| { | |
| "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": -0.07764632999897003, | |
| "logits/rejected": 0.0654495358467102, | |
| "logps/chosen": -548.1927490234375, | |
| "logps/rejected": -445.62200927734375, | |
| "loss": 0.21, | |
| "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": 0.013010119087994099, | |
| "logits/rejected": 0.14687097072601318, | |
| "logps/chosen": -445.48004150390625, | |
| "logps/rejected": -474.8713073730469, | |
| "loss": 0.2064, | |
| "rewards/accuracies": 0.3333333432674408, | |
| "rewards/chosen": 0.0016270489431917667, | |
| "rewards/margins": -3.060871677007526e-05, | |
| "rewards/rejected": 0.0016576575580984354, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.04200610890984535, | |
| "logits/rejected": 0.11701512336730957, | |
| "logps/chosen": -480.87042236328125, | |
| "logps/rejected": -453.9954528808594, | |
| "loss": 0.2121, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.002087645698338747, | |
| "rewards/margins": 4.953762254444882e-05, | |
| "rewards/rejected": 0.002038107719272375, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.08031486719846725, | |
| "logits/rejected": 0.17916838824748993, | |
| "logps/chosen": -533.2267456054688, | |
| "logps/rejected": -509.03192138671875, | |
| "loss": 0.2044, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": 0.001751051051542163, | |
| "rewards/margins": 0.0012214130256325006, | |
| "rewards/rejected": 0.0005296378512866795, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.02258063293993473, | |
| "logits/rejected": 0.07948300242424011, | |
| "logps/chosen": -482.404296875, | |
| "logps/rejected": -482.5491638183594, | |
| "loss": 0.2131, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.0007890533888712525, | |
| "rewards/margins": 0.002143475692719221, | |
| "rewards/rejected": -0.002932528965175152, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.12140221893787384, | |
| "logits/rejected": 0.09028098732233047, | |
| "logps/chosen": -442.23797607421875, | |
| "logps/rejected": -465.88214111328125, | |
| "loss": 0.2105, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.009099994786083698, | |
| "rewards/margins": 0.002986655803397298, | |
| "rewards/rejected": -0.012086650356650352, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.053414009511470795, | |
| "logits/rejected": 0.18009646236896515, | |
| "logps/chosen": -528.6683959960938, | |
| "logps/rejected": -486.89361572265625, | |
| "loss": 0.1982, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.015702595934271812, | |
| "rewards/margins": 0.0040222653187811375, | |
| "rewards/rejected": -0.019724860787391663, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.07392764091491699, | |
| "logits/rejected": 0.11563147604465485, | |
| "logps/chosen": -534.7530517578125, | |
| "logps/rejected": -519.1510009765625, | |
| "loss": 0.2073, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.02576013281941414, | |
| "rewards/margins": 0.006931398995220661, | |
| "rewards/rejected": -0.032691530883312225, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.0789504274725914, | |
| "logits/rejected": -0.028738999739289284, | |
| "logps/chosen": -516.7589111328125, | |
| "logps/rejected": -550.5140991210938, | |
| "loss": 0.1998, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.02832980453968048, | |
| "rewards/margins": 0.013306397013366222, | |
| "rewards/rejected": -0.04163619875907898, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.007517285645008087, | |
| "logits/rejected": 0.06739212572574615, | |
| "logps/chosen": -530.9830932617188, | |
| "logps/rejected": -524.5928955078125, | |
| "loss": 0.2029, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": -0.06454873085021973, | |
| "rewards/margins": 0.008291224017739296, | |
| "rewards/rejected": -0.07283996045589447, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.023702986538410187, | |
| "logits/rejected": 0.037184469401836395, | |
| "logps/chosen": -489.89581298828125, | |
| "logps/rejected": -484.0779724121094, | |
| "loss": 0.1966, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.07171402126550674, | |
| "rewards/margins": 0.024490345269441605, | |
| "rewards/rejected": -0.09620436280965805, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.09514528512954712, | |
| "logits/rejected": 0.029327446594834328, | |
| "logps/chosen": -453.439697265625, | |
| "logps/rejected": -483.3421325683594, | |
| "loss": 0.1919, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.06944920122623444, | |
| "rewards/margins": 0.02075176127254963, | |
| "rewards/rejected": -0.09020096063613892, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.052310071885585785, | |
| "logits/rejected": 0.023372933268547058, | |
| "logps/chosen": -465.06622314453125, | |
| "logps/rejected": -487.2718200683594, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.045129794627428055, | |
| "rewards/margins": 0.017120003700256348, | |
| "rewards/rejected": -0.0622497983276844, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.040262121707201004, | |
| "logits/rejected": 0.2257479429244995, | |
| "logps/chosen": -475.408935546875, | |
| "logps/rejected": -475.36669921875, | |
| "loss": 0.1949, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": -0.03786056488752365, | |
| "rewards/margins": 0.022892726585268974, | |
| "rewards/rejected": -0.060753293335437775, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.16925258934497833, | |
| "logits/rejected": 0.07987479865550995, | |
| "logps/chosen": -481.04083251953125, | |
| "logps/rejected": -517.6348876953125, | |
| "loss": 0.1979, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.07286442816257477, | |
| "rewards/margins": 0.0315692201256752, | |
| "rewards/rejected": -0.10443364083766937, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.021633749827742577, | |
| "logits/rejected": 0.06488915532827377, | |
| "logps/chosen": -476.21661376953125, | |
| "logps/rejected": -496.6123046875, | |
| "loss": 0.1908, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.10675158351659775, | |
| "rewards/margins": 0.03962003067135811, | |
| "rewards/rejected": -0.14637163281440735, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.05234859511256218, | |
| "logits/rejected": 0.02733474411070347, | |
| "logps/chosen": -506.60382080078125, | |
| "logps/rejected": -550.2049560546875, | |
| "loss": 0.1971, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.12149013578891754, | |
| "rewards/margins": 0.03788003325462341, | |
| "rewards/rejected": -0.15937016904354095, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.00539384875446558, | |
| "logits/rejected": 0.001964741852134466, | |
| "logps/chosen": -469.29840087890625, | |
| "logps/rejected": -486.98504638671875, | |
| "loss": 0.2055, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.0949864536523819, | |
| "rewards/margins": 0.023784717544913292, | |
| "rewards/rejected": -0.11877117305994034, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.03078017197549343, | |
| "logits/rejected": 0.06562785059213638, | |
| "logps/chosen": -467.913818359375, | |
| "logps/rejected": -488.88165283203125, | |
| "loss": 0.1832, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.08011925220489502, | |
| "rewards/margins": 0.03867987170815468, | |
| "rewards/rejected": -0.1187991127371788, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": -0.012808884494006634, | |
| "logits/rejected": 0.07382214069366455, | |
| "logps/chosen": -470.2506408691406, | |
| "logps/rejected": -483.8114318847656, | |
| "loss": 0.2019, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.08619646728038788, | |
| "rewards/margins": 0.021857351064682007, | |
| "rewards/rejected": -0.10805382579565048, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": 0.11153273284435272, | |
| "logits/rejected": 0.0497104674577713, | |
| "logps/chosen": -451.3377380371094, | |
| "logps/rejected": -478.8421936035156, | |
| "loss": 0.1942, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.0883895754814148, | |
| "rewards/margins": 0.030626222491264343, | |
| "rewards/rejected": -0.11901579797267914, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.08737102895975113, | |
| "logits/rejected": 0.11837353557348251, | |
| "logps/chosen": -519.548828125, | |
| "logps/rejected": -513.87255859375, | |
| "loss": 0.1924, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.11379089206457138, | |
| "rewards/margins": 0.03358360007405281, | |
| "rewards/rejected": -0.1473744958639145, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.14104892313480377, | |
| "logits/rejected": 0.05087919905781746, | |
| "logps/chosen": -525.2891235351562, | |
| "logps/rejected": -516.625, | |
| "loss": 0.2024, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.10455243289470673, | |
| "rewards/margins": 0.0242080669850111, | |
| "rewards/rejected": -0.12876050174236298, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": 0.09732341766357422, | |
| "logits/rejected": 0.06271795928478241, | |
| "logps/chosen": -514.5999755859375, | |
| "logps/rejected": -543.45068359375, | |
| "loss": 0.2006, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.1029946357011795, | |
| "rewards/margins": 0.033023733645677567, | |
| "rewards/rejected": -0.13601836562156677, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": 0.06812410056591034, | |
| "logits/rejected": 0.025554204359650612, | |
| "logps/chosen": -516.6173095703125, | |
| "logps/rejected": -508.93511962890625, | |
| "loss": 0.2079, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.0961836650967598, | |
| "rewards/margins": 0.020848505198955536, | |
| "rewards/rejected": -0.11703218519687653, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": 0.052641429007053375, | |
| "logits/rejected": 0.11277129501104355, | |
| "logps/chosen": -499.9371032714844, | |
| "logps/rejected": -524.9378662109375, | |
| "loss": 0.2003, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.11769168078899384, | |
| "rewards/margins": 0.02954215370118618, | |
| "rewards/rejected": -0.14723384380340576, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": 0.1458149254322052, | |
| "logits/rejected": 0.18058298528194427, | |
| "logps/chosen": -461.63262939453125, | |
| "logps/rejected": -501.23077392578125, | |
| "loss": 0.195, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.1235838383436203, | |
| "rewards/margins": 0.02654830738902092, | |
| "rewards/rejected": -0.15013214945793152, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": 0.07014019787311554, | |
| "logits/rejected": 0.16694627702236176, | |
| "logps/chosen": -472.4525451660156, | |
| "logps/rejected": -473.22857666015625, | |
| "loss": 0.1957, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.12277780473232269, | |
| "rewards/margins": 0.01745077781379223, | |
| "rewards/rejected": -0.14022858440876007, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": 0.11594029515981674, | |
| "logits/rejected": 0.14667686820030212, | |
| "logps/chosen": -454.9984436035156, | |
| "logps/rejected": -468.27764892578125, | |
| "loss": 0.1927, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.14300423860549927, | |
| "rewards/margins": 0.037807125598192215, | |
| "rewards/rejected": -0.18081137537956238, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": 0.08191246539354324, | |
| "logits/rejected": 0.13855530321598053, | |
| "logps/chosen": -496.906982421875, | |
| "logps/rejected": -526.3932495117188, | |
| "loss": 0.1933, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.12783119082450867, | |
| "rewards/margins": 0.03418290615081787, | |
| "rewards/rejected": -0.16201409697532654, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": 0.10729117691516876, | |
| "logits/rejected": 0.1382167488336563, | |
| "logps/chosen": -525.2513427734375, | |
| "logps/rejected": -557.9173583984375, | |
| "loss": 0.1848, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.10007178783416748, | |
| "rewards/margins": 0.03654087707400322, | |
| "rewards/rejected": -0.136612668633461, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": 0.03605717048048973, | |
| "logits/rejected": 0.08247358351945877, | |
| "logps/chosen": -476.7731018066406, | |
| "logps/rejected": -517.1722412109375, | |
| "loss": 0.1901, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.0743183121085167, | |
| "rewards/margins": 0.028351595625281334, | |
| "rewards/rejected": -0.10266990959644318, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": 0.03332200273871422, | |
| "logits/rejected": 0.09907984733581543, | |
| "logps/chosen": -449.68212890625, | |
| "logps/rejected": -509.12554931640625, | |
| "loss": 0.1926, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.05537894368171692, | |
| "rewards/margins": 0.05774994567036629, | |
| "rewards/rejected": -0.11312887817621231, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": 0.05158273130655289, | |
| "logits/rejected": 0.10417857021093369, | |
| "logps/chosen": -498.09197998046875, | |
| "logps/rejected": -511.01611328125, | |
| "loss": 0.1957, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.06420507282018661, | |
| "rewards/margins": 0.012455371208488941, | |
| "rewards/rejected": -0.07666043937206268, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": 0.015417411923408508, | |
| "logits/rejected": 0.09142233431339264, | |
| "logps/chosen": -502.4671936035156, | |
| "logps/rejected": -528.1393432617188, | |
| "loss": 0.187, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.0401633158326149, | |
| "rewards/margins": 0.03752883896231651, | |
| "rewards/rejected": -0.07769216597080231, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": 0.07565109431743622, | |
| "logits/rejected": 0.14677348732948303, | |
| "logps/chosen": -490.3994140625, | |
| "logps/rejected": -483.0618591308594, | |
| "loss": 0.182, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.000983755337074399, | |
| "rewards/margins": 0.04362579435110092, | |
| "rewards/rejected": -0.04460955038666725, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": -0.012587164528667927, | |
| "logits/rejected": 0.17615702748298645, | |
| "logps/chosen": -460.7539978027344, | |
| "logps/rejected": -496.0394592285156, | |
| "loss": 0.1765, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.004288173280656338, | |
| "rewards/margins": 0.045940086245536804, | |
| "rewards/rejected": -0.05022825673222542, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": 0.05247374624013901, | |
| "logits/rejected": 0.06078363209962845, | |
| "logps/chosen": -500.44384765625, | |
| "logps/rejected": -511.1734313964844, | |
| "loss": 0.189, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.0024426232557743788, | |
| "rewards/margins": 0.04215018078684807, | |
| "rewards/rejected": -0.03970755636692047, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": 0.04354017972946167, | |
| "logits/rejected": 0.12157300859689713, | |
| "logps/chosen": -477.10601806640625, | |
| "logps/rejected": -487.3301696777344, | |
| "loss": 0.1834, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": 0.022890402004122734, | |
| "rewards/margins": 0.03842080011963844, | |
| "rewards/rejected": -0.01553039439022541, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": 0.04744056984782219, | |
| "logits/rejected": 0.07940800487995148, | |
| "logps/chosen": -472.5918884277344, | |
| "logps/rejected": -496.16387939453125, | |
| "loss": 0.1833, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.019969001412391663, | |
| "rewards/margins": 0.04812890291213989, | |
| "rewards/rejected": -0.02815990522503853, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": 0.04268989711999893, | |
| "logits/rejected": 0.07916353642940521, | |
| "logps/chosen": -486.46746826171875, | |
| "logps/rejected": -520.3671875, | |
| "loss": 0.1851, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": 0.029743488878011703, | |
| "rewards/margins": 0.04279039055109024, | |
| "rewards/rejected": -0.013046903535723686, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": -0.027468422427773476, | |
| "logits/rejected": -0.04074520990252495, | |
| "logps/chosen": -461.34814453125, | |
| "logps/rejected": -480.4091796875, | |
| "loss": 0.1842, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": 0.058516185730695724, | |
| "rewards/margins": 0.04239815101027489, | |
| "rewards/rejected": 0.016118034720420837, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": 0.029957780614495277, | |
| "logits/rejected": 0.1293373703956604, | |
| "logps/chosen": -476.6888732910156, | |
| "logps/rejected": -474.581298828125, | |
| "loss": 0.1863, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": 0.06844775378704071, | |
| "rewards/margins": 0.0345512256026268, | |
| "rewards/rejected": 0.03389652818441391, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": 0.12380461394786835, | |
| "logits/rejected": 0.08030427992343903, | |
| "logps/chosen": -501.05419921875, | |
| "logps/rejected": -540.5864868164062, | |
| "loss": 0.1833, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.08108590543270111, | |
| "rewards/margins": 0.04042986035346985, | |
| "rewards/rejected": 0.040656037628650665, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": 0.07574252784252167, | |
| "logits/rejected": -0.00224049249663949, | |
| "logps/chosen": -482.531982421875, | |
| "logps/rejected": -499.04180908203125, | |
| "loss": 0.1822, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.0877564549446106, | |
| "rewards/margins": 0.03748675435781479, | |
| "rewards/rejected": 0.05026970058679581, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": 0.10215386003255844, | |
| "logits/rejected": 0.025258684530854225, | |
| "logps/chosen": -455.39825439453125, | |
| "logps/rejected": -464.11016845703125, | |
| "loss": 0.1929, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.11565859615802765, | |
| "rewards/margins": 0.04002625495195389, | |
| "rewards/rejected": 0.07563233375549316, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": 0.07208961248397827, | |
| "logits/rejected": 0.05293772369623184, | |
| "logps/chosen": -422.0474548339844, | |
| "logps/rejected": -466.1214904785156, | |
| "loss": 0.1867, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.13345687091350555, | |
| "rewards/margins": 0.04373367875814438, | |
| "rewards/rejected": 0.08972318470478058, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": 0.08753691613674164, | |
| "logits/rejected": 0.13366851210594177, | |
| "logps/chosen": -493.6244201660156, | |
| "logps/rejected": -481.61505126953125, | |
| "loss": 0.1971, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.15995556116104126, | |
| "rewards/margins": 0.030286097899079323, | |
| "rewards/rejected": 0.1296694576740265, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": 0.040917035192251205, | |
| "logits/rejected": 0.049655430018901825, | |
| "logps/chosen": -496.69207763671875, | |
| "logps/rejected": -510.538818359375, | |
| "loss": 0.1827, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.16031843423843384, | |
| "rewards/margins": 0.047999344766139984, | |
| "rewards/rejected": 0.11231908947229385, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": -0.02139192819595337, | |
| "logits/rejected": 0.07283684611320496, | |
| "logps/chosen": -455.2607421875, | |
| "logps/rejected": -484.5707092285156, | |
| "loss": 0.2065, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.13160817325115204, | |
| "rewards/margins": 0.03165828809142113, | |
| "rewards/rejected": 0.09994987398386002, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": 0.043560974299907684, | |
| "logits/rejected": 0.0819135308265686, | |
| "logps/chosen": -448.69793701171875, | |
| "logps/rejected": -472.7060546875, | |
| "loss": 0.1844, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.10212703794240952, | |
| "rewards/margins": 0.038218505680561066, | |
| "rewards/rejected": 0.06390853226184845, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": 0.005092448554933071, | |
| "logits/rejected": 0.10958313941955566, | |
| "logps/chosen": -484.2552185058594, | |
| "logps/rejected": -514.481689453125, | |
| "loss": 0.1942, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.09142176806926727, | |
| "rewards/margins": 0.05243723839521408, | |
| "rewards/rejected": 0.038984514772892, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": 0.09447602182626724, | |
| "logits/rejected": 0.08024080842733383, | |
| "logps/chosen": -485.6512756347656, | |
| "logps/rejected": -482.0237731933594, | |
| "loss": 0.1891, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.07477296888828278, | |
| "rewards/margins": 0.05883296579122543, | |
| "rewards/rejected": 0.015940004959702492, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": 0.03767072409391403, | |
| "logits/rejected": 0.02741827443242073, | |
| "logps/chosen": -440.016357421875, | |
| "logps/rejected": -472.63873291015625, | |
| "loss": 0.1917, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.08402599394321442, | |
| "rewards/margins": 0.061398398131132126, | |
| "rewards/rejected": 0.02262760140001774, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.00708337128162384, | |
| "logits/rejected": 0.055642373859882355, | |
| "logps/chosen": -446.92498779296875, | |
| "logps/rejected": -466.16912841796875, | |
| "loss": 0.1827, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.07480621337890625, | |
| "rewards/margins": 0.033216286450624466, | |
| "rewards/rejected": 0.04158991575241089, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": 0.035645630210638046, | |
| "logits/rejected": 0.019252488389611244, | |
| "logps/chosen": -421.6669921875, | |
| "logps/rejected": -462.30279541015625, | |
| "loss": 0.1782, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.08298784494400024, | |
| "rewards/margins": 0.05089961364865303, | |
| "rewards/rejected": 0.032088227570056915, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": 0.03386590629816055, | |
| "logits/rejected": 0.0352010652422905, | |
| "logps/chosen": -497.4388732910156, | |
| "logps/rejected": -539.6199951171875, | |
| "loss": 0.184, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.07756824791431427, | |
| "rewards/margins": 0.06193417310714722, | |
| "rewards/rejected": 0.01563408225774765, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": 0.06176334619522095, | |
| "logits/rejected": 0.05909004807472229, | |
| "logps/chosen": -479.58502197265625, | |
| "logps/rejected": -504.15118408203125, | |
| "loss": 0.179, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": 0.07643002271652222, | |
| "rewards/margins": 0.055131517350673676, | |
| "rewards/rejected": 0.021298503503203392, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": 0.10232093185186386, | |
| "logits/rejected": 0.03354436904191971, | |
| "logps/chosen": -504.35552978515625, | |
| "logps/rejected": -500.55242919921875, | |
| "loss": 0.188, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.0535406656563282, | |
| "rewards/margins": 0.045547787100076675, | |
| "rewards/rejected": 0.007992876693606377, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": 0.08256170898675919, | |
| "logits/rejected": 0.011359182186424732, | |
| "logps/chosen": -423.5184631347656, | |
| "logps/rejected": -462.12921142578125, | |
| "loss": 0.1847, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.06936874985694885, | |
| "rewards/margins": 0.04546703025698662, | |
| "rewards/rejected": 0.023901721462607384, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": 0.05957316234707832, | |
| "logits/rejected": 0.02687280997633934, | |
| "logps/chosen": -463.92340087890625, | |
| "logps/rejected": -488.68524169921875, | |
| "loss": 0.1841, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.093163400888443, | |
| "rewards/margins": 0.05278082564473152, | |
| "rewards/rejected": 0.04038257524371147, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": 0.03663691133260727, | |
| "logits/rejected": 0.06973805278539658, | |
| "logps/chosen": -460.896728515625, | |
| "logps/rejected": -508.50408935546875, | |
| "loss": 0.1778, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.0800352543592453, | |
| "rewards/margins": 0.05085815116763115, | |
| "rewards/rejected": 0.029177110642194748, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": 0.028588274493813515, | |
| "logits/rejected": 0.081044040620327, | |
| "logps/chosen": -504.3495178222656, | |
| "logps/rejected": -515.7532958984375, | |
| "loss": 0.1676, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.07437646389007568, | |
| "rewards/margins": 0.04987211152911186, | |
| "rewards/rejected": 0.02450435608625412, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": 0.03259013965725899, | |
| "logits/rejected": 0.0049851625226438046, | |
| "logps/chosen": -485.429443359375, | |
| "logps/rejected": -495.66802978515625, | |
| "loss": 0.1771, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.07386799156665802, | |
| "rewards/margins": 0.05358237773180008, | |
| "rewards/rejected": 0.02028561569750309, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": 0.040917474776506424, | |
| "logits/rejected": 0.03524303436279297, | |
| "logps/chosen": -474.40252685546875, | |
| "logps/rejected": -492.69415283203125, | |
| "loss": 0.1848, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.08967775106430054, | |
| "rewards/margins": 0.04113636165857315, | |
| "rewards/rejected": 0.04854138568043709, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.0005787253612652421, | |
| "logits/rejected": 0.05223802477121353, | |
| "logps/chosen": -469.7098693847656, | |
| "logps/rejected": -505.01739501953125, | |
| "loss": 0.1979, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.08548668771982193, | |
| "rewards/margins": 0.057396382093429565, | |
| "rewards/rejected": 0.028090301901102066, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": 0.09496835619211197, | |
| "logits/rejected": -0.0715503841638565, | |
| "logps/chosen": -422.630859375, | |
| "logps/rejected": -483.22479248046875, | |
| "loss": 0.1824, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.08553356677293777, | |
| "rewards/margins": 0.059073131531476974, | |
| "rewards/rejected": 0.02646043337881565, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": -0.00498700188472867, | |
| "logits/rejected": 0.013168242760002613, | |
| "logps/chosen": -445.13861083984375, | |
| "logps/rejected": -466.5567932128906, | |
| "loss": 0.1776, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.10062942653894424, | |
| "rewards/margins": 0.04562296345829964, | |
| "rewards/rejected": 0.055006466805934906, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": 0.0634564682841301, | |
| "logits/rejected": 0.03239821270108223, | |
| "logps/chosen": -454.5580139160156, | |
| "logps/rejected": -471.24456787109375, | |
| "loss": 0.1788, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.11137165874242783, | |
| "rewards/margins": 0.06377272307872772, | |
| "rewards/rejected": 0.0475989505648613, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": 0.13198313117027283, | |
| "logits/rejected": -0.011327924206852913, | |
| "logps/chosen": -447.65301513671875, | |
| "logps/rejected": -485.073974609375, | |
| "loss": 0.1763, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.11758910119533539, | |
| "rewards/margins": 0.05051087588071823, | |
| "rewards/rejected": 0.06707822531461716, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": 0.01844317652285099, | |
| "logits/rejected": 0.046270035207271576, | |
| "logps/chosen": -461.8832092285156, | |
| "logps/rejected": -497.8064880371094, | |
| "loss": 0.1902, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": 0.11230479180812836, | |
| "rewards/margins": 0.038510359823703766, | |
| "rewards/rejected": 0.07379443198442459, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.0005433745682239532, | |
| "logits/rejected": 0.07239210605621338, | |
| "logps/chosen": -440.7948303222656, | |
| "logps/rejected": -482.3118591308594, | |
| "loss": 0.1805, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.11701339483261108, | |
| "rewards/margins": 0.05948741361498833, | |
| "rewards/rejected": 0.05752598121762276, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": 0.004731953144073486, | |
| "logits/rejected": 0.06949619203805923, | |
| "logps/chosen": -495.1181640625, | |
| "logps/rejected": -494.14727783203125, | |
| "loss": 0.1907, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.11087504774332047, | |
| "rewards/margins": 0.05067601799964905, | |
| "rewards/rejected": 0.060199033468961716, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": 0.05472816899418831, | |
| "logits/rejected": 0.12369577586650848, | |
| "logps/chosen": -454.3465270996094, | |
| "logps/rejected": -466.5943298339844, | |
| "loss": 0.1862, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.09320452064275742, | |
| "rewards/margins": 0.04919914901256561, | |
| "rewards/rejected": 0.0440053753554821, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": 0.049330659210681915, | |
| "logits/rejected": -0.05037214607000351, | |
| "logps/chosen": -426.7115173339844, | |
| "logps/rejected": -468.99169921875, | |
| "loss": 0.178, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": 0.1173645481467247, | |
| "rewards/margins": 0.06448864191770554, | |
| "rewards/rejected": 0.05287591367959976, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": 0.018861666321754456, | |
| "logits/rejected": 0.04406842216849327, | |
| "logps/chosen": -471.93438720703125, | |
| "logps/rejected": -496.49993896484375, | |
| "loss": 0.1862, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.11358505487442017, | |
| "rewards/margins": 0.046424929052591324, | |
| "rewards/rejected": 0.06716012954711914, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": 0.009539002552628517, | |
| "logits/rejected": 0.11306466907262802, | |
| "logps/chosen": -520.7066650390625, | |
| "logps/rejected": -506.4190368652344, | |
| "loss": 0.1811, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.11439801752567291, | |
| "rewards/margins": 0.042585574090480804, | |
| "rewards/rejected": 0.07181242108345032, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": 0.05899236351251602, | |
| "logits/rejected": -0.012183750048279762, | |
| "logps/chosen": -465.54583740234375, | |
| "logps/rejected": -492.99627685546875, | |
| "loss": 0.1822, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.11408071219921112, | |
| "rewards/margins": 0.05036580562591553, | |
| "rewards/rejected": 0.06371492147445679, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": 0.04782986640930176, | |
| "logits/rejected": 0.0820665955543518, | |
| "logps/chosen": -446.2923278808594, | |
| "logps/rejected": -473.13330078125, | |
| "loss": 0.1747, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": 0.10825562477111816, | |
| "rewards/margins": 0.04990454763174057, | |
| "rewards/rejected": 0.058351077139377594, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": -0.04199203476309776, | |
| "logits/rejected": 0.034929946064949036, | |
| "logps/chosen": -446.6978454589844, | |
| "logps/rejected": -435.03521728515625, | |
| "loss": 0.1749, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": 0.09576048702001572, | |
| "rewards/margins": 0.045824047178030014, | |
| "rewards/rejected": 0.0499364398419857, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": 0.026381831616163254, | |
| "logits/rejected": 0.07602212578058243, | |
| "logps/chosen": -467.7432556152344, | |
| "logps/rejected": -516.2023315429688, | |
| "loss": 0.1863, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.11635438352823257, | |
| "rewards/margins": 0.06682081520557404, | |
| "rewards/rejected": 0.04953356832265854, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": 0.04555773735046387, | |
| "logits/rejected": 0.0836203470826149, | |
| "logps/chosen": -432.5563049316406, | |
| "logps/rejected": -469.57196044921875, | |
| "loss": 0.1816, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.11761243641376495, | |
| "rewards/margins": 0.05147726088762283, | |
| "rewards/rejected": 0.06613515317440033, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": 0.06100771576166153, | |
| "logits/rejected": -0.052108317613601685, | |
| "logps/chosen": -411.35906982421875, | |
| "logps/rejected": -439.3312072753906, | |
| "loss": 0.1838, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.12352752685546875, | |
| "rewards/margins": 0.057254206389188766, | |
| "rewards/rejected": 0.06627332419157028, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": 0.0011749922996386886, | |
| "logits/rejected": -0.003739739302545786, | |
| "logps/chosen": -458.92919921875, | |
| "logps/rejected": -491.05615234375, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.11504600197076797, | |
| "rewards/margins": 0.04097086191177368, | |
| "rewards/rejected": 0.0740751400589943, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": 0.06734688580036163, | |
| "logits/rejected": 0.08023635298013687, | |
| "logps/chosen": -494.57342529296875, | |
| "logps/rejected": -526.6209716796875, | |
| "loss": 0.1857, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.1077314019203186, | |
| "rewards/margins": 0.04536820203065872, | |
| "rewards/rejected": 0.06236319988965988, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": 0.03414052724838257, | |
| "logits/rejected": 0.0967903584241867, | |
| "logps/chosen": -406.6197204589844, | |
| "logps/rejected": -463.4610900878906, | |
| "loss": 0.1849, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": 0.12588708102703094, | |
| "rewards/margins": 0.05215496942400932, | |
| "rewards/rejected": 0.07373212277889252, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": 0.046077877283096313, | |
| "logits/rejected": -0.010932741686701775, | |
| "logps/chosen": -486.12933349609375, | |
| "logps/rejected": -490.4742736816406, | |
| "loss": 0.1796, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.11952760070562363, | |
| "rewards/margins": 0.052067436277866364, | |
| "rewards/rejected": 0.06746017187833786, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": 0.08364848792552948, | |
| "logits/rejected": -0.016565704718232155, | |
| "logps/chosen": -421.2782287597656, | |
| "logps/rejected": -448.32061767578125, | |
| "loss": 0.1673, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": 0.12084163725376129, | |
| "rewards/margins": 0.05717577412724495, | |
| "rewards/rejected": 0.06366586685180664, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": 0.05045370012521744, | |
| "logits/rejected": -0.015437754802405834, | |
| "logps/chosen": -436.0224609375, | |
| "logps/rejected": -480.992431640625, | |
| "loss": 0.1805, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.1131979450583458, | |
| "rewards/margins": 0.05461297556757927, | |
| "rewards/rejected": 0.058584969490766525, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": 0.0034905076026916504, | |
| "logits/rejected": 0.10833680629730225, | |
| "logps/chosen": -497.405517578125, | |
| "logps/rejected": -531.6438598632812, | |
| "loss": 0.1902, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.10887560993432999, | |
| "rewards/margins": 0.053155817091464996, | |
| "rewards/rejected": 0.05571979284286499, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": 0.09158419072628021, | |
| "logits/rejected": 0.12686944007873535, | |
| "logps/chosen": -453.55206298828125, | |
| "logps/rejected": -507.8880920410156, | |
| "loss": 0.1797, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.11705918610095978, | |
| "rewards/margins": 0.061758268624544144, | |
| "rewards/rejected": 0.055300913751125336, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": -0.005747838411480188, | |
| "logits/rejected": 0.02690596505999565, | |
| "logps/chosen": -485.0928649902344, | |
| "logps/rejected": -488.77349853515625, | |
| "loss": 0.1824, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.10394562780857086, | |
| "rewards/margins": 0.04496193677186966, | |
| "rewards/rejected": 0.058983683586120605, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": 0.04409351199865341, | |
| "logits/rejected": 0.023466479033231735, | |
| "logps/chosen": -516.257080078125, | |
| "logps/rejected": -529.1635131835938, | |
| "loss": 0.186, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.11666563898324966, | |
| "rewards/margins": 0.0654011145234108, | |
| "rewards/rejected": 0.051264531910419464, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": 0.029077816754579544, | |
| "logits/rejected": 0.1013563722372055, | |
| "logps/chosen": -449.57379150390625, | |
| "logps/rejected": -469.8511657714844, | |
| "loss": 0.1967, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": 0.10130287706851959, | |
| "rewards/margins": 0.050037987530231476, | |
| "rewards/rejected": 0.05126488208770752, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.1892308688214647, | |
| "train_runtime": 7960.6994, | |
| "train_samples_per_second": 3.769, | |
| "train_steps_per_second": 0.118 | |
| } | |
| ], | |
| "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 | |
| } | |