616111755, | |
| "rewards/margins": 0.06889108568429947, | |
| "rewards/rejected": -0.24848568439483643, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": 0.04719340056180954, | |
| "logits/rejected": -0.004802383482456207, | |
| "logps/chosen": -631.9532470703125, | |
| "logps/rejected": -683.6273193359375, | |
| "loss": 0.3369, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.20903649926185608, | |
| "rewards/margins": 0.048954375088214874, | |
| "rewards/rejected": -0.25799083709716797, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.06102383881807327, | |
| "logits/rejected": -0.040299173444509506, | |
| "logps/chosen": -619.0484008789062, | |
| "logps/rejected": -694.1537475585938, | |
| "loss": 0.3325, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.20765328407287598, | |
| "rewards/margins": 0.06029241159558296, | |
| "rewards/rejected": -0.2679457366466522, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": -0.0017960123950615525, | |
| "logits/rejected": 0.0005670204991474748, | |
| "logps/chosen": -625.3905029296875, | |
| "logps/rejected": -722.47705078125, | |
| "loss": 0.3109, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.2023344784975052, | |
| "rewards/margins": 0.08977413177490234, | |
| "rewards/rejected": -0.29210859537124634, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": -0.0029361299239099026, | |
| "logits/rejected": -0.00539384176954627, | |
| "logps/chosen": -571.76611328125, | |
| "logps/rejected": -648.0147705078125, | |
| "loss": 0.3257, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.17501702904701233, | |
| "rewards/margins": 0.07722989469766617, | |
| "rewards/rejected": -0.2522469162940979, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": -0.009814179502427578, | |
| "logits/rejected": 0.030293231830000877, | |
| "logps/chosen": -626.85400390625, | |
| "logps/rejected": -675.728759765625, | |
| "loss": 0.3388, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.21242670714855194, | |
| "rewards/margins": 0.046240173280239105, | |
| "rewards/rejected": -0.25866684317588806, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": 0.014827275648713112, | |
| "logits/rejected": -0.03816215693950653, | |
| "logps/chosen": -608.7626953125, | |
| "logps/rejected": -721.3744506835938, | |
| "loss": 0.3021, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.19283899664878845, | |
| "rewards/margins": 0.09409373998641968, | |
| "rewards/rejected": -0.28693273663520813, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": 0.009239016100764275, | |
| "logits/rejected": 0.0575685016810894, | |
| "logps/chosen": -571.3810424804688, | |
| "logps/rejected": -638.7860107421875, | |
| "loss": 0.3399, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.19390597939491272, | |
| "rewards/margins": 0.06127419322729111, | |
| "rewards/rejected": -0.2551801800727844, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": -0.11210556328296661, | |
| "logits/rejected": 0.02634884975850582, | |
| "logps/chosen": -591.846435546875, | |
| "logps/rejected": -665.6798706054688, | |
| "loss": 0.3203, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.18739867210388184, | |
| "rewards/margins": 0.07172352820634842, | |
| "rewards/rejected": -0.25912219285964966, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": -0.0635216236114502, | |
| "logits/rejected": 0.02560749091207981, | |
| "logps/chosen": -595.9927978515625, | |
| "logps/rejected": -671.517578125, | |
| "loss": 0.3428, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.20496638119220734, | |
| "rewards/margins": 0.08088068664073944, | |
| "rewards/rejected": -0.2858470678329468, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": -0.06199071928858757, | |
| "logits/rejected": -0.03149358555674553, | |
| "logps/chosen": -600.124755859375, | |
| "logps/rejected": -636.0254516601562, | |
| "loss": 0.3202, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.20770713686943054, | |
| "rewards/margins": 0.05365163832902908, | |
| "rewards/rejected": -0.26135876774787903, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": -0.039042651653289795, | |
| "logits/rejected": -0.12910965085029602, | |
| "logps/chosen": -588.3599853515625, | |
| "logps/rejected": -666.7010498046875, | |
| "loss": 0.3153, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.19020356237888336, | |
| "rewards/margins": 0.0770055502653122, | |
| "rewards/rejected": -0.26720911264419556, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": -0.021715031936764717, | |
| "logits/rejected": 0.05355725437402725, | |
| "logps/chosen": -592.3232421875, | |
| "logps/rejected": -638.9842529296875, | |
| "loss": 0.33, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.19993874430656433, | |
| "rewards/margins": 0.04968953877687454, | |
| "rewards/rejected": -0.24962830543518066, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": -0.06499023735523224, | |
| "logits/rejected": -0.037022653967142105, | |
| "logps/chosen": -583.3277587890625, | |
| "logps/rejected": -623.016845703125, | |
| "loss": 0.3404, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.18943414092063904, | |
| "rewards/margins": 0.073947012424469, | |
| "rewards/rejected": -0.2633811831474304, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": -0.08982490003108978, | |
| "logits/rejected": -0.010354668833315372, | |
| "logps/chosen": -636.1207275390625, | |
| "logps/rejected": -710.9208374023438, | |
| "loss": 0.3256, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.21537241339683533, | |
| "rewards/margins": 0.06745601445436478, | |
| "rewards/rejected": -0.2828284204006195, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": -0.08715302497148514, | |
| "logits/rejected": 0.018399229273200035, | |
| "logps/chosen": -614.1312866210938, | |
| "logps/rejected": -668.2174682617188, | |
| "loss": 0.3167, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.20197662711143494, | |
| "rewards/margins": 0.07475915551185608, | |
| "rewards/rejected": -0.276735782623291, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": -0.05179240554571152, | |
| "logits/rejected": -0.03785177320241928, | |
| "logps/chosen": -565.5599365234375, | |
| "logps/rejected": -678.8760986328125, | |
| "loss": 0.3032, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.17605264484882355, | |
| "rewards/margins": 0.10278842598199844, | |
| "rewards/rejected": -0.2788410484790802, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": -0.06850672513246536, | |
| "logits/rejected": -0.0343458354473114, | |
| "logps/chosen": -548.1768798828125, | |
| "logps/rejected": -605.9397583007812, | |
| "loss": 0.3453, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": -0.1724565625190735, | |
| "rewards/margins": 0.0625961571931839, | |
| "rewards/rejected": -0.23505273461341858, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": -0.14548218250274658, | |
| "logits/rejected": 0.019901303574442863, | |
| "logps/chosen": -573.3082275390625, | |
| "logps/rejected": -650.7000732421875, | |
| "loss": 0.3149, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.20980414748191833, | |
| "rewards/margins": 0.07451895624399185, | |
| "rewards/rejected": -0.284323126077652, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": -0.04322751238942146, | |
| "logits/rejected": -0.028643226251006126, | |
| "logps/chosen": -515.6051025390625, | |
| "logps/rejected": -579.2161865234375, | |
| "loss": 0.3281, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": -0.17263229191303253, | |
| "rewards/margins": 0.06705036014318466, | |
| "rewards/rejected": -0.2396826446056366, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": -0.06822247058153152, | |
| "logits/rejected": -0.01675710827112198, | |
| "logps/chosen": -676.0777587890625, | |
| "logps/rejected": -730.0950927734375, | |
| "loss": 0.3367, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.2302449494600296, | |
| "rewards/margins": 0.058430016040802, | |
| "rewards/rejected": -0.288674920797348, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": 0.05342342332005501, | |
| "logits/rejected": -0.031858671456575394, | |
| "logps/chosen": -590.1318969726562, | |
| "logps/rejected": -664.7706298828125, | |
| "loss": 0.3159, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.18570013344287872, | |
| "rewards/margins": 0.07330085337162018, | |
| "rewards/rejected": -0.2590009868144989, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": -0.10356259346008301, | |
| "logits/rejected": 0.014384737238287926, | |
| "logps/chosen": -551.7988891601562, | |
| "logps/rejected": -598.3231201171875, | |
| "loss": 0.3249, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.17019973695278168, | |
| "rewards/margins": 0.07013366371393204, | |
| "rewards/rejected": -0.24033339321613312, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": 0.0009760938701219857, | |
| "logits/rejected": 0.020660245791077614, | |
| "logps/chosen": -624.172119140625, | |
| "logps/rejected": -702.16845703125, | |
| "loss": 0.3345, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.19060316681861877, | |
| "rewards/margins": 0.07204783707857132, | |
| "rewards/rejected": -0.2626509964466095, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": -0.052719853818416595, | |
| "logits/rejected": 0.04510030895471573, | |
| "logps/chosen": -581.1450805664062, | |
| "logps/rejected": -657.2957763671875, | |
| "loss": 0.3416, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.19285398721694946, | |
| "rewards/margins": 0.057259153574705124, | |
| "rewards/rejected": -0.2501131296157837, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.3375262364570206, | |
| "train_runtime": 5783.6809, | |
| "train_samples_per_second": 2.594, | |
| "train_steps_per_second": 0.162 | |
| } | |
| ], | |
| "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 | |
| } | |