| { | |
| "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": 0.5103299021720886, | |
| "logits/rejected": 0.45260655879974365, | |
| "logps/chosen": -462.9068603515625, | |
| "logps/rejected": -475.4417419433594, | |
| "loss": 0.2285, | |
| "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": 0.31862032413482666, | |
| "logits/rejected": 0.314761757850647, | |
| "logps/chosen": -292.8707275390625, | |
| "logps/rejected": -296.7555847167969, | |
| "loss": 0.2107, | |
| "rewards/accuracies": 0.2361111044883728, | |
| "rewards/chosen": 0.00015502750466112047, | |
| "rewards/margins": -4.963619721820578e-05, | |
| "rewards/rejected": 0.0002046637237071991, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.31885525584220886, | |
| "logits/rejected": 0.3816074728965759, | |
| "logps/chosen": -255.02761840820312, | |
| "logps/rejected": -261.906005859375, | |
| "loss": 0.2118, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": -0.00024467133334837854, | |
| "rewards/margins": 3.665524491225369e-05, | |
| "rewards/rejected": -0.0002813266182783991, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.306856244802475, | |
| "logits/rejected": 0.26547059416770935, | |
| "logps/chosen": -271.63592529296875, | |
| "logps/rejected": -279.66046142578125, | |
| "loss": 0.2112, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.00020208054047543555, | |
| "rewards/margins": -1.5414796507684514e-05, | |
| "rewards/rejected": -0.00018666575488168746, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.3179105222225189, | |
| "logits/rejected": 0.37513914704322815, | |
| "logps/chosen": -291.2155456542969, | |
| "logps/rejected": -309.3365173339844, | |
| "loss": 0.2082, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": -0.0003900996525771916, | |
| "rewards/margins": -8.274164429167286e-05, | |
| "rewards/rejected": -0.00030735801556147635, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.3564581871032715, | |
| "logits/rejected": 0.328947514295578, | |
| "logps/chosen": -265.4711608886719, | |
| "logps/rejected": -290.76092529296875, | |
| "loss": 0.1932, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -4.1023264202522114e-05, | |
| "rewards/margins": 8.93441028892994e-06, | |
| "rewards/rejected": -4.995765630155802e-05, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.3046855330467224, | |
| "logits/rejected": 0.3436508774757385, | |
| "logps/chosen": -353.6334228515625, | |
| "logps/rejected": -360.71929931640625, | |
| "loss": 0.2105, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": -8.662064647069201e-05, | |
| "rewards/margins": 0.00014803092926740646, | |
| "rewards/rejected": -0.00023465156846214086, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.31014540791511536, | |
| "logits/rejected": 0.3247402608394623, | |
| "logps/chosen": -270.9269104003906, | |
| "logps/rejected": -301.1846008300781, | |
| "loss": 0.2008, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": -0.0006521933828480542, | |
| "rewards/margins": -0.00015351924230344594, | |
| "rewards/rejected": -0.0004986740532331169, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.2668500542640686, | |
| "logits/rejected": 0.2699311673641205, | |
| "logps/chosen": -279.9768371582031, | |
| "logps/rejected": -305.3199157714844, | |
| "loss": 0.2116, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.001971206162124872, | |
| "rewards/margins": 0.00017751348786987364, | |
| "rewards/rejected": -0.002148719970136881, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.29766160249710083, | |
| "logits/rejected": 0.3357502818107605, | |
| "logps/chosen": -335.7276306152344, | |
| "logps/rejected": -333.04852294921875, | |
| "loss": 0.2104, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -0.0029864246025681496, | |
| "rewards/margins": 0.00020421878434717655, | |
| "rewards/rejected": -0.003190643386915326, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.32819682359695435, | |
| "logits/rejected": 0.3049379885196686, | |
| "logps/chosen": -341.62030029296875, | |
| "logps/rejected": -362.6277160644531, | |
| "loss": 0.2148, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.002841348061338067, | |
| "rewards/margins": 0.0008320774650201201, | |
| "rewards/rejected": -0.003673425642773509, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.25718414783477783, | |
| "logits/rejected": 0.22829556465148926, | |
| "logps/chosen": -262.38275146484375, | |
| "logps/rejected": -263.3617248535156, | |
| "loss": 0.212, | |
| "rewards/accuracies": 0.21250000596046448, | |
| "rewards/chosen": -0.002418057294562459, | |
| "rewards/margins": -0.0002635159471537918, | |
| "rewards/rejected": -0.002154541201889515, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.2612871825695038, | |
| "logits/rejected": 0.20154158771038055, | |
| "logps/chosen": -331.4267883300781, | |
| "logps/rejected": -325.03582763671875, | |
| "loss": 0.2083, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.0029244977049529552, | |
| "rewards/margins": 0.00017272381228394806, | |
| "rewards/rejected": -0.003097221488133073, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.22134217619895935, | |
| "logits/rejected": 0.236228346824646, | |
| "logps/chosen": -259.1029968261719, | |
| "logps/rejected": -295.55267333984375, | |
| "loss": 0.2013, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -0.0034433710388839245, | |
| "rewards/margins": 0.0002571121440269053, | |
| "rewards/rejected": -0.003700483590364456, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.24982547760009766, | |
| "logits/rejected": 0.255920946598053, | |
| "logps/chosen": -319.05792236328125, | |
| "logps/rejected": -311.7349853515625, | |
| "loss": 0.2154, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": -0.003797318320721388, | |
| "rewards/margins": 0.001411179080605507, | |
| "rewards/rejected": -0.005208497401326895, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.19847659766674042, | |
| "logits/rejected": 0.2311740666627884, | |
| "logps/chosen": -297.9468994140625, | |
| "logps/rejected": -310.4879455566406, | |
| "loss": 0.2133, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -9.477639105170965e-05, | |
| "rewards/margins": 0.0019024491775780916, | |
| "rewards/rejected": -0.0019972254522144794, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.19259515404701233, | |
| "logits/rejected": 0.15325775742530823, | |
| "logps/chosen": -304.2120361328125, | |
| "logps/rejected": -318.15814208984375, | |
| "loss": 0.2097, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.002039076527580619, | |
| "rewards/margins": -0.0007348087383434176, | |
| "rewards/rejected": 0.002773885615170002, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.2160247564315796, | |
| "logits/rejected": 0.1700972020626068, | |
| "logps/chosen": -297.9866943359375, | |
| "logps/rejected": -340.97308349609375, | |
| "loss": 0.2019, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.005410970654338598, | |
| "rewards/margins": 0.0020456851925700903, | |
| "rewards/rejected": 0.003365286160260439, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.13720422983169556, | |
| "logits/rejected": 0.17013493180274963, | |
| "logps/chosen": -263.00592041015625, | |
| "logps/rejected": -285.49493408203125, | |
| "loss": 0.203, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.006990239955484867, | |
| "rewards/margins": 0.004235823173075914, | |
| "rewards/rejected": 0.0027544170152395964, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.1691962480545044, | |
| "logits/rejected": 0.18480929732322693, | |
| "logps/chosen": -260.39678955078125, | |
| "logps/rejected": -315.83148193359375, | |
| "loss": 0.2089, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": 0.009463444352149963, | |
| "rewards/margins": 0.0045218681916594505, | |
| "rewards/rejected": 0.0049415766261518, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": 0.08825172483921051, | |
| "logits/rejected": 0.15803390741348267, | |
| "logps/chosen": -288.59637451171875, | |
| "logps/rejected": -313.7530822753906, | |
| "loss": 0.2001, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.008411525748670101, | |
| "rewards/margins": 0.00624421751126647, | |
| "rewards/rejected": 0.0021673087030649185, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.09255161136388779, | |
| "logits/rejected": 0.08204738795757294, | |
| "logps/chosen": -287.4985046386719, | |
| "logps/rejected": -305.6175231933594, | |
| "loss": 0.2157, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.0021202426869422197, | |
| "rewards/margins": 0.00259004021063447, | |
| "rewards/rejected": -0.00046979711623862386, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.07696634531021118, | |
| "logits/rejected": 0.0523553304374218, | |
| "logps/chosen": -276.10125732421875, | |
| "logps/rejected": -297.9394836425781, | |
| "loss": 0.213, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.0040213195607066154, | |
| "rewards/margins": 0.005335698835551739, | |
| "rewards/rejected": -0.001314379507675767, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": 0.11084002256393433, | |
| "logits/rejected": 0.025312194600701332, | |
| "logps/chosen": -257.67864990234375, | |
| "logps/rejected": -297.58941650390625, | |
| "loss": 0.2099, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.00831601768732071, | |
| "rewards/margins": 0.004710269160568714, | |
| "rewards/rejected": 0.0036057487595826387, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": 0.1027657762169838, | |
| "logits/rejected": 0.07919518649578094, | |
| "logps/chosen": -280.58990478515625, | |
| "logps/rejected": -299.32000732421875, | |
| "loss": 0.2063, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.01294595468789339, | |
| "rewards/margins": 0.004328696057200432, | |
| "rewards/rejected": 0.008617257699370384, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": 0.03375561907887459, | |
| "logits/rejected": 0.06997523456811905, | |
| "logps/chosen": -232.6851043701172, | |
| "logps/rejected": -268.51544189453125, | |
| "loss": 0.2106, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.0066901808604598045, | |
| "rewards/margins": 0.004283884074538946, | |
| "rewards/rejected": 0.0024062972515821457, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": 0.018621403723955154, | |
| "logits/rejected": 0.031526364386081696, | |
| "logps/chosen": -255.9957275390625, | |
| "logps/rejected": -266.628173828125, | |
| "loss": 0.2014, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.0048848409205675125, | |
| "rewards/margins": 0.006752178072929382, | |
| "rewards/rejected": -0.0018673368031159043, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": 0.012805347330868244, | |
| "logits/rejected": 0.07229005545377731, | |
| "logps/chosen": -299.4040222167969, | |
| "logps/rejected": -313.70330810546875, | |
| "loss": 0.2113, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.009505004622042179, | |
| "rewards/margins": 0.009923577308654785, | |
| "rewards/rejected": -0.00041857213363982737, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": 0.015601048246026039, | |
| "logits/rejected": 0.0003178860933985561, | |
| "logps/chosen": -202.97805786132812, | |
| "logps/rejected": -224.36019897460938, | |
| "loss": 0.2096, | |
| "rewards/accuracies": 0.24375000596046448, | |
| "rewards/chosen": 0.00423224363476038, | |
| "rewards/margins": 0.0043428512290120125, | |
| "rewards/rejected": -0.00011060731776524335, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": -0.05607699602842331, | |
| "logits/rejected": -0.054743945598602295, | |
| "logps/chosen": -289.1156005859375, | |
| "logps/rejected": -311.282470703125, | |
| "loss": 0.1986, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.009154532104730606, | |
| "rewards/margins": 0.0036413397174328566, | |
| "rewards/rejected": 0.0055131930857896805, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": -0.047469399869441986, | |
| "logits/rejected": -0.019701087847352028, | |
| "logps/chosen": -285.7122497558594, | |
| "logps/rejected": -288.4371032714844, | |
| "loss": 0.2106, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.014235032722353935, | |
| "rewards/margins": 0.002786442870274186, | |
| "rewards/rejected": 0.011448590084910393, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": -0.08851752430200577, | |
| "logits/rejected": -0.04262327030301094, | |
| "logps/chosen": -283.068115234375, | |
| "logps/rejected": -291.9867248535156, | |
| "loss": 0.1975, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.0208656694740057, | |
| "rewards/margins": 0.0043783774599432945, | |
| "rewards/rejected": 0.016487296670675278, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": -0.017280403524637222, | |
| "logits/rejected": -0.0075028399005532265, | |
| "logps/chosen": -257.34454345703125, | |
| "logps/rejected": -289.70953369140625, | |
| "loss": 0.2046, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.013227777555584908, | |
| "rewards/margins": 0.0045712715946137905, | |
| "rewards/rejected": 0.008656504563987255, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": -0.04185862094163895, | |
| "logits/rejected": -0.07007891684770584, | |
| "logps/chosen": -287.055419921875, | |
| "logps/rejected": -345.94244384765625, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": 0.01658710278570652, | |
| "rewards/margins": 0.01041549164801836, | |
| "rewards/rejected": 0.00617161113768816, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": -0.1421864926815033, | |
| "logits/rejected": -0.06375002861022949, | |
| "logps/chosen": -326.5301208496094, | |
| "logps/rejected": -310.8834228515625, | |
| "loss": 0.2112, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.002902784151956439, | |
| "rewards/margins": -0.0022658538073301315, | |
| "rewards/rejected": -0.0006369305774569511, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": -0.10621865093708038, | |
| "logits/rejected": -0.07717886567115784, | |
| "logps/chosen": -333.44171142578125, | |
| "logps/rejected": -341.5768737792969, | |
| "loss": 0.2066, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.006154628936201334, | |
| "rewards/margins": 0.004178780596703291, | |
| "rewards/rejected": 0.0019758488051593304, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": -0.12794511020183563, | |
| "logits/rejected": -0.042172081768512726, | |
| "logps/chosen": -287.95379638671875, | |
| "logps/rejected": -327.0418701171875, | |
| "loss": 0.212, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.010754548013210297, | |
| "rewards/margins": 0.009464221075177193, | |
| "rewards/rejected": 0.0012903271708637476, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": -0.039213500916957855, | |
| "logits/rejected": -0.06362026929855347, | |
| "logps/chosen": -317.4774169921875, | |
| "logps/rejected": -290.7700500488281, | |
| "loss": 0.2108, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.007657637353986502, | |
| "rewards/margins": 0.001145793590694666, | |
| "rewards/rejected": 0.00651184469461441, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": -0.08858563005924225, | |
| "logits/rejected": -0.11657065153121948, | |
| "logps/chosen": -254.9708251953125, | |
| "logps/rejected": -272.658935546875, | |
| "loss": 0.2083, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.009081033989787102, | |
| "rewards/margins": 0.008375957608222961, | |
| "rewards/rejected": 0.0007050756248645484, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": -0.17941537499427795, | |
| "logits/rejected": -0.11935138702392578, | |
| "logps/chosen": -268.1692810058594, | |
| "logps/rejected": -290.07513427734375, | |
| "loss": 0.2068, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.009544593282043934, | |
| "rewards/margins": 0.0030928333289921284, | |
| "rewards/rejected": 0.006451760418713093, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": -0.0844186395406723, | |
| "logits/rejected": -0.16169007122516632, | |
| "logps/chosen": -207.09371948242188, | |
| "logps/rejected": -256.3302917480469, | |
| "loss": 0.2001, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.0057321893982589245, | |
| "rewards/margins": 0.006300176493823528, | |
| "rewards/rejected": -0.0005679865134879947, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": -0.14396438002586365, | |
| "logits/rejected": -0.12671178579330444, | |
| "logps/chosen": -260.51361083984375, | |
| "logps/rejected": -311.62823486328125, | |
| "loss": 0.2088, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": 0.01238117553293705, | |
| "rewards/margins": 0.013218119740486145, | |
| "rewards/rejected": -0.0008369443821720779, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": -0.15392616391181946, | |
| "logits/rejected": -0.13130106031894684, | |
| "logps/chosen": -273.2947082519531, | |
| "logps/rejected": -266.83770751953125, | |
| "loss": 0.2103, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.008025998249650002, | |
| "rewards/margins": 0.003167554037645459, | |
| "rewards/rejected": 0.004858444444835186, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": -0.12201549857854843, | |
| "logits/rejected": -0.13493959605693817, | |
| "logps/chosen": -292.46600341796875, | |
| "logps/rejected": -327.0417175292969, | |
| "loss": 0.2043, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.007237803190946579, | |
| "rewards/margins": 0.00763185927644372, | |
| "rewards/rejected": -0.0003940565511584282, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": -0.18244734406471252, | |
| "logits/rejected": -0.19711770117282867, | |
| "logps/chosen": -238.47006225585938, | |
| "logps/rejected": -273.8080139160156, | |
| "loss": 0.2071, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.006393319461494684, | |
| "rewards/margins": 0.004871468059718609, | |
| "rewards/rejected": 0.0015218511689454317, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": -0.1901955008506775, | |
| "logits/rejected": -0.2543238401412964, | |
| "logps/chosen": -316.27459716796875, | |
| "logps/rejected": -338.86798095703125, | |
| "loss": 0.2094, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.006844863295555115, | |
| "rewards/margins": 0.0016734026139602065, | |
| "rewards/rejected": 0.005171460565179586, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": -0.24599988758563995, | |
| "logits/rejected": -0.2185782939195633, | |
| "logps/chosen": -298.28271484375, | |
| "logps/rejected": -333.5831604003906, | |
| "loss": 0.2115, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.011538187973201275, | |
| "rewards/margins": 0.006993747316300869, | |
| "rewards/rejected": 0.004544440656900406, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": -0.24635422229766846, | |
| "logits/rejected": -0.19935253262519836, | |
| "logps/chosen": -315.05218505859375, | |
| "logps/rejected": -328.00213623046875, | |
| "loss": 0.2077, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.007171535398811102, | |
| "rewards/margins": 0.004863684065639973, | |
| "rewards/rejected": 0.002307850867509842, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": -0.198508158326149, | |
| "logits/rejected": -0.16920627653598785, | |
| "logps/chosen": -279.35040283203125, | |
| "logps/rejected": -301.2700500488281, | |
| "loss": 0.1993, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.007105141878128052, | |
| "rewards/margins": 0.0047775073908269405, | |
| "rewards/rejected": 0.0023276349529623985, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": -0.24639983475208282, | |
| "logits/rejected": -0.24618887901306152, | |
| "logps/chosen": -284.9212341308594, | |
| "logps/rejected": -297.228271484375, | |
| "loss": 0.2031, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.0051613058894872665, | |
| "rewards/margins": 0.004140165634453297, | |
| "rewards/rejected": 0.0010211400222033262, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": -0.26932382583618164, | |
| "logits/rejected": -0.24082553386688232, | |
| "logps/chosen": -272.0877990722656, | |
| "logps/rejected": -298.43389892578125, | |
| "loss": 0.2158, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.007115568965673447, | |
| "rewards/margins": 0.0035896040499210358, | |
| "rewards/rejected": 0.003525964915752411, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": -0.23894211649894714, | |
| "logits/rejected": -0.18848392367362976, | |
| "logps/chosen": -248.10400390625, | |
| "logps/rejected": -285.8645935058594, | |
| "loss": 0.2031, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.01616227999329567, | |
| "rewards/margins": 0.007524468004703522, | |
| "rewards/rejected": 0.008637811988592148, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": -0.22787991166114807, | |
| "logits/rejected": -0.20557613670825958, | |
| "logps/chosen": -230.24072265625, | |
| "logps/rejected": -255.6363983154297, | |
| "loss": 0.2083, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.011957610957324505, | |
| "rewards/margins": 0.006479016039520502, | |
| "rewards/rejected": 0.005478595849126577, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": -0.2212837040424347, | |
| "logits/rejected": -0.19795656204223633, | |
| "logps/chosen": -331.7243347167969, | |
| "logps/rejected": -322.8146667480469, | |
| "loss": 0.2026, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.006065345369279385, | |
| "rewards/margins": 0.00012618684559129179, | |
| "rewards/rejected": 0.005939158145338297, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.23584437370300293, | |
| "logits/rejected": -0.22455132007598877, | |
| "logps/chosen": -276.896484375, | |
| "logps/rejected": -315.27154541015625, | |
| "loss": 0.2034, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.009026501327753067, | |
| "rewards/margins": 0.0023224998731166124, | |
| "rewards/rejected": 0.006704001221805811, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": -0.20962193608283997, | |
| "logits/rejected": -0.20796938240528107, | |
| "logps/chosen": -241.4213409423828, | |
| "logps/rejected": -284.7034606933594, | |
| "loss": 0.2051, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.012364699505269527, | |
| "rewards/margins": 0.010054551064968109, | |
| "rewards/rejected": 0.0023101491387933493, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": -0.2906285524368286, | |
| "logits/rejected": -0.24792306125164032, | |
| "logps/chosen": -259.7020568847656, | |
| "logps/rejected": -283.08673095703125, | |
| "loss": 0.204, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.008635065518319607, | |
| "rewards/margins": 0.004379413556307554, | |
| "rewards/rejected": 0.004255651496350765, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": -0.3118959367275238, | |
| "logits/rejected": -0.25501811504364014, | |
| "logps/chosen": -276.6239013671875, | |
| "logps/rejected": -296.7880554199219, | |
| "loss": 0.2075, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": 0.014445966109633446, | |
| "rewards/margins": 0.011028407141566277, | |
| "rewards/rejected": 0.0034175582695752382, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": -0.31071215867996216, | |
| "logits/rejected": -0.27819615602493286, | |
| "logps/chosen": -312.6679992675781, | |
| "logps/rejected": -342.43603515625, | |
| "loss": 0.216, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.008847331628203392, | |
| "rewards/margins": 0.008721059188246727, | |
| "rewards/rejected": 0.00012627345859073102, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": -0.291133314371109, | |
| "logits/rejected": -0.31414347887039185, | |
| "logps/chosen": -303.83502197265625, | |
| "logps/rejected": -353.6219482421875, | |
| "loss": 0.2042, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.009907987900078297, | |
| "rewards/margins": 0.008294956758618355, | |
| "rewards/rejected": 0.0016130302101373672, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": -0.32893937826156616, | |
| "logits/rejected": -0.34056320786476135, | |
| "logps/chosen": -279.0770263671875, | |
| "logps/rejected": -284.71649169921875, | |
| "loss": 0.2059, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.0037155388854444027, | |
| "rewards/margins": 0.0032048276625573635, | |
| "rewards/rejected": 0.000510711339302361, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": -0.41928038001060486, | |
| "logits/rejected": -0.3275998830795288, | |
| "logps/chosen": -322.4991149902344, | |
| "logps/rejected": -333.2740478515625, | |
| "loss": 0.2, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.005991303361952305, | |
| "rewards/margins": 0.004061755258589983, | |
| "rewards/rejected": 0.0019295478705316782, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": -0.382260799407959, | |
| "logits/rejected": -0.40049606561660767, | |
| "logps/chosen": -250.11538696289062, | |
| "logps/rejected": -292.5018615722656, | |
| "loss": 0.2112, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.003382362425327301, | |
| "rewards/margins": 0.007972141727805138, | |
| "rewards/rejected": -0.004589778836816549, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": -0.38077178597450256, | |
| "logits/rejected": -0.3729470372200012, | |
| "logps/chosen": -257.94268798828125, | |
| "logps/rejected": -269.142822265625, | |
| "loss": 0.2012, | |
| "rewards/accuracies": 0.21250000596046448, | |
| "rewards/chosen": 0.0004567865398712456, | |
| "rewards/margins": 0.0030794304329901934, | |
| "rewards/rejected": -0.0026226439513266087, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": -0.3339672386646271, | |
| "logits/rejected": -0.32460784912109375, | |
| "logps/chosen": -285.16754150390625, | |
| "logps/rejected": -296.7590026855469, | |
| "loss": 0.2119, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": -0.0008748549735173583, | |
| "rewards/margins": 0.006057094782590866, | |
| "rewards/rejected": -0.0069319503381848335, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.3941500782966614, | |
| "logits/rejected": -0.3795050084590912, | |
| "logps/chosen": -330.65777587890625, | |
| "logps/rejected": -311.400146484375, | |
| "loss": 0.2109, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": -0.001492818701080978, | |
| "rewards/margins": 0.0023579366970807314, | |
| "rewards/rejected": -0.003850755747407675, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": -0.34223705530166626, | |
| "logits/rejected": -0.3617275357246399, | |
| "logps/chosen": -322.3763122558594, | |
| "logps/rejected": -316.99615478515625, | |
| "loss": 0.2077, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.0017126466846093535, | |
| "rewards/margins": 0.003889720421284437, | |
| "rewards/rejected": -0.005602366756647825, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": -0.34621626138687134, | |
| "logits/rejected": -0.35351529717445374, | |
| "logps/chosen": -324.47442626953125, | |
| "logps/rejected": -359.94256591796875, | |
| "loss": 0.2096, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.002571218181401491, | |
| "rewards/margins": 0.00809338316321373, | |
| "rewards/rejected": -0.010664600878953934, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": -0.38271594047546387, | |
| "logits/rejected": -0.3791079521179199, | |
| "logps/chosen": -278.4394226074219, | |
| "logps/rejected": -310.6518859863281, | |
| "loss": 0.2058, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.0030440581031143665, | |
| "rewards/margins": 0.011548629961907864, | |
| "rewards/rejected": -0.00850457139313221, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": -0.3675210475921631, | |
| "logits/rejected": -0.4030851423740387, | |
| "logps/chosen": -281.1656494140625, | |
| "logps/rejected": -319.8685607910156, | |
| "loss": 0.2095, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.0020102262496948242, | |
| "rewards/margins": 0.010647585615515709, | |
| "rewards/rejected": -0.00863736029714346, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": -0.3838180601596832, | |
| "logits/rejected": -0.36158618330955505, | |
| "logps/chosen": -293.8908386230469, | |
| "logps/rejected": -322.5982360839844, | |
| "loss": 0.2055, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.003357111709192395, | |
| "rewards/margins": 0.009485922753810883, | |
| "rewards/rejected": -0.01284303329885006, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.3621710240840912, | |
| "logits/rejected": -0.3666265606880188, | |
| "logps/chosen": -317.04534912109375, | |
| "logps/rejected": -355.4456481933594, | |
| "loss": 0.2086, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.0012256987392902374, | |
| "rewards/margins": 0.009772216901183128, | |
| "rewards/rejected": -0.010997915640473366, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": -0.38925933837890625, | |
| "logits/rejected": -0.34841951727867126, | |
| "logps/chosen": -233.68087768554688, | |
| "logps/rejected": -276.94183349609375, | |
| "loss": 0.2102, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.0036917359102517366, | |
| "rewards/margins": 0.009330306202173233, | |
| "rewards/rejected": -0.005638569127768278, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": -0.42922061681747437, | |
| "logits/rejected": -0.37054792046546936, | |
| "logps/chosen": -305.6702880859375, | |
| "logps/rejected": -336.8018798828125, | |
| "loss": 0.2114, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.0010867089731618762, | |
| "rewards/margins": 0.007067729718983173, | |
| "rewards/rejected": -0.005981021095067263, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": -0.36909401416778564, | |
| "logits/rejected": -0.3725087344646454, | |
| "logps/chosen": -314.70892333984375, | |
| "logps/rejected": -341.8924255371094, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": -0.0023520714603364468, | |
| "rewards/margins": 0.00565011752769351, | |
| "rewards/rejected": -0.008002189919352531, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": -0.3493804633617401, | |
| "logits/rejected": -0.36864691972732544, | |
| "logps/chosen": -318.31158447265625, | |
| "logps/rejected": -329.7109680175781, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.004697232972830534, | |
| "rewards/margins": 0.0050050243735313416, | |
| "rewards/rejected": -0.009702257812023163, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": -0.3421717882156372, | |
| "logits/rejected": -0.3551585376262665, | |
| "logps/chosen": -284.5762939453125, | |
| "logps/rejected": -303.1432189941406, | |
| "loss": 0.2021, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.006993536837399006, | |
| "rewards/margins": 0.010174809023737907, | |
| "rewards/rejected": -0.017168346792459488, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": -0.3510192930698395, | |
| "logits/rejected": -0.351273775100708, | |
| "logps/chosen": -283.2718505859375, | |
| "logps/rejected": -298.7680358886719, | |
| "loss": 0.2054, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.001104214577935636, | |
| "rewards/margins": 0.008227458223700523, | |
| "rewards/rejected": -0.00933167152106762, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": -0.3711737394332886, | |
| "logits/rejected": -0.35552793741226196, | |
| "logps/chosen": -317.1517028808594, | |
| "logps/rejected": -300.7435607910156, | |
| "loss": 0.2134, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.0016167514258995652, | |
| "rewards/margins": -0.0006161117926239967, | |
| "rewards/rejected": -0.0010006397496908903, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": -0.382574200630188, | |
| "logits/rejected": -0.37562495470046997, | |
| "logps/chosen": -311.616943359375, | |
| "logps/rejected": -327.84442138671875, | |
| "loss": 0.203, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.003906398545950651, | |
| "rewards/margins": 0.0055776028893888, | |
| "rewards/rejected": -0.009484002366662025, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": -0.3587433695793152, | |
| "logits/rejected": -0.3470574915409088, | |
| "logps/chosen": -262.9179382324219, | |
| "logps/rejected": -290.76214599609375, | |
| "loss": 0.2072, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.0015628261025995016, | |
| "rewards/margins": 0.004087894223630428, | |
| "rewards/rejected": -0.005650720559060574, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": -0.3712787926197052, | |
| "logits/rejected": -0.3702451288700104, | |
| "logps/chosen": -261.8824768066406, | |
| "logps/rejected": -319.24224853515625, | |
| "loss": 0.2093, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": 0.0052637443877756596, | |
| "rewards/margins": 0.014261695556342602, | |
| "rewards/rejected": -0.008997952565550804, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": -0.32240134477615356, | |
| "logits/rejected": -0.37527790665626526, | |
| "logps/chosen": -248.02835083007812, | |
| "logps/rejected": -285.5946350097656, | |
| "loss": 0.1976, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.0007311357767321169, | |
| "rewards/margins": 0.01329697947949171, | |
| "rewards/rejected": -0.012565843760967255, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": -0.37857678532600403, | |
| "logits/rejected": -0.35946762561798096, | |
| "logps/chosen": -203.0977020263672, | |
| "logps/rejected": -246.66049194335938, | |
| "loss": 0.1988, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.0019016917794942856, | |
| "rewards/margins": 0.00954529084265232, | |
| "rewards/rejected": -0.007643598131835461, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": -0.3962993621826172, | |
| "logits/rejected": -0.4342166781425476, | |
| "logps/chosen": -274.3848571777344, | |
| "logps/rejected": -307.4154357910156, | |
| "loss": 0.2018, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.004304953850805759, | |
| "rewards/margins": 0.008778423070907593, | |
| "rewards/rejected": -0.004473468754440546, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": -0.36348485946655273, | |
| "logits/rejected": -0.3567599058151245, | |
| "logps/chosen": -286.14520263671875, | |
| "logps/rejected": -310.6136474609375, | |
| "loss": 0.1949, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.0005280302138999104, | |
| "rewards/margins": 0.006353036966174841, | |
| "rewards/rejected": -0.006881065666675568, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": -0.3500184118747711, | |
| "logits/rejected": -0.342162549495697, | |
| "logps/chosen": -286.3838806152344, | |
| "logps/rejected": -341.6246337890625, | |
| "loss": 0.2074, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.00207870383746922, | |
| "rewards/margins": 0.011935941874980927, | |
| "rewards/rejected": -0.014014643616974354, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": -0.39309799671173096, | |
| "logits/rejected": -0.33321046829223633, | |
| "logps/chosen": -252.35006713867188, | |
| "logps/rejected": -278.26580810546875, | |
| "loss": 0.2062, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.001301859738305211, | |
| "rewards/margins": 0.009171300567686558, | |
| "rewards/rejected": -0.00786944106221199, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": -0.38764262199401855, | |
| "logits/rejected": -0.3758700489997864, | |
| "logps/chosen": -274.4250183105469, | |
| "logps/rejected": -305.24371337890625, | |
| "loss": 0.2028, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.004850239958614111, | |
| "rewards/margins": 0.006510905921459198, | |
| "rewards/rejected": -0.011361144483089447, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": -0.35581859946250916, | |
| "logits/rejected": -0.36488813161849976, | |
| "logps/chosen": -298.48529052734375, | |
| "logps/rejected": -312.73004150390625, | |
| "loss": 0.2046, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.0006998656317591667, | |
| "rewards/margins": 0.0070778196677565575, | |
| "rewards/rejected": -0.007777686230838299, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": -0.3638850152492523, | |
| "logits/rejected": -0.3694193363189697, | |
| "logps/chosen": -304.8964538574219, | |
| "logps/rejected": -322.5799560546875, | |
| "loss": 0.1952, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 4.08724881708622e-05, | |
| "rewards/margins": 0.00868744682520628, | |
| "rewards/rejected": -0.008646574802696705, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": -0.40093159675598145, | |
| "logits/rejected": -0.35080868005752563, | |
| "logps/chosen": -306.1406555175781, | |
| "logps/rejected": -354.2425842285156, | |
| "loss": 0.2078, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.0017454406479373574, | |
| "rewards/margins": 0.01031532697379589, | |
| "rewards/rejected": -0.012060767039656639, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": -0.4077305793762207, | |
| "logits/rejected": -0.3492675721645355, | |
| "logps/chosen": -302.4758605957031, | |
| "logps/rejected": -299.92266845703125, | |
| "loss": 0.2094, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": -0.0018849109765142202, | |
| "rewards/margins": 0.002199392067268491, | |
| "rewards/rejected": -0.004084303043782711, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": -0.35477060079574585, | |
| "logits/rejected": -0.3815353512763977, | |
| "logps/chosen": -273.3544006347656, | |
| "logps/rejected": -309.14599609375, | |
| "loss": 0.2057, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.0007420766050927341, | |
| "rewards/margins": 0.01061691902577877, | |
| "rewards/rejected": -0.009874843060970306, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.029720235659702103, | |
| "train_runtime": 1648.0199, | |
| "train_samples_per_second": 18.204, | |
| "train_steps_per_second": 0.569 | |
| } | |
| ], | |
| "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 | |
| } | |