| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.5165289256198347, | |
| "eval_steps": 500, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.000000000000001e-07, | |
| "logits/chosen": -2.103288412094116, | |
| "logits/rejected": -2.1592373847961426, | |
| "logps/chosen": -170.5673828125, | |
| "logps/rejected": -191.9125213623047, | |
| "loss": 0.6936, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.005827450659126043, | |
| "rewards/margins": -9.014667011797428e-05, | |
| "rewards/rejected": 0.005917596165090799, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "logits/chosen": -2.0918476581573486, | |
| "logits/rejected": -2.176135540008545, | |
| "logps/chosen": -155.1693115234375, | |
| "logps/rejected": -168.37811279296875, | |
| "loss": 0.6902, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.008325956761837006, | |
| "rewards/margins": 0.006542182061821222, | |
| "rewards/rejected": 0.0017837753985077143, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.5e-06, | |
| "logits/chosen": -2.2100210189819336, | |
| "logits/rejected": -2.1960439682006836, | |
| "logps/chosen": -179.06021118164062, | |
| "logps/rejected": -204.23802185058594, | |
| "loss": 0.696, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.005401517264544964, | |
| "rewards/margins": -0.004398156888782978, | |
| "rewards/rejected": -0.0010033606085926294, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "logits/chosen": -2.2304933071136475, | |
| "logits/rejected": -2.2267584800720215, | |
| "logps/chosen": -192.56463623046875, | |
| "logps/rejected": -182.64608764648438, | |
| "loss": 0.6873, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.025449421256780624, | |
| "rewards/margins": 0.013228798285126686, | |
| "rewards/rejected": 0.012220621109008789, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.5e-06, | |
| "logits/chosen": -2.280231475830078, | |
| "logits/rejected": -2.241490125656128, | |
| "logps/chosen": -113.12672424316406, | |
| "logps/rejected": -117.45411682128906, | |
| "loss": 0.6853, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.007502961438149214, | |
| "rewards/margins": 0.016341591253876686, | |
| "rewards/rejected": -0.008838629350066185, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3e-06, | |
| "logits/chosen": -2.2110278606414795, | |
| "logits/rejected": -2.180075168609619, | |
| "logps/chosen": -173.43307495117188, | |
| "logps/rejected": -165.2748260498047, | |
| "loss": 0.6762, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.0310593843460083, | |
| "rewards/margins": 0.03597269207239151, | |
| "rewards/rejected": -0.00491330586373806, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.5000000000000004e-06, | |
| "logits/chosen": -2.0283565521240234, | |
| "logits/rejected": -2.079817295074463, | |
| "logps/chosen": -125.85430145263672, | |
| "logps/rejected": -131.1006317138672, | |
| "loss": 0.7011, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.018836330622434616, | |
| "rewards/margins": -0.014867208898067474, | |
| "rewards/rejected": -0.003969121258705854, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.000000000000001e-06, | |
| "logits/chosen": -2.197159767150879, | |
| "logits/rejected": -2.2359261512756348, | |
| "logps/chosen": -197.13482666015625, | |
| "logps/rejected": -206.99850463867188, | |
| "loss": 0.6962, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.008057355880737305, | |
| "rewards/margins": -0.004953860770910978, | |
| "rewards/rejected": -0.0031034951098263264, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5e-06, | |
| "logits/chosen": -2.043184518814087, | |
| "logits/rejected": -2.0166640281677246, | |
| "logps/chosen": -153.27444458007812, | |
| "logps/rejected": -155.04905700683594, | |
| "loss": 0.6957, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.008291959762573242, | |
| "rewards/margins": -0.004631852265447378, | |
| "rewards/rejected": 0.012923812493681908, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-06, | |
| "logits/chosen": -2.0842106342315674, | |
| "logits/rejected": -2.0971806049346924, | |
| "logps/chosen": -149.14915466308594, | |
| "logps/rejected": -148.91160583496094, | |
| "loss": 0.705, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.01141366921365261, | |
| "rewards/margins": -0.022213315591216087, | |
| "rewards/rejected": 0.010799647308886051, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.500000000000001e-06, | |
| "logits/chosen": -2.235450506210327, | |
| "logits/rejected": -2.1357128620147705, | |
| "logps/chosen": -197.30487060546875, | |
| "logps/rejected": -172.14366149902344, | |
| "loss": 0.7011, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.013692856766283512, | |
| "rewards/margins": -0.01506800763309002, | |
| "rewards/rejected": 0.0013751504011452198, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6e-06, | |
| "logits/chosen": -2.22318172454834, | |
| "logits/rejected": -2.2130980491638184, | |
| "logps/chosen": -208.3831787109375, | |
| "logps/rejected": -211.73146057128906, | |
| "loss": 0.6788, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.01979823224246502, | |
| "rewards/margins": 0.02999591827392578, | |
| "rewards/rejected": -0.010197687894105911, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "logits/chosen": -2.2513983249664307, | |
| "logits/rejected": -2.197139263153076, | |
| "logps/chosen": -203.53457641601562, | |
| "logps/rejected": -192.02703857421875, | |
| "loss": 0.7039, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.00548624899238348, | |
| "rewards/margins": -0.02056746557354927, | |
| "rewards/rejected": 0.015081215649843216, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.000000000000001e-06, | |
| "logits/chosen": -2.172175407409668, | |
| "logits/rejected": -2.323552370071411, | |
| "logps/chosen": -149.00196838378906, | |
| "logps/rejected": -205.64434814453125, | |
| "loss": 0.71, | |
| "rewards/accuracies": 0.1875, | |
| "rewards/chosen": -0.03644166141748428, | |
| "rewards/margins": -0.033078860491514206, | |
| "rewards/rejected": -0.003362798597663641, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.5e-06, | |
| "logits/chosen": -2.2747740745544434, | |
| "logits/rejected": -2.30997371673584, | |
| "logps/chosen": -173.9440460205078, | |
| "logps/rejected": -152.3206787109375, | |
| "loss": 0.6856, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.02293362468481064, | |
| "rewards/margins": 0.016007710248231888, | |
| "rewards/rejected": 0.006925915367901325, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.000000000000001e-06, | |
| "logits/chosen": -2.3504209518432617, | |
| "logits/rejected": -2.2377350330352783, | |
| "logps/chosen": -176.5444793701172, | |
| "logps/rejected": -157.48834228515625, | |
| "loss": 0.6908, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.02712235413491726, | |
| "rewards/margins": 0.00521473865956068, | |
| "rewards/rejected": 0.021907614544034004, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.500000000000002e-06, | |
| "logits/chosen": -2.120563507080078, | |
| "logits/rejected": -2.0712292194366455, | |
| "logps/chosen": -207.5531005859375, | |
| "logps/rejected": -254.9108123779297, | |
| "loss": 0.6819, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.020979978144168854, | |
| "rewards/margins": 0.023171592503786087, | |
| "rewards/rejected": -0.0021916148252785206, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9e-06, | |
| "logits/chosen": -2.123784065246582, | |
| "logits/rejected": -2.1854984760284424, | |
| "logps/chosen": -168.7614288330078, | |
| "logps/rejected": -187.71261596679688, | |
| "loss": 0.7007, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.0059951068833470345, | |
| "rewards/margins": -0.013646721839904785, | |
| "rewards/rejected": 0.007651614956557751, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.5e-06, | |
| "logits/chosen": -2.0693728923797607, | |
| "logits/rejected": -2.1303658485412598, | |
| "logps/chosen": -166.02944946289062, | |
| "logps/rejected": -198.49871826171875, | |
| "loss": 0.6946, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.013394879177212715, | |
| "rewards/margins": -0.0023463736288249493, | |
| "rewards/rejected": 0.01574125327169895, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1e-05, | |
| "logits/chosen": -2.248774528503418, | |
| "logits/rejected": -2.2679812908172607, | |
| "logps/chosen": -155.03526306152344, | |
| "logps/rejected": -168.04348754882812, | |
| "loss": 0.6887, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.013049841858446598, | |
| "rewards/margins": 0.009325623512268066, | |
| "rewards/rejected": 0.0037242178805172443, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.05e-05, | |
| "logits/chosen": -2.054776906967163, | |
| "logits/rejected": -1.9716964960098267, | |
| "logps/chosen": -187.37937927246094, | |
| "logps/rejected": -171.3694305419922, | |
| "loss": 0.677, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.010295724496245384, | |
| "rewards/margins": 0.03319711610674858, | |
| "rewards/rejected": -0.022901393473148346, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "logits/chosen": -2.0806198120117188, | |
| "logits/rejected": -2.0407967567443848, | |
| "logps/chosen": -179.2973175048828, | |
| "logps/rejected": -150.33853149414062, | |
| "loss": 0.6918, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.016437100246548653, | |
| "rewards/margins": 0.0046033854596316814, | |
| "rewards/rejected": 0.011833716183900833, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.1500000000000002e-05, | |
| "logits/chosen": -2.104949474334717, | |
| "logits/rejected": -2.222316265106201, | |
| "logps/chosen": -137.92550659179688, | |
| "logps/rejected": -148.96446228027344, | |
| "loss": 0.6826, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.020548105239868164, | |
| "rewards/margins": 0.02232079580426216, | |
| "rewards/rejected": -0.0017726896330714226, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2e-05, | |
| "logits/chosen": -2.3451733589172363, | |
| "logits/rejected": -2.267653226852417, | |
| "logps/chosen": -170.1451416015625, | |
| "logps/rejected": -150.92538452148438, | |
| "loss": 0.6952, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.007013177499175072, | |
| "rewards/margins": -0.002862358232960105, | |
| "rewards/rejected": 0.00987553596496582, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.25e-05, | |
| "logits/chosen": -1.9284687042236328, | |
| "logits/rejected": -2.018852710723877, | |
| "logps/chosen": -108.83004760742188, | |
| "logps/rejected": -138.14512634277344, | |
| "loss": 0.6933, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.0071326494216918945, | |
| "rewards/margins": 0.0002952106297016144, | |
| "rewards/rejected": -0.007427859120070934, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "logits/chosen": -2.2073161602020264, | |
| "logits/rejected": -2.1905336380004883, | |
| "logps/chosen": -209.46640014648438, | |
| "logps/rejected": -205.64620971679688, | |
| "loss": 0.701, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.0032413718290627003, | |
| "rewards/margins": -0.01478672120720148, | |
| "rewards/rejected": 0.011545348912477493, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "logits/chosen": -2.0893516540527344, | |
| "logits/rejected": -2.224111557006836, | |
| "logps/chosen": -166.69960021972656, | |
| "logps/rejected": -180.09710693359375, | |
| "loss": 0.6964, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.004743862431496382, | |
| "rewards/margins": -0.0056123011745512486, | |
| "rewards/rejected": 0.0008684396743774414, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "logits/chosen": -2.149770975112915, | |
| "logits/rejected": -2.1601545810699463, | |
| "logps/chosen": -136.2816162109375, | |
| "logps/rejected": -161.84671020507812, | |
| "loss": 0.7068, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.017784714698791504, | |
| "rewards/margins": -0.026406075805425644, | |
| "rewards/rejected": 0.00862135924398899, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.45e-05, | |
| "logits/chosen": -1.9762641191482544, | |
| "logits/rejected": -2.0807995796203613, | |
| "logps/chosen": -149.5594940185547, | |
| "logps/rejected": -172.4080810546875, | |
| "loss": 0.6758, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.012678837403655052, | |
| "rewards/margins": 0.03670220449566841, | |
| "rewards/rejected": -0.02402336895465851, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5e-05, | |
| "logits/chosen": -2.112738609313965, | |
| "logits/rejected": -2.120100736618042, | |
| "logps/chosen": -174.06729125976562, | |
| "logps/rejected": -162.5001983642578, | |
| "loss": 0.6923, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.008127642795443535, | |
| "rewards/margins": 0.0025386326014995575, | |
| "rewards/rejected": -0.010666273534297943, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.55e-05, | |
| "logits/chosen": -2.1441712379455566, | |
| "logits/rejected": -2.1569695472717285, | |
| "logps/chosen": -206.64749145507812, | |
| "logps/rejected": -226.67657470703125, | |
| "loss": 0.7127, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.031477928161621094, | |
| "rewards/margins": -0.03634142875671387, | |
| "rewards/rejected": 0.004863499663770199, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "logits/chosen": -2.174954891204834, | |
| "logits/rejected": -2.2214908599853516, | |
| "logps/chosen": -132.75070190429688, | |
| "logps/rejected": -137.4167022705078, | |
| "loss": 0.6907, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.00288999080657959, | |
| "rewards/margins": 0.006665444932878017, | |
| "rewards/rejected": -0.009555434808135033, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.65e-05, | |
| "logits/chosen": -2.0139670372009277, | |
| "logits/rejected": -2.1149113178253174, | |
| "logps/chosen": -125.44695281982422, | |
| "logps/rejected": -136.34800720214844, | |
| "loss": 0.7071, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.0032261379528790712, | |
| "rewards/margins": -0.02571232244372368, | |
| "rewards/rejected": 0.022486185654997826, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "logits/chosen": -2.151669979095459, | |
| "logits/rejected": -2.225001335144043, | |
| "logps/chosen": -142.0449981689453, | |
| "logps/rejected": -152.09034729003906, | |
| "loss": 0.6993, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.00544746033847332, | |
| "rewards/margins": -0.011389017105102539, | |
| "rewards/rejected": 0.005941559094935656, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.75e-05, | |
| "logits/chosen": -2.112931251525879, | |
| "logits/rejected": -2.1539905071258545, | |
| "logps/chosen": -140.09375, | |
| "logps/rejected": -163.06671142578125, | |
| "loss": 0.6858, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.02165074460208416, | |
| "rewards/margins": 0.016152428463101387, | |
| "rewards/rejected": 0.005498316138982773, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.8e-05, | |
| "logits/chosen": -2.1201136112213135, | |
| "logits/rejected": -2.212294816970825, | |
| "logps/chosen": -142.1851348876953, | |
| "logps/rejected": -161.80767822265625, | |
| "loss": 0.6967, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.007724761962890625, | |
| "rewards/margins": -0.0059874774888157845, | |
| "rewards/rejected": -0.0017372854053974152, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.85e-05, | |
| "logits/chosen": -2.0521628856658936, | |
| "logits/rejected": -2.2256691455841064, | |
| "logps/chosen": -149.46946716308594, | |
| "logps/rejected": -218.28085327148438, | |
| "loss": 0.6829, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.01561835128813982, | |
| "rewards/margins": 0.026017257943749428, | |
| "rewards/rejected": -0.041635606437921524, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9e-05, | |
| "logits/chosen": -2.1336724758148193, | |
| "logits/rejected": -2.129721164703369, | |
| "logps/chosen": -149.40090942382812, | |
| "logps/rejected": -166.29310607910156, | |
| "loss": 0.7004, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.05927868187427521, | |
| "rewards/margins": -0.012753820046782494, | |
| "rewards/rejected": -0.046524859964847565, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9500000000000003e-05, | |
| "logits/chosen": -2.2073593139648438, | |
| "logits/rejected": -2.2479546070098877, | |
| "logps/chosen": -172.16322326660156, | |
| "logps/rejected": -165.34413146972656, | |
| "loss": 0.6911, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.009036087431013584, | |
| "rewards/margins": 0.0075013842433691025, | |
| "rewards/rejected": -0.01653747446835041, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2e-05, | |
| "logits/chosen": -2.0156097412109375, | |
| "logits/rejected": -1.9676055908203125, | |
| "logps/chosen": -214.78041076660156, | |
| "logps/rejected": -248.21835327148438, | |
| "loss": 0.6774, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.03332924842834473, | |
| "rewards/margins": 0.03387260064482689, | |
| "rewards/rejected": -0.06720185279846191, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.05e-05, | |
| "logits/chosen": -2.1184558868408203, | |
| "logits/rejected": -2.127263069152832, | |
| "logps/chosen": -205.9291534423828, | |
| "logps/rejected": -200.1770782470703, | |
| "loss": 0.694, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.048839375376701355, | |
| "rewards/margins": 0.00204639695584774, | |
| "rewards/rejected": -0.050885774195194244, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1e-05, | |
| "logits/chosen": -2.094801902770996, | |
| "logits/rejected": -2.121784210205078, | |
| "logps/chosen": -175.4020538330078, | |
| "logps/rejected": -171.9228973388672, | |
| "loss": 0.6745, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.0025298111140727997, | |
| "rewards/margins": 0.04258604347705841, | |
| "rewards/rejected": -0.04511585459113121, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.15e-05, | |
| "logits/chosen": -2.1637444496154785, | |
| "logits/rejected": -2.254055976867676, | |
| "logps/chosen": -172.897216796875, | |
| "logps/rejected": -212.38848876953125, | |
| "loss": 0.6949, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.06465272605419159, | |
| "rewards/margins": 0.0003746009897440672, | |
| "rewards/rejected": -0.06502733379602432, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "logits/chosen": -2.0817313194274902, | |
| "logits/rejected": -2.127499580383301, | |
| "logps/chosen": -149.39105224609375, | |
| "logps/rejected": -151.25680541992188, | |
| "loss": 0.7009, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.054700277745723724, | |
| "rewards/margins": -0.01434707548469305, | |
| "rewards/rejected": -0.0403532050549984, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.25e-05, | |
| "logits/chosen": -2.2211873531341553, | |
| "logits/rejected": -2.1942520141601562, | |
| "logps/chosen": -136.8941650390625, | |
| "logps/rejected": -131.68629455566406, | |
| "loss": 0.6864, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.06061294302344322, | |
| "rewards/margins": 0.01509325485676527, | |
| "rewards/rejected": -0.07570619881153107, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "logits/chosen": -2.0270614624023438, | |
| "logits/rejected": -2.0136606693267822, | |
| "logps/chosen": -154.61752319335938, | |
| "logps/rejected": -159.65757751464844, | |
| "loss": 0.6724, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.04580743610858917, | |
| "rewards/margins": 0.04787466675043106, | |
| "rewards/rejected": -0.09368209540843964, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.35e-05, | |
| "logits/chosen": -2.0510787963867188, | |
| "logits/rejected": -2.1365280151367188, | |
| "logps/chosen": -160.6487274169922, | |
| "logps/rejected": -154.50369262695312, | |
| "loss": 0.6853, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.056206539273262024, | |
| "rewards/margins": 0.01876838132739067, | |
| "rewards/rejected": -0.0749749168753624, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.4e-05, | |
| "logits/chosen": -2.1573550701141357, | |
| "logits/rejected": -2.0711829662323, | |
| "logps/chosen": -166.64385986328125, | |
| "logps/rejected": -196.8815155029297, | |
| "loss": 0.7056, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.09673108905553818, | |
| "rewards/margins": -0.02166934125125408, | |
| "rewards/rejected": -0.07506175339221954, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.45e-05, | |
| "logits/chosen": -2.1508450508117676, | |
| "logits/rejected": -2.1660666465759277, | |
| "logps/chosen": -178.22775268554688, | |
| "logps/rejected": -196.37791442871094, | |
| "loss": 0.6902, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.12391133606433868, | |
| "rewards/margins": 0.008201269432902336, | |
| "rewards/rejected": -0.13211259245872498, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.5e-05, | |
| "logits/chosen": -2.102323532104492, | |
| "logits/rejected": -2.1379528045654297, | |
| "logps/chosen": -157.07907104492188, | |
| "logps/rejected": -160.67617797851562, | |
| "loss": 0.6855, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.07163415104150772, | |
| "rewards/margins": 0.018704529851675034, | |
| "rewards/rejected": -0.09033869206905365, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.5500000000000003e-05, | |
| "logits/chosen": -2.1440117359161377, | |
| "logits/rejected": -2.1146342754364014, | |
| "logps/chosen": -176.24920654296875, | |
| "logps/rejected": -171.60125732421875, | |
| "loss": 0.6785, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.07399649173021317, | |
| "rewards/margins": 0.03202385827898979, | |
| "rewards/rejected": -0.10602036118507385, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "logits/chosen": -2.1452462673187256, | |
| "logits/rejected": -2.1233859062194824, | |
| "logps/chosen": -152.2857208251953, | |
| "logps/rejected": -143.5636444091797, | |
| "loss": 0.7112, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.09882183372974396, | |
| "rewards/margins": -0.03147630766034126, | |
| "rewards/rejected": -0.0673455223441124, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6500000000000004e-05, | |
| "logits/chosen": -2.1651771068573, | |
| "logits/rejected": -2.1493921279907227, | |
| "logps/chosen": -183.37535095214844, | |
| "logps/rejected": -185.5274658203125, | |
| "loss": 0.7173, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.1233779639005661, | |
| "rewards/margins": -0.04217071086168289, | |
| "rewards/rejected": -0.08120725303888321, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "logits/chosen": -2.171654462814331, | |
| "logits/rejected": -2.1836743354797363, | |
| "logps/chosen": -164.41943359375, | |
| "logps/rejected": -169.02845764160156, | |
| "loss": 0.7049, | |
| "rewards/accuracies": 0.1875, | |
| "rewards/chosen": -0.1076451763510704, | |
| "rewards/margins": -0.021807147189974785, | |
| "rewards/rejected": -0.08583803474903107, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.7500000000000004e-05, | |
| "logits/chosen": -2.255077362060547, | |
| "logits/rejected": -2.2395968437194824, | |
| "logps/chosen": -181.29159545898438, | |
| "logps/rejected": -181.9247283935547, | |
| "loss": 0.7198, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.11303920298814774, | |
| "rewards/margins": -0.047478508204221725, | |
| "rewards/rejected": -0.06556069850921631, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "logits/chosen": -2.1476848125457764, | |
| "logits/rejected": -2.1060516834259033, | |
| "logps/chosen": -186.89280700683594, | |
| "logps/rejected": -210.42250061035156, | |
| "loss": 0.6694, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.10679897665977478, | |
| "rewards/margins": 0.051741939038038254, | |
| "rewards/rejected": -0.15854091942310333, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "logits/chosen": -2.1386842727661133, | |
| "logits/rejected": -2.120420455932617, | |
| "logps/chosen": -151.2351531982422, | |
| "logps/rejected": -141.44479370117188, | |
| "loss": 0.689, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.10497517883777618, | |
| "rewards/margins": 0.013038091361522675, | |
| "rewards/rejected": -0.11801326274871826, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.9e-05, | |
| "logits/chosen": -2.047849416732788, | |
| "logits/rejected": -2.1129918098449707, | |
| "logps/chosen": -132.04562377929688, | |
| "logps/rejected": -172.14659118652344, | |
| "loss": 0.6764, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.09353986382484436, | |
| "rewards/margins": 0.0432942658662796, | |
| "rewards/rejected": -0.13683411478996277, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.95e-05, | |
| "logits/chosen": -2.0588788986206055, | |
| "logits/rejected": -2.0961172580718994, | |
| "logps/chosen": -153.62033081054688, | |
| "logps/rejected": -163.29586791992188, | |
| "loss": 0.6737, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.128899484872818, | |
| "rewards/margins": 0.04542210325598717, | |
| "rewards/rejected": -0.17432159185409546, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3e-05, | |
| "logits/chosen": -2.1147661209106445, | |
| "logits/rejected": -2.2322981357574463, | |
| "logps/chosen": -127.71258544921875, | |
| "logps/rejected": -147.38790893554688, | |
| "loss": 0.6874, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.11440041661262512, | |
| "rewards/margins": 0.020626328885555267, | |
| "rewards/rejected": -0.135026752948761, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.05e-05, | |
| "logits/chosen": -2.2986626625061035, | |
| "logits/rejected": -2.1925692558288574, | |
| "logps/chosen": -177.45851135253906, | |
| "logps/rejected": -164.67543029785156, | |
| "loss": 0.6952, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.14733535051345825, | |
| "rewards/margins": 0.000783134251832962, | |
| "rewards/rejected": -0.14811846613883972, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.1e-05, | |
| "logits/chosen": -2.2345943450927734, | |
| "logits/rejected": -2.2452545166015625, | |
| "logps/chosen": -241.3500518798828, | |
| "logps/rejected": -254.5419464111328, | |
| "loss": 0.719, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.20402950048446655, | |
| "rewards/margins": -0.043932244181632996, | |
| "rewards/rejected": -0.16009727120399475, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.15e-05, | |
| "logits/chosen": -1.8611196279525757, | |
| "logits/rejected": -1.9434685707092285, | |
| "logps/chosen": -169.63336181640625, | |
| "logps/rejected": -175.9423065185547, | |
| "loss": 0.6778, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.20865049958229065, | |
| "rewards/margins": 0.0381101630628109, | |
| "rewards/rejected": -0.24676063656806946, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "logits/chosen": -2.072692632675171, | |
| "logits/rejected": -2.011317729949951, | |
| "logps/chosen": -162.03518676757812, | |
| "logps/rejected": -168.4072723388672, | |
| "loss": 0.6419, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.15871694684028625, | |
| "rewards/margins": 0.1158316433429718, | |
| "rewards/rejected": -0.27454859018325806, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.2500000000000004e-05, | |
| "logits/chosen": -1.9928253889083862, | |
| "logits/rejected": -2.0193428993225098, | |
| "logps/chosen": -162.2337646484375, | |
| "logps/rejected": -174.1917724609375, | |
| "loss": 0.7201, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.16654035449028015, | |
| "rewards/margins": -0.048485275357961655, | |
| "rewards/rejected": -0.1180550754070282, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.3e-05, | |
| "logits/chosen": -2.159943103790283, | |
| "logits/rejected": -2.1084258556365967, | |
| "logps/chosen": -278.35736083984375, | |
| "logps/rejected": -280.3484802246094, | |
| "loss": 0.6671, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.29622915387153625, | |
| "rewards/margins": 0.05588364228606224, | |
| "rewards/rejected": -0.3521128296852112, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.35e-05, | |
| "logits/chosen": -2.2655584812164307, | |
| "logits/rejected": -2.2335681915283203, | |
| "logps/chosen": -198.4883575439453, | |
| "logps/rejected": -177.21963500976562, | |
| "loss": 0.6969, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.23526902496814728, | |
| "rewards/margins": 0.006288906559348106, | |
| "rewards/rejected": -0.24155797064304352, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "logits/chosen": -2.07804012298584, | |
| "logits/rejected": -2.1780683994293213, | |
| "logps/chosen": -159.1077423095703, | |
| "logps/rejected": -162.15577697753906, | |
| "loss": 0.6687, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.21029870212078094, | |
| "rewards/margins": 0.062353264540433884, | |
| "rewards/rejected": -0.2726519703865051, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.45e-05, | |
| "logits/chosen": -2.1590423583984375, | |
| "logits/rejected": -2.1173341274261475, | |
| "logps/chosen": -171.05960083007812, | |
| "logps/rejected": -172.3741912841797, | |
| "loss": 0.7086, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.21330544352531433, | |
| "rewards/margins": -0.01934635452926159, | |
| "rewards/rejected": -0.1939591020345688, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.5e-05, | |
| "logits/chosen": -2.1891655921936035, | |
| "logits/rejected": -2.165578603744507, | |
| "logps/chosen": -173.93069458007812, | |
| "logps/rejected": -166.1029510498047, | |
| "loss": 0.6574, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.14292821288108826, | |
| "rewards/margins": 0.08482673764228821, | |
| "rewards/rejected": -0.22775495052337646, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.55e-05, | |
| "logits/chosen": -2.0817506313323975, | |
| "logits/rejected": -2.1412384510040283, | |
| "logps/chosen": -200.84112548828125, | |
| "logps/rejected": -192.84481811523438, | |
| "loss": 0.7133, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.1910141557455063, | |
| "rewards/margins": -0.033721309155225754, | |
| "rewards/rejected": -0.15729284286499023, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.6e-05, | |
| "logits/chosen": -2.2072203159332275, | |
| "logits/rejected": -2.2581570148468018, | |
| "logps/chosen": -159.04298400878906, | |
| "logps/rejected": -169.8282928466797, | |
| "loss": 0.6775, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.27095648646354675, | |
| "rewards/margins": 0.035364869982004166, | |
| "rewards/rejected": -0.30632132291793823, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.65e-05, | |
| "logits/chosen": -2.135335922241211, | |
| "logits/rejected": -2.103652000427246, | |
| "logps/chosen": -185.53627014160156, | |
| "logps/rejected": -185.88861083984375, | |
| "loss": 0.7066, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2883676588535309, | |
| "rewards/margins": -0.014560367912054062, | |
| "rewards/rejected": -0.2738072872161865, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.7e-05, | |
| "logits/chosen": -2.047358751296997, | |
| "logits/rejected": -2.0611729621887207, | |
| "logps/chosen": -165.90042114257812, | |
| "logps/rejected": -171.07815551757812, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.1897939294576645, | |
| "rewards/margins": 0.07893175631761551, | |
| "rewards/rejected": -0.2687256634235382, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "logits/chosen": -2.15533709526062, | |
| "logits/rejected": -2.1560285091400146, | |
| "logps/chosen": -179.14215087890625, | |
| "logps/rejected": -201.46534729003906, | |
| "loss": 0.6959, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2652670443058014, | |
| "rewards/margins": 0.019084777683019638, | |
| "rewards/rejected": -0.28435182571411133, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.8e-05, | |
| "logits/chosen": -1.9366830587387085, | |
| "logits/rejected": -1.94514799118042, | |
| "logps/chosen": -139.78160095214844, | |
| "logps/rejected": -185.41348266601562, | |
| "loss": 0.6614, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.1186312660574913, | |
| "rewards/margins": 0.08789843320846558, | |
| "rewards/rejected": -0.20652970671653748, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.85e-05, | |
| "logits/chosen": -2.075505256652832, | |
| "logits/rejected": -2.181002140045166, | |
| "logps/chosen": -126.95040130615234, | |
| "logps/rejected": -155.3976593017578, | |
| "loss": 0.6595, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.16201543807983398, | |
| "rewards/margins": 0.09039945900440216, | |
| "rewards/rejected": -0.25241488218307495, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "logits/chosen": -2.110365629196167, | |
| "logits/rejected": -2.075228691101074, | |
| "logps/chosen": -150.39419555664062, | |
| "logps/rejected": -141.55215454101562, | |
| "loss": 0.7161, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.18655307590961456, | |
| "rewards/margins": -0.03298550099134445, | |
| "rewards/rejected": -0.15356756746768951, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.9500000000000005e-05, | |
| "logits/chosen": -2.1718859672546387, | |
| "logits/rejected": -2.149282932281494, | |
| "logps/chosen": -161.88609313964844, | |
| "logps/rejected": -151.4117431640625, | |
| "loss": 0.7148, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.22528785467147827, | |
| "rewards/margins": -0.025278834626078606, | |
| "rewards/rejected": -0.2000090330839157, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4e-05, | |
| "logits/chosen": -1.9775056838989258, | |
| "logits/rejected": -2.018721580505371, | |
| "logps/chosen": -158.59869384765625, | |
| "logps/rejected": -157.62229919433594, | |
| "loss": 0.7018, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4199330806732178, | |
| "rewards/margins": -0.005412150174379349, | |
| "rewards/rejected": -0.41452091932296753, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.05e-05, | |
| "logits/chosen": -2.133749485015869, | |
| "logits/rejected": -2.110414981842041, | |
| "logps/chosen": -187.2969970703125, | |
| "logps/rejected": -171.6340789794922, | |
| "loss": 0.7587, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.30970045924186707, | |
| "rewards/margins": -0.09912550449371338, | |
| "rewards/rejected": -0.2105749547481537, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.1e-05, | |
| "logits/chosen": -2.0524086952209473, | |
| "logits/rejected": -2.072053909301758, | |
| "logps/chosen": -153.37196350097656, | |
| "logps/rejected": -169.2631072998047, | |
| "loss": 0.708, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3111407458782196, | |
| "rewards/margins": -0.010960865765810013, | |
| "rewards/rejected": -0.3001798689365387, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.15e-05, | |
| "logits/chosen": -2.322385311126709, | |
| "logits/rejected": -2.243534564971924, | |
| "logps/chosen": -139.49378967285156, | |
| "logps/rejected": -134.7334747314453, | |
| "loss": 0.6372, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.28934672474861145, | |
| "rewards/margins": 0.1364654153585434, | |
| "rewards/rejected": -0.4258121848106384, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.2e-05, | |
| "logits/chosen": -2.071685314178467, | |
| "logits/rejected": -2.143737316131592, | |
| "logps/chosen": -194.6321563720703, | |
| "logps/rejected": -208.46151733398438, | |
| "loss": 0.6989, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.2532384991645813, | |
| "rewards/margins": 0.005628681741654873, | |
| "rewards/rejected": -0.25886717438697815, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.25e-05, | |
| "logits/chosen": -2.1286940574645996, | |
| "logits/rejected": -2.0751113891601562, | |
| "logps/chosen": -139.53895568847656, | |
| "logps/rejected": -159.332763671875, | |
| "loss": 0.7189, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.25435367226600647, | |
| "rewards/margins": -0.02771497517824173, | |
| "rewards/rejected": -0.22663868963718414, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.3e-05, | |
| "logits/chosen": -2.067699670791626, | |
| "logits/rejected": -2.1141469478607178, | |
| "logps/chosen": -149.09188842773438, | |
| "logps/rejected": -143.4701690673828, | |
| "loss": 0.6949, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3173486590385437, | |
| "rewards/margins": 0.01528053916990757, | |
| "rewards/rejected": -0.3326292037963867, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.35e-05, | |
| "logits/chosen": -2.0156338214874268, | |
| "logits/rejected": -1.958490252494812, | |
| "logps/chosen": -150.90170288085938, | |
| "logps/rejected": -148.94139099121094, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.23088550567626953, | |
| "rewards/margins": 0.029866689816117287, | |
| "rewards/rejected": -0.2607521712779999, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "logits/chosen": -2.0316781997680664, | |
| "logits/rejected": -1.9842568635940552, | |
| "logps/chosen": -173.40185546875, | |
| "logps/rejected": -171.54295349121094, | |
| "loss": 0.7099, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.2735195457935333, | |
| "rewards/margins": -0.008416082710027695, | |
| "rewards/rejected": -0.26510342955589294, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.4500000000000004e-05, | |
| "logits/chosen": -2.212313413619995, | |
| "logits/rejected": -2.129784345626831, | |
| "logps/chosen": -165.49900817871094, | |
| "logps/rejected": -163.41685485839844, | |
| "loss": 0.6781, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.2438955008983612, | |
| "rewards/margins": 0.058673955500125885, | |
| "rewards/rejected": -0.3025694489479065, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5e-05, | |
| "logits/chosen": -2.0904877185821533, | |
| "logits/rejected": -2.1187663078308105, | |
| "logps/chosen": -136.18760681152344, | |
| "logps/rejected": -156.7350616455078, | |
| "loss": 0.7426, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.24850907921791077, | |
| "rewards/margins": -0.04654031991958618, | |
| "rewards/rejected": -0.20196877419948578, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.55e-05, | |
| "logits/chosen": -2.071505069732666, | |
| "logits/rejected": -2.0013296604156494, | |
| "logps/chosen": -126.41056060791016, | |
| "logps/rejected": -110.86978912353516, | |
| "loss": 0.7034, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.23601453006267548, | |
| "rewards/margins": 0.0055437590926885605, | |
| "rewards/rejected": -0.24155831336975098, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.600000000000001e-05, | |
| "logits/chosen": -2.0711007118225098, | |
| "logits/rejected": -2.0291450023651123, | |
| "logps/chosen": -182.40138244628906, | |
| "logps/rejected": -183.68246459960938, | |
| "loss": 0.7034, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.3310171365737915, | |
| "rewards/margins": -0.006781145930290222, | |
| "rewards/rejected": -0.3242359459400177, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.6500000000000005e-05, | |
| "logits/chosen": -2.010193109512329, | |
| "logits/rejected": -1.9630932807922363, | |
| "logps/chosen": -169.18971252441406, | |
| "logps/rejected": -173.6016082763672, | |
| "loss": 0.6432, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.35679909586906433, | |
| "rewards/margins": 0.1112755835056305, | |
| "rewards/rejected": -0.4680746793746948, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.7e-05, | |
| "logits/chosen": -2.238032817840576, | |
| "logits/rejected": -2.218147039413452, | |
| "logps/chosen": -172.18995666503906, | |
| "logps/rejected": -185.42132568359375, | |
| "loss": 0.6778, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2448209524154663, | |
| "rewards/margins": 0.062459129840135574, | |
| "rewards/rejected": -0.3072800934314728, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.75e-05, | |
| "logits/chosen": -2.000636339187622, | |
| "logits/rejected": -2.040536642074585, | |
| "logps/chosen": -146.30796813964844, | |
| "logps/rejected": -154.10507202148438, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.21792376041412354, | |
| "rewards/margins": 0.12938888370990753, | |
| "rewards/rejected": -0.3473126292228699, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8e-05, | |
| "logits/chosen": -2.2271435260772705, | |
| "logits/rejected": -2.1643269062042236, | |
| "logps/chosen": -196.40042114257812, | |
| "logps/rejected": -170.08030700683594, | |
| "loss": 0.6553, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.2668391764163971, | |
| "rewards/margins": 0.08902429789304733, | |
| "rewards/rejected": -0.3558635115623474, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.85e-05, | |
| "logits/chosen": -2.2633023262023926, | |
| "logits/rejected": -2.2394962310791016, | |
| "logps/chosen": -164.4300537109375, | |
| "logps/rejected": -163.46556091308594, | |
| "loss": 0.6129, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.20561742782592773, | |
| "rewards/margins": 0.20790418982505798, | |
| "rewards/rejected": -0.4135216474533081, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9e-05, | |
| "logits/chosen": -2.354341983795166, | |
| "logits/rejected": -2.3002171516418457, | |
| "logps/chosen": -204.92608642578125, | |
| "logps/rejected": -184.40310668945312, | |
| "loss": 0.7651, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3943729102611542, | |
| "rewards/margins": -0.058540064841508865, | |
| "rewards/rejected": -0.3358328342437744, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9500000000000004e-05, | |
| "logits/chosen": -2.152163028717041, | |
| "logits/rejected": -1.9742796421051025, | |
| "logps/chosen": -181.94400024414062, | |
| "logps/rejected": -157.1391143798828, | |
| "loss": 0.7108, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.45200178027153015, | |
| "rewards/margins": -0.008755732327699661, | |
| "rewards/rejected": -0.4432460367679596, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5e-05, | |
| "logits/chosen": -2.1966938972473145, | |
| "logits/rejected": -2.180880308151245, | |
| "logps/chosen": -164.84500122070312, | |
| "logps/rejected": -162.55015563964844, | |
| "loss": 0.7124, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.45281216502189636, | |
| "rewards/margins": 0.0067802779376506805, | |
| "rewards/rejected": -0.45959243178367615, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.999983625432647e-05, | |
| "logits/chosen": -2.185654640197754, | |
| "logits/rejected": -2.174161911010742, | |
| "logps/chosen": -178.64385986328125, | |
| "logps/rejected": -170.22705078125, | |
| "loss": 0.6557, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.2829210162162781, | |
| "rewards/margins": 0.11552949994802475, | |
| "rewards/rejected": -0.3984505236148834, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999934501945087e-05, | |
| "logits/chosen": -2.2388956546783447, | |
| "logits/rejected": -2.3327696323394775, | |
| "logps/chosen": -188.85293579101562, | |
| "logps/rejected": -204.47427368164062, | |
| "loss": 0.7025, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2629534602165222, | |
| "rewards/margins": 0.021605387330055237, | |
| "rewards/rejected": -0.28455886244773865, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9998526301808224e-05, | |
| "logits/chosen": -2.111642360687256, | |
| "logits/rejected": -2.178530693054199, | |
| "logps/chosen": -168.89016723632812, | |
| "logps/rejected": -228.85340881347656, | |
| "loss": 0.6831, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.34425827860832214, | |
| "rewards/margins": 0.06618466973304749, | |
| "rewards/rejected": -0.41044288873672485, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999738011212344e-05, | |
| "logits/chosen": -2.111250638961792, | |
| "logits/rejected": -2.113889217376709, | |
| "logps/chosen": -138.0966796875, | |
| "logps/rejected": -153.44723510742188, | |
| "loss": 0.7545, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.38345861434936523, | |
| "rewards/margins": -0.08550610393285751, | |
| "rewards/rejected": -0.2979525327682495, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.99959064654112e-05, | |
| "logits/chosen": -2.2508976459503174, | |
| "logits/rejected": -2.2036538124084473, | |
| "logps/chosen": -215.6832733154297, | |
| "logps/rejected": -238.4425506591797, | |
| "loss": 0.6683, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.3765934109687805, | |
| "rewards/margins": 0.21266478300094604, | |
| "rewards/rejected": -0.5892581939697266, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9994105380975785e-05, | |
| "logits/chosen": -2.184192180633545, | |
| "logits/rejected": -2.1407535076141357, | |
| "logps/chosen": -173.4888916015625, | |
| "logps/rejected": -173.6526336669922, | |
| "loss": 0.5906, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.1949945092201233, | |
| "rewards/margins": 0.25307750701904297, | |
| "rewards/rejected": -0.44807201623916626, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999197688241076e-05, | |
| "logits/chosen": -2.066265106201172, | |
| "logits/rejected": -2.0473287105560303, | |
| "logps/chosen": -163.60227966308594, | |
| "logps/rejected": -162.29248046875, | |
| "loss": 0.7303, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3194810748100281, | |
| "rewards/margins": -0.05436049401760101, | |
| "rewards/rejected": -0.26512062549591064, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.998952099759874e-05, | |
| "logits/chosen": -1.989267110824585, | |
| "logits/rejected": -2.111635446548462, | |
| "logps/chosen": -126.28399658203125, | |
| "logps/rejected": -151.67735290527344, | |
| "loss": 0.7078, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.264290452003479, | |
| "rewards/margins": -0.01643555797636509, | |
| "rewards/rejected": -0.24785491824150085, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9986737758710946e-05, | |
| "logits/chosen": -2.0882601737976074, | |
| "logits/rejected": -2.1823153495788574, | |
| "logps/chosen": -150.013916015625, | |
| "logps/rejected": -155.45712280273438, | |
| "loss": 0.7514, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.37010160088539124, | |
| "rewards/margins": -0.04745666682720184, | |
| "rewards/rejected": -0.3226448893547058, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.998362720220684e-05, | |
| "logits/chosen": -2.0734307765960693, | |
| "logits/rejected": -2.0464093685150146, | |
| "logps/chosen": -152.87559509277344, | |
| "logps/rejected": -171.39698791503906, | |
| "loss": 0.6578, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2491399347782135, | |
| "rewards/margins": 0.08997855335474014, | |
| "rewards/rejected": -0.33911848068237305, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9980189368833656e-05, | |
| "logits/chosen": -2.0747406482696533, | |
| "logits/rejected": -2.0951552391052246, | |
| "logps/chosen": -144.61111450195312, | |
| "logps/rejected": -154.14492797851562, | |
| "loss": 0.6906, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.2565230131149292, | |
| "rewards/margins": 0.036110781133174896, | |
| "rewards/rejected": -0.2926337718963623, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9976424303625815e-05, | |
| "logits/chosen": -2.039541244506836, | |
| "logits/rejected": -2.0199966430664062, | |
| "logps/chosen": -254.40066528320312, | |
| "logps/rejected": -241.01333618164062, | |
| "loss": 0.7487, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.5225521922111511, | |
| "rewards/margins": -0.07920274883508682, | |
| "rewards/rejected": -0.4433494508266449, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.997233205590436e-05, | |
| "logits/chosen": -2.121485948562622, | |
| "logits/rejected": -2.132420301437378, | |
| "logps/chosen": -156.11683654785156, | |
| "logps/rejected": -165.07196044921875, | |
| "loss": 0.7828, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.2995274066925049, | |
| "rewards/margins": -0.1235617995262146, | |
| "rewards/rejected": -0.1759655922651291, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9967912679276316e-05, | |
| "logits/chosen": -1.991243600845337, | |
| "logits/rejected": -2.0190415382385254, | |
| "logps/chosen": -152.22360229492188, | |
| "logps/rejected": -166.73895263671875, | |
| "loss": 0.6907, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2022009938955307, | |
| "rewards/margins": 0.020534606650471687, | |
| "rewards/rejected": -0.22273561358451843, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.996316623163401e-05, | |
| "logits/chosen": -2.069793701171875, | |
| "logits/rejected": -2.0622239112854004, | |
| "logps/chosen": -155.5286865234375, | |
| "logps/rejected": -164.81361389160156, | |
| "loss": 0.688, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.34616801142692566, | |
| "rewards/margins": 0.026021815836429596, | |
| "rewards/rejected": -0.37218981981277466, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.995809277515424e-05, | |
| "logits/chosen": -1.7386094331741333, | |
| "logits/rejected": -1.8783282041549683, | |
| "logps/chosen": -173.74142456054688, | |
| "logps/rejected": -157.15963745117188, | |
| "loss": 0.6324, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.20083683729171753, | |
| "rewards/margins": 0.16446277499198914, | |
| "rewards/rejected": -0.36529961228370667, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.995269237629755e-05, | |
| "logits/chosen": -2.0975406169891357, | |
| "logits/rejected": -2.261514902114868, | |
| "logps/chosen": -218.3932342529297, | |
| "logps/rejected": -240.0115509033203, | |
| "loss": 0.7039, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.42158380150794983, | |
| "rewards/margins": 0.023422667756676674, | |
| "rewards/rejected": -0.44500645995140076, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9946965105807275e-05, | |
| "logits/chosen": -2.0053865909576416, | |
| "logits/rejected": -2.0759031772613525, | |
| "logps/chosen": -152.17633056640625, | |
| "logps/rejected": -170.76083374023438, | |
| "loss": 0.5858, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.15409651398658752, | |
| "rewards/margins": 0.26004791259765625, | |
| "rewards/rejected": -0.4141444265842438, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9940911038708686e-05, | |
| "logits/chosen": -1.943060278892517, | |
| "logits/rejected": -2.0972752571105957, | |
| "logps/chosen": -134.478759765625, | |
| "logps/rejected": -203.55897521972656, | |
| "loss": 0.6855, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3040854036808014, | |
| "rewards/margins": 0.03960128873586655, | |
| "rewards/rejected": -0.34368667006492615, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.993453025430797e-05, | |
| "logits/chosen": -2.053199529647827, | |
| "logits/rejected": -2.086787462234497, | |
| "logps/chosen": -163.63150024414062, | |
| "logps/rejected": -168.98843383789062, | |
| "loss": 0.6745, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3369709849357605, | |
| "rewards/margins": 0.05357670038938522, | |
| "rewards/rejected": -0.3905476927757263, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.992782283619118e-05, | |
| "logits/chosen": -1.9108364582061768, | |
| "logits/rejected": -2.0180165767669678, | |
| "logps/chosen": -258.45269775390625, | |
| "logps/rejected": -297.9960632324219, | |
| "loss": 0.7029, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5936928987503052, | |
| "rewards/margins": 0.01269455999135971, | |
| "rewards/rejected": -0.6063875555992126, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.99207888722232e-05, | |
| "logits/chosen": -2.013960123062134, | |
| "logits/rejected": -2.0115973949432373, | |
| "logps/chosen": -176.5162811279297, | |
| "logps/rejected": -168.38345336914062, | |
| "loss": 0.6733, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3322643041610718, | |
| "rewards/margins": 0.07553744316101074, | |
| "rewards/rejected": -0.40780171751976013, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9913428454546494e-05, | |
| "logits/chosen": -1.939697265625, | |
| "logits/rejected": -1.9972270727157593, | |
| "logps/chosen": -151.5918731689453, | |
| "logps/rejected": -176.63543701171875, | |
| "loss": 0.7238, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.3320300877094269, | |
| "rewards/margins": 0.008037537336349487, | |
| "rewards/rejected": -0.34006762504577637, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9905741679580007e-05, | |
| "logits/chosen": -1.7919318675994873, | |
| "logits/rejected": -1.8278323411941528, | |
| "logps/chosen": -133.59677124023438, | |
| "logps/rejected": -150.1385498046875, | |
| "loss": 0.672, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.28024911880493164, | |
| "rewards/margins": 0.09316328912973404, | |
| "rewards/rejected": -0.3734124004840851, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.989772864801782e-05, | |
| "logits/chosen": -2.1430420875549316, | |
| "logits/rejected": -2.1535253524780273, | |
| "logps/chosen": -189.2075958251953, | |
| "logps/rejected": -200.56422424316406, | |
| "loss": 0.6673, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2726452648639679, | |
| "rewards/margins": 0.09557943046092987, | |
| "rewards/rejected": -0.3682246804237366, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.988938946482786e-05, | |
| "logits/chosen": -2.0016047954559326, | |
| "logits/rejected": -2.002544641494751, | |
| "logps/chosen": -180.3876495361328, | |
| "logps/rejected": -171.05935668945312, | |
| "loss": 0.7887, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.42258885502815247, | |
| "rewards/margins": -0.15555229783058167, | |
| "rewards/rejected": -0.2670365869998932, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9880724239250565e-05, | |
| "logits/chosen": -1.9078752994537354, | |
| "logits/rejected": -1.9413063526153564, | |
| "logps/chosen": -162.1499786376953, | |
| "logps/rejected": -203.156982421875, | |
| "loss": 0.6717, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4850098490715027, | |
| "rewards/margins": 0.09099780023097992, | |
| "rewards/rejected": -0.5760077238082886, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.987173308479738e-05, | |
| "logits/chosen": -2.0715203285217285, | |
| "logits/rejected": -1.9801918268203735, | |
| "logps/chosen": -175.82017517089844, | |
| "logps/rejected": -195.7940673828125, | |
| "loss": 0.6384, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5006306171417236, | |
| "rewards/margins": 0.2475757896900177, | |
| "rewards/rejected": -0.748206377029419, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.986241611924932e-05, | |
| "logits/chosen": -2.0729923248291016, | |
| "logits/rejected": -2.1148691177368164, | |
| "logps/chosen": -151.3936767578125, | |
| "logps/rejected": -147.7472381591797, | |
| "loss": 0.7244, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.4272467792034149, | |
| "rewards/margins": -0.02306976355612278, | |
| "rewards/rejected": -0.4041770100593567, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.98527734646554e-05, | |
| "logits/chosen": -1.9960330724716187, | |
| "logits/rejected": -2.0567307472229004, | |
| "logps/chosen": -155.43038940429688, | |
| "logps/rejected": -192.47073364257812, | |
| "loss": 0.6894, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.403178334236145, | |
| "rewards/margins": 0.03879673033952713, | |
| "rewards/rejected": -0.44197511672973633, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.984280524733107e-05, | |
| "logits/chosen": -2.037220001220703, | |
| "logits/rejected": -2.1089906692504883, | |
| "logps/chosen": -165.5675811767578, | |
| "logps/rejected": -177.08114624023438, | |
| "loss": 0.7429, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5028324723243713, | |
| "rewards/margins": -0.01752498745918274, | |
| "rewards/rejected": -0.4853074550628662, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.983251159785651e-05, | |
| "logits/chosen": -1.9660859107971191, | |
| "logits/rejected": -1.9128999710083008, | |
| "logps/chosen": -149.2632598876953, | |
| "logps/rejected": -158.37025451660156, | |
| "loss": 0.6813, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.45628058910369873, | |
| "rewards/margins": 0.05837492644786835, | |
| "rewards/rejected": -0.5146555304527283, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.982189265107499e-05, | |
| "logits/chosen": -2.093052387237549, | |
| "logits/rejected": -2.0723822116851807, | |
| "logps/chosen": -218.46633911132812, | |
| "logps/rejected": -182.45477294921875, | |
| "loss": 0.7192, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.520983099937439, | |
| "rewards/margins": -0.025474410504102707, | |
| "rewards/rejected": -0.49550867080688477, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.981094854609101e-05, | |
| "logits/chosen": -2.179332971572876, | |
| "logits/rejected": -2.219320774078369, | |
| "logps/chosen": -177.7232666015625, | |
| "logps/rejected": -194.99066162109375, | |
| "loss": 0.6147, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.5284168720245361, | |
| "rewards/margins": 0.20047883689403534, | |
| "rewards/rejected": -0.7288957238197327, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.979967942626858e-05, | |
| "logits/chosen": -2.1552772521972656, | |
| "logits/rejected": -2.1887803077697754, | |
| "logps/chosen": -168.3583984375, | |
| "logps/rejected": -219.6854248046875, | |
| "loss": 0.5772, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.38589900732040405, | |
| "rewards/margins": 0.27822399139404297, | |
| "rewards/rejected": -0.6641228795051575, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.978808543922925e-05, | |
| "logits/chosen": -2.143907070159912, | |
| "logits/rejected": -2.1659560203552246, | |
| "logps/chosen": -187.9869384765625, | |
| "logps/rejected": -199.4171142578125, | |
| "loss": 0.7769, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6709006428718567, | |
| "rewards/margins": -0.10143251717090607, | |
| "rewards/rejected": -0.5694682002067566, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977616673685026e-05, | |
| "logits/chosen": -2.0695879459381104, | |
| "logits/rejected": -2.173468589782715, | |
| "logps/chosen": -180.43331909179688, | |
| "logps/rejected": -217.425048828125, | |
| "loss": 0.688, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5364802479743958, | |
| "rewards/margins": 0.12391349673271179, | |
| "rewards/rejected": -0.6603937745094299, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9763923475262464e-05, | |
| "logits/chosen": -2.2173025608062744, | |
| "logits/rejected": -2.219977855682373, | |
| "logps/chosen": -152.98944091796875, | |
| "logps/rejected": -158.1151123046875, | |
| "loss": 0.6759, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.22611242532730103, | |
| "rewards/margins": 0.08070263266563416, | |
| "rewards/rejected": -0.3068150281906128, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.975135581484836e-05, | |
| "logits/chosen": -2.240402936935425, | |
| "logits/rejected": -2.2945306301116943, | |
| "logps/chosen": -233.8233184814453, | |
| "logps/rejected": -231.96136474609375, | |
| "loss": 0.8167, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6244059801101685, | |
| "rewards/margins": -0.066192626953125, | |
| "rewards/rejected": -0.5582134127616882, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9738463920239955e-05, | |
| "logits/chosen": -2.2763917446136475, | |
| "logits/rejected": -2.253692626953125, | |
| "logps/chosen": -149.86184692382812, | |
| "logps/rejected": -153.2465057373047, | |
| "loss": 0.7391, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.4595310091972351, | |
| "rewards/margins": -0.06695618480443954, | |
| "rewards/rejected": -0.39257481694221497, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.972524796031659e-05, | |
| "logits/chosen": -2.093257427215576, | |
| "logits/rejected": -2.1301755905151367, | |
| "logps/chosen": -188.2993621826172, | |
| "logps/rejected": -183.14076232910156, | |
| "loss": 0.8359, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.7259301543235779, | |
| "rewards/margins": -0.19261422753334045, | |
| "rewards/rejected": -0.533315896987915, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.971170810820279e-05, | |
| "logits/chosen": -2.1239709854125977, | |
| "logits/rejected": -2.1164982318878174, | |
| "logps/chosen": -161.5440216064453, | |
| "logps/rejected": -165.34368896484375, | |
| "loss": 0.6823, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.43793031573295593, | |
| "rewards/margins": 0.09565365314483643, | |
| "rewards/rejected": -0.53358393907547, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.969784454126591e-05, | |
| "logits/chosen": -2.10398006439209, | |
| "logits/rejected": -2.153364419937134, | |
| "logps/chosen": -162.19198608398438, | |
| "logps/rejected": -176.05035400390625, | |
| "loss": 0.7656, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.7570656538009644, | |
| "rewards/margins": -0.09328162670135498, | |
| "rewards/rejected": -0.6637839674949646, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9683657441113884e-05, | |
| "logits/chosen": -1.9448606967926025, | |
| "logits/rejected": -1.9559727907180786, | |
| "logps/chosen": -198.11264038085938, | |
| "logps/rejected": -197.9307098388672, | |
| "loss": 0.7806, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6761496067047119, | |
| "rewards/margins": -0.03766350448131561, | |
| "rewards/rejected": -0.6384860873222351, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.966914699359282e-05, | |
| "logits/chosen": -2.157869815826416, | |
| "logits/rejected": -2.1131656169891357, | |
| "logps/chosen": -172.5656280517578, | |
| "logps/rejected": -186.9047393798828, | |
| "loss": 0.6644, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.39786607027053833, | |
| "rewards/margins": 0.10626877844333649, | |
| "rewards/rejected": -0.5041348934173584, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.965431338878456e-05, | |
| "logits/chosen": -2.0113039016723633, | |
| "logits/rejected": -2.0353922843933105, | |
| "logps/chosen": -161.632080078125, | |
| "logps/rejected": -186.112060546875, | |
| "loss": 0.7521, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.5831659436225891, | |
| "rewards/margins": 0.03063490055501461, | |
| "rewards/rejected": -0.6138008832931519, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9639156821004184e-05, | |
| "logits/chosen": -2.1529879570007324, | |
| "logits/rejected": -2.150418996810913, | |
| "logps/chosen": -209.376708984375, | |
| "logps/rejected": -202.34423828125, | |
| "loss": 0.669, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.6463437080383301, | |
| "rewards/margins": 0.10935439169406891, | |
| "rewards/rejected": -0.7556981444358826, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.962367748879748e-05, | |
| "logits/chosen": -1.9734456539154053, | |
| "logits/rejected": -1.9453670978546143, | |
| "logps/chosen": -137.22714233398438, | |
| "logps/rejected": -132.848388671875, | |
| "loss": 0.7078, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.39396798610687256, | |
| "rewards/margins": 0.0391971729695797, | |
| "rewards/rejected": -0.43316513299942017, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.960787559493836e-05, | |
| "logits/chosen": -2.1786108016967773, | |
| "logits/rejected": -2.0795602798461914, | |
| "logps/chosen": -206.21754455566406, | |
| "logps/rejected": -200.25201416015625, | |
| "loss": 0.7455, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5245193243026733, | |
| "rewards/margins": -0.07819047570228577, | |
| "rewards/rejected": -0.4463288486003876, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.959175134642614e-05, | |
| "logits/chosen": -2.0515449047088623, | |
| "logits/rejected": -2.150259494781494, | |
| "logps/chosen": -152.1802215576172, | |
| "logps/rejected": -189.8648681640625, | |
| "loss": 0.6417, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4843188226222992, | |
| "rewards/margins": 0.27961939573287964, | |
| "rewards/rejected": -0.7639381885528564, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.957530495448292e-05, | |
| "logits/chosen": -2.147951602935791, | |
| "logits/rejected": -2.095947504043579, | |
| "logps/chosen": -186.4452667236328, | |
| "logps/rejected": -178.0105438232422, | |
| "loss": 0.8001, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6299381256103516, | |
| "rewards/margins": -0.16357335448265076, | |
| "rewards/rejected": -0.4663648307323456, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.955853663455072e-05, | |
| "logits/chosen": -2.175231695175171, | |
| "logits/rejected": -2.0836031436920166, | |
| "logps/chosen": -180.3741455078125, | |
| "logps/rejected": -164.85044860839844, | |
| "loss": 0.8194, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.5521696209907532, | |
| "rewards/margins": -0.15574771165847778, | |
| "rewards/rejected": -0.3964219093322754, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.954144660628875e-05, | |
| "logits/chosen": -2.202047824859619, | |
| "logits/rejected": -2.1592371463775635, | |
| "logps/chosen": -158.53955078125, | |
| "logps/rejected": -128.51901245117188, | |
| "loss": 0.6143, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.22061151266098022, | |
| "rewards/margins": 0.20456114411354065, | |
| "rewards/rejected": -0.4251726567745209, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.952403509357044e-05, | |
| "logits/chosen": -2.3511812686920166, | |
| "logits/rejected": -2.3003928661346436, | |
| "logps/chosen": -215.75103759765625, | |
| "logps/rejected": -195.51934814453125, | |
| "loss": 0.6105, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3606884181499481, | |
| "rewards/margins": 0.20525097846984863, | |
| "rewards/rejected": -0.5659393668174744, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9506302324480605e-05, | |
| "logits/chosen": -2.2802467346191406, | |
| "logits/rejected": -2.2563934326171875, | |
| "logps/chosen": -156.922119140625, | |
| "logps/rejected": -165.9469451904297, | |
| "loss": 0.6965, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.45227211713790894, | |
| "rewards/margins": 0.004478612914681435, | |
| "rewards/rejected": -0.456750750541687, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.948824853131236e-05, | |
| "logits/chosen": -2.140197515487671, | |
| "logits/rejected": -2.1911277770996094, | |
| "logps/chosen": -145.484619140625, | |
| "logps/rejected": -171.29197692871094, | |
| "loss": 0.6379, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2703571319580078, | |
| "rewards/margins": 0.15410497784614563, | |
| "rewards/rejected": -0.42446210980415344, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.946987395056416e-05, | |
| "logits/chosen": -2.2976553440093994, | |
| "logits/rejected": -2.323747158050537, | |
| "logps/chosen": -179.03887939453125, | |
| "logps/rejected": -164.2755889892578, | |
| "loss": 0.7579, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.43789076805114746, | |
| "rewards/margins": -0.08924071490764618, | |
| "rewards/rejected": -0.3486500680446625, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945117882293666e-05, | |
| "logits/chosen": -2.3072831630706787, | |
| "logits/rejected": -2.2856385707855225, | |
| "logps/chosen": -216.36965942382812, | |
| "logps/rejected": -217.8874053955078, | |
| "loss": 0.7023, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.48600953817367554, | |
| "rewards/margins": 0.011942245066165924, | |
| "rewards/rejected": -0.49795183539390564, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9432163393329544e-05, | |
| "logits/chosen": -2.2269880771636963, | |
| "logits/rejected": -2.2329916954040527, | |
| "logps/chosen": -151.3675537109375, | |
| "logps/rejected": -156.10838317871094, | |
| "loss": 0.6478, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.386996865272522, | |
| "rewards/margins": 0.13896134495735168, | |
| "rewards/rejected": -0.5259581804275513, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.941282791083836e-05, | |
| "logits/chosen": -2.1749000549316406, | |
| "logits/rejected": -2.209458589553833, | |
| "logps/chosen": -158.9986114501953, | |
| "logps/rejected": -189.9351806640625, | |
| "loss": 0.6596, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.4211054742336273, | |
| "rewards/margins": 0.09134037792682648, | |
| "rewards/rejected": -0.5124458074569702, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.939317262875125e-05, | |
| "logits/chosen": -2.20003604888916, | |
| "logits/rejected": -2.208467721939087, | |
| "logps/chosen": -140.90966796875, | |
| "logps/rejected": -156.3489532470703, | |
| "loss": 0.7092, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5110758543014526, | |
| "rewards/margins": 0.0035289786756038666, | |
| "rewards/rejected": -0.5146048069000244, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.937319780454559e-05, | |
| "logits/chosen": -2.119629383087158, | |
| "logits/rejected": -2.2122950553894043, | |
| "logps/chosen": -173.94345092773438, | |
| "logps/rejected": -187.1260986328125, | |
| "loss": 0.7071, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4379804730415344, | |
| "rewards/margins": 0.03866587579250336, | |
| "rewards/rejected": -0.4766463339328766, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.935290369988468e-05, | |
| "logits/chosen": -2.2016067504882812, | |
| "logits/rejected": -2.1686062812805176, | |
| "logps/chosen": -180.24510192871094, | |
| "logps/rejected": -176.13328552246094, | |
| "loss": 0.7777, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5541373491287231, | |
| "rewards/margins": -0.1263955533504486, | |
| "rewards/rejected": -0.4277418255805969, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.933229058061425e-05, | |
| "logits/chosen": -2.1250979900360107, | |
| "logits/rejected": -2.1696457862854004, | |
| "logps/chosen": -197.02017211914062, | |
| "logps/rejected": -205.6886749267578, | |
| "loss": 0.7329, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4308657646179199, | |
| "rewards/margins": -0.018534280359745026, | |
| "rewards/rejected": -0.4123315215110779, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.931135871675905e-05, | |
| "logits/chosen": -2.2878334522247314, | |
| "logits/rejected": -2.2978017330169678, | |
| "logps/chosen": -139.80540466308594, | |
| "logps/rejected": -135.60267639160156, | |
| "loss": 0.6976, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.35316216945648193, | |
| "rewards/margins": 0.019971035420894623, | |
| "rewards/rejected": -0.37313324213027954, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.929010838251923e-05, | |
| "logits/chosen": -2.159762382507324, | |
| "logits/rejected": -2.2488741874694824, | |
| "logps/chosen": -148.69459533691406, | |
| "logps/rejected": -180.4253692626953, | |
| "loss": 0.7472, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.45069968700408936, | |
| "rewards/margins": -0.067423515021801, | |
| "rewards/rejected": -0.3832761347293854, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.926853985626682e-05, | |
| "logits/chosen": -2.008970260620117, | |
| "logits/rejected": -2.031158208847046, | |
| "logps/chosen": -156.30447387695312, | |
| "logps/rejected": -173.08717346191406, | |
| "loss": 0.719, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5630744099617004, | |
| "rewards/margins": 0.002446278929710388, | |
| "rewards/rejected": -0.565520703792572, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924665342054204e-05, | |
| "logits/chosen": -2.3275699615478516, | |
| "logits/rejected": -2.3046929836273193, | |
| "logps/chosen": -217.69931030273438, | |
| "logps/rejected": -190.64454650878906, | |
| "loss": 0.8186, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.5474885702133179, | |
| "rewards/margins": -0.20113232731819153, | |
| "rewards/rejected": -0.34635621309280396, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9224449362049654e-05, | |
| "logits/chosen": -2.0889933109283447, | |
| "logits/rejected": -2.06156063079834, | |
| "logps/chosen": -174.817138671875, | |
| "logps/rejected": -157.86846923828125, | |
| "loss": 0.6693, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.3793041706085205, | |
| "rewards/margins": 0.1282089203596115, | |
| "rewards/rejected": -0.5075130462646484, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.920192797165511e-05, | |
| "logits/chosen": -2.0895235538482666, | |
| "logits/rejected": -2.0723814964294434, | |
| "logps/chosen": -167.65811157226562, | |
| "logps/rejected": -158.02890014648438, | |
| "loss": 0.8024, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.7083860635757446, | |
| "rewards/margins": -0.1765305995941162, | |
| "rewards/rejected": -0.531855583190918, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.917908954438084e-05, | |
| "logits/chosen": -2.287144422531128, | |
| "logits/rejected": -2.1813595294952393, | |
| "logps/chosen": -173.13775634765625, | |
| "logps/rejected": -176.2907257080078, | |
| "loss": 0.7204, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.45302245020866394, | |
| "rewards/margins": 0.024594204500317574, | |
| "rewards/rejected": -0.47761663794517517, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9155934379402335e-05, | |
| "logits/chosen": -2.252314329147339, | |
| "logits/rejected": -2.3036305904388428, | |
| "logps/chosen": -146.82371520996094, | |
| "logps/rejected": -164.65354919433594, | |
| "loss": 0.7665, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5438223481178284, | |
| "rewards/margins": -0.07613788545131683, | |
| "rewards/rejected": -0.4676845073699951, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.913246278004425e-05, | |
| "logits/chosen": -2.1716370582580566, | |
| "logits/rejected": -2.1546061038970947, | |
| "logps/chosen": -213.76388549804688, | |
| "logps/rejected": -192.2274169921875, | |
| "loss": 0.7251, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.4964134693145752, | |
| "rewards/margins": 0.02723705768585205, | |
| "rewards/rejected": -0.5236505270004272, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.91086750537764e-05, | |
| "logits/chosen": -2.133942127227783, | |
| "logits/rejected": -2.118009567260742, | |
| "logps/chosen": -151.01699829101562, | |
| "logps/rejected": -164.3484649658203, | |
| "loss": 0.7692, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.48824450373649597, | |
| "rewards/margins": 0.013736486434936523, | |
| "rewards/rejected": -0.5019809603691101, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.908457151220976e-05, | |
| "logits/chosen": -2.1285924911499023, | |
| "logits/rejected": -2.175334930419922, | |
| "logps/chosen": -179.7931671142578, | |
| "logps/rejected": -184.04031372070312, | |
| "loss": 0.6895, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3827531337738037, | |
| "rewards/margins": 0.1502717286348343, | |
| "rewards/rejected": -0.5330249071121216, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9060152471092414e-05, | |
| "logits/chosen": -2.1608972549438477, | |
| "logits/rejected": -2.086259603500366, | |
| "logps/chosen": -165.98989868164062, | |
| "logps/rejected": -148.0418243408203, | |
| "loss": 0.7032, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.23593486845493317, | |
| "rewards/margins": 0.017191678285598755, | |
| "rewards/rejected": -0.2531265318393707, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9035418250305314e-05, | |
| "logits/chosen": -2.120004177093506, | |
| "logits/rejected": -2.181701183319092, | |
| "logps/chosen": -153.8833465576172, | |
| "logps/rejected": -167.73036193847656, | |
| "loss": 0.5838, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.5051828622817993, | |
| "rewards/margins": 0.26960399746894836, | |
| "rewards/rejected": -0.7747868299484253, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9010369173858204e-05, | |
| "logits/chosen": -2.1011998653411865, | |
| "logits/rejected": -2.186149835586548, | |
| "logps/chosen": -158.24081420898438, | |
| "logps/rejected": -180.07054138183594, | |
| "loss": 0.8031, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.5937004685401917, | |
| "rewards/margins": -0.1665870100259781, | |
| "rewards/rejected": -0.42711353302001953, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8985005569885325e-05, | |
| "logits/chosen": -2.1024258136749268, | |
| "logits/rejected": -2.161679267883301, | |
| "logps/chosen": -195.25772094726562, | |
| "logps/rejected": -193.77999877929688, | |
| "loss": 0.8584, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.7584065198898315, | |
| "rewards/margins": -0.21715182065963745, | |
| "rewards/rejected": -0.5412546992301941, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.89593277706411e-05, | |
| "logits/chosen": -2.0745136737823486, | |
| "logits/rejected": -2.2386059761047363, | |
| "logps/chosen": -145.7827606201172, | |
| "logps/rejected": -167.44580078125, | |
| "loss": 0.7023, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.2986881732940674, | |
| "rewards/margins": 0.007890693843364716, | |
| "rewards/rejected": -0.3065788745880127, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.893333611249581e-05, | |
| "logits/chosen": -2.385495185852051, | |
| "logits/rejected": -2.412301778793335, | |
| "logps/chosen": -179.8057098388672, | |
| "logps/rejected": -182.3574981689453, | |
| "loss": 0.691, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3335916996002197, | |
| "rewards/margins": 0.100555419921875, | |
| "rewards/rejected": -0.4341471195220947, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.890703093593118e-05, | |
| "logits/chosen": -2.1535239219665527, | |
| "logits/rejected": -2.219959259033203, | |
| "logps/chosen": -202.92239379882812, | |
| "logps/rejected": -213.8419952392578, | |
| "loss": 0.7452, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5874291658401489, | |
| "rewards/margins": -0.023649999871850014, | |
| "rewards/rejected": -0.5637791752815247, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8880412585535926e-05, | |
| "logits/chosen": -2.031306743621826, | |
| "logits/rejected": -1.9869210720062256, | |
| "logps/chosen": -185.4857635498047, | |
| "logps/rejected": -195.2773895263672, | |
| "loss": 0.7235, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.444195419549942, | |
| "rewards/margins": -0.03562233969569206, | |
| "rewards/rejected": -0.40857309103012085, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.885348141000122e-05, | |
| "logits/chosen": -2.041801691055298, | |
| "logits/rejected": -2.128411054611206, | |
| "logps/chosen": -173.86544799804688, | |
| "logps/rejected": -193.53025817871094, | |
| "loss": 0.7379, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.3505318760871887, | |
| "rewards/margins": -0.0760502815246582, | |
| "rewards/rejected": -0.2744815945625305, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8826237762116144e-05, | |
| "logits/chosen": -2.1389496326446533, | |
| "logits/rejected": -2.0953948497772217, | |
| "logps/chosen": -191.95359802246094, | |
| "logps/rejected": -191.35183715820312, | |
| "loss": 0.8064, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.5560134053230286, | |
| "rewards/margins": -0.18388979136943817, | |
| "rewards/rejected": -0.3721236288547516, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.879868199876305e-05, | |
| "logits/chosen": -1.9923707246780396, | |
| "logits/rejected": -2.1531779766082764, | |
| "logps/chosen": -130.49436950683594, | |
| "logps/rejected": -178.189697265625, | |
| "loss": 0.6928, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.28136155009269714, | |
| "rewards/margins": 0.135418102145195, | |
| "rewards/rejected": -0.41677963733673096, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.877081448091291e-05, | |
| "logits/chosen": -2.062145233154297, | |
| "logits/rejected": -2.079340934753418, | |
| "logps/chosen": -173.77001953125, | |
| "logps/rejected": -186.27011108398438, | |
| "loss": 0.7799, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.4552696943283081, | |
| "rewards/margins": -0.11873035877943039, | |
| "rewards/rejected": -0.3365393579006195, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.874263557362056e-05, | |
| "logits/chosen": -1.9422199726104736, | |
| "logits/rejected": -1.9917988777160645, | |
| "logps/chosen": -154.36346435546875, | |
| "logps/rejected": -187.7787322998047, | |
| "loss": 0.6659, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.33184733986854553, | |
| "rewards/margins": 0.12450146675109863, | |
| "rewards/rejected": -0.4563487768173218, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.871414564601992e-05, | |
| "logits/chosen": -2.1566262245178223, | |
| "logits/rejected": -2.126441478729248, | |
| "logps/chosen": -169.04275512695312, | |
| "logps/rejected": -172.1595458984375, | |
| "loss": 0.8547, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.5513345003128052, | |
| "rewards/margins": -0.21477870643138885, | |
| "rewards/rejected": -0.3365558385848999, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.868534507131919e-05, | |
| "logits/chosen": -2.154214859008789, | |
| "logits/rejected": -2.18147873878479, | |
| "logps/chosen": -199.698486328125, | |
| "logps/rejected": -246.1205596923828, | |
| "loss": 0.7063, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5538137555122375, | |
| "rewards/margins": 0.07623854279518127, | |
| "rewards/rejected": -0.6300523281097412, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.865623422679593e-05, | |
| "logits/chosen": -2.311129570007324, | |
| "logits/rejected": -2.2521347999572754, | |
| "logps/chosen": -207.49635314941406, | |
| "logps/rejected": -215.36251831054688, | |
| "loss": 0.708, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3663349747657776, | |
| "rewards/margins": 0.038322776556015015, | |
| "rewards/rejected": -0.4046577215194702, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.862681349379212e-05, | |
| "logits/chosen": -2.201547384262085, | |
| "logits/rejected": -2.1968986988067627, | |
| "logps/chosen": -167.12265014648438, | |
| "logps/rejected": -184.99134826660156, | |
| "loss": 0.7268, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4530975818634033, | |
| "rewards/margins": -0.0044959532096982, | |
| "rewards/rejected": -0.4486016035079956, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8597083257709194e-05, | |
| "logits/chosen": -2.122652530670166, | |
| "logits/rejected": -2.1181819438934326, | |
| "logps/chosen": -181.10531616210938, | |
| "logps/rejected": -199.0648956298828, | |
| "loss": 0.7389, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.4301409125328064, | |
| "rewards/margins": -0.04280529171228409, | |
| "rewards/rejected": -0.3873356580734253, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.856704390800294e-05, | |
| "logits/chosen": -2.100674629211426, | |
| "logits/rejected": -2.028221845626831, | |
| "logps/chosen": -172.59988403320312, | |
| "logps/rejected": -162.9868927001953, | |
| "loss": 0.6582, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.4320454001426697, | |
| "rewards/margins": 0.1003347635269165, | |
| "rewards/rejected": -0.5323801040649414, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8536695838178456e-05, | |
| "logits/chosen": -2.1120193004608154, | |
| "logits/rejected": -2.0844578742980957, | |
| "logps/chosen": -120.8355712890625, | |
| "logps/rejected": -124.54715728759766, | |
| "loss": 0.632, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.16351774334907532, | |
| "rewards/margins": 0.15973550081253052, | |
| "rewards/rejected": -0.3232532739639282, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.850603944578494e-05, | |
| "logits/chosen": -2.1671504974365234, | |
| "logits/rejected": -2.30904483795166, | |
| "logps/chosen": -162.22195434570312, | |
| "logps/rejected": -193.69979858398438, | |
| "loss": 0.6579, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3427208364009857, | |
| "rewards/margins": 0.1314423829317093, | |
| "rewards/rejected": -0.4741632044315338, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.847507513241053e-05, | |
| "logits/chosen": -2.1925673484802246, | |
| "logits/rejected": -2.1313178539276123, | |
| "logps/chosen": -166.4717559814453, | |
| "logps/rejected": -157.5460968017578, | |
| "loss": 0.744, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.5565357208251953, | |
| "rewards/margins": -0.05020551383495331, | |
| "rewards/rejected": -0.5063301920890808, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.844380330367701e-05, | |
| "logits/chosen": -2.0932931900024414, | |
| "logits/rejected": -2.0991854667663574, | |
| "logps/chosen": -218.98275756835938, | |
| "logps/rejected": -256.97625732421875, | |
| "loss": 0.737, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.7028021216392517, | |
| "rewards/margins": -0.03586144745349884, | |
| "rewards/rejected": -0.6669406294822693, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.841222436923451e-05, | |
| "logits/chosen": -2.060140371322632, | |
| "logits/rejected": -2.0552310943603516, | |
| "logps/chosen": -156.42568969726562, | |
| "logps/rejected": -159.0933837890625, | |
| "loss": 0.6313, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.20694246888160706, | |
| "rewards/margins": 0.18116910755634308, | |
| "rewards/rejected": -0.38811156153678894, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.8380338742756157e-05, | |
| "logits/chosen": -2.1082582473754883, | |
| "logits/rejected": -2.144632339477539, | |
| "logps/chosen": -168.499755859375, | |
| "logps/rejected": -153.81903076171875, | |
| "loss": 0.7304, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.37163180112838745, | |
| "rewards/margins": -0.0361662432551384, | |
| "rewards/rejected": -0.33546555042266846, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.834814684193261e-05, | |
| "logits/chosen": -2.0787179470062256, | |
| "logits/rejected": -2.087756395339966, | |
| "logps/chosen": -172.99032592773438, | |
| "logps/rejected": -177.86279296875, | |
| "loss": 0.7514, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6183903217315674, | |
| "rewards/margins": -0.032746944576501846, | |
| "rewards/rejected": -0.5856433510780334, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.831564908846661e-05, | |
| "logits/chosen": -2.1304032802581787, | |
| "logits/rejected": -2.1413257122039795, | |
| "logps/chosen": -142.9176483154297, | |
| "logps/rejected": -132.79454040527344, | |
| "loss": 0.7649, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.45798230171203613, | |
| "rewards/margins": -0.09754021465778351, | |
| "rewards/rejected": -0.36044207215309143, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.8282845908067507e-05, | |
| "logits/chosen": -2.056964635848999, | |
| "logits/rejected": -2.063077449798584, | |
| "logps/chosen": -152.53311157226562, | |
| "logps/rejected": -161.74180603027344, | |
| "loss": 0.613, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.1394699364900589, | |
| "rewards/margins": 0.18046818673610687, | |
| "rewards/rejected": -0.31993812322616577, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.82497377304456e-05, | |
| "logits/chosen": -2.1834802627563477, | |
| "logits/rejected": -2.1908788681030273, | |
| "logps/chosen": -175.36349487304688, | |
| "logps/rejected": -176.58810424804688, | |
| "loss": 0.7565, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4943966269493103, | |
| "rewards/margins": -0.07095511257648468, | |
| "rewards/rejected": -0.4234415292739868, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.821632498930656e-05, | |
| "logits/chosen": -1.855932593345642, | |
| "logits/rejected": -1.9100501537322998, | |
| "logps/chosen": -171.09329223632812, | |
| "logps/rejected": -199.5696563720703, | |
| "loss": 0.758, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5612733960151672, | |
| "rewards/margins": -0.07660679519176483, | |
| "rewards/rejected": -0.4846665561199188, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.818260812234572e-05, | |
| "logits/chosen": -2.092160940170288, | |
| "logits/rejected": -2.1146998405456543, | |
| "logps/chosen": -159.81150817871094, | |
| "logps/rejected": -166.06539916992188, | |
| "loss": 0.7463, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.515409529209137, | |
| "rewards/margins": -0.049059510231018066, | |
| "rewards/rejected": -0.4663500487804413, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.8148587571242373e-05, | |
| "logits/chosen": -2.0305914878845215, | |
| "logits/rejected": -2.0486207008361816, | |
| "logps/chosen": -160.466064453125, | |
| "logps/rejected": -183.69647216796875, | |
| "loss": 0.6401, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5715111494064331, | |
| "rewards/margins": 0.16600874066352844, | |
| "rewards/rejected": -0.7375199198722839, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.811426378165398e-05, | |
| "logits/chosen": -1.965864658355713, | |
| "logits/rejected": -1.960595965385437, | |
| "logps/chosen": -152.71310424804688, | |
| "logps/rejected": -151.248779296875, | |
| "loss": 0.7115, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5327611565589905, | |
| "rewards/margins": 0.03953855484724045, | |
| "rewards/rejected": -0.5722997188568115, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.807963720321028e-05, | |
| "logits/chosen": -2.0893819332122803, | |
| "logits/rejected": -2.1314423084259033, | |
| "logps/chosen": -164.28872680664062, | |
| "logps/rejected": -195.37356567382812, | |
| "loss": 0.6274, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3729438781738281, | |
| "rewards/margins": 0.19530445337295532, | |
| "rewards/rejected": -0.5682482719421387, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.804470828950748e-05, | |
| "logits/chosen": -1.9575481414794922, | |
| "logits/rejected": -1.9728269577026367, | |
| "logps/chosen": -165.69314575195312, | |
| "logps/rejected": -176.58261108398438, | |
| "loss": 0.7058, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.39593279361724854, | |
| "rewards/margins": 0.025891020894050598, | |
| "rewards/rejected": -0.42182376980781555, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.800947749810224e-05, | |
| "logits/chosen": -2.065213680267334, | |
| "logits/rejected": -2.153409719467163, | |
| "logps/chosen": -140.4229278564453, | |
| "logps/rejected": -171.43482971191406, | |
| "loss": 0.608, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.49880582094192505, | |
| "rewards/margins": 0.2511264681816101, | |
| "rewards/rejected": -0.7499323487281799, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.7973945290505766e-05, | |
| "logits/chosen": -2.0265920162200928, | |
| "logits/rejected": -2.0881428718566895, | |
| "logps/chosen": -156.8954315185547, | |
| "logps/rejected": -243.28848266601562, | |
| "loss": 0.6198, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4524546265602112, | |
| "rewards/margins": 0.26209649443626404, | |
| "rewards/rejected": -0.7145511507987976, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.793811213217766e-05, | |
| "logits/chosen": -2.080693483352661, | |
| "logits/rejected": -2.118926525115967, | |
| "logps/chosen": -231.95574951171875, | |
| "logps/rejected": -232.67112731933594, | |
| "loss": 0.7117, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6412327885627747, | |
| "rewards/margins": 0.06916885077953339, | |
| "rewards/rejected": -0.7104016542434692, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.7901978492519894e-05, | |
| "logits/chosen": -2.065901279449463, | |
| "logits/rejected": -2.1083710193634033, | |
| "logps/chosen": -179.43934631347656, | |
| "logps/rejected": -177.6961669921875, | |
| "loss": 0.6881, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.45454099774360657, | |
| "rewards/margins": 0.08787879347801208, | |
| "rewards/rejected": -0.5424197912216187, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.786554484487064e-05, | |
| "logits/chosen": -1.9663487672805786, | |
| "logits/rejected": -2.065662145614624, | |
| "logps/chosen": -176.4106903076172, | |
| "logps/rejected": -197.2572021484375, | |
| "loss": 0.6107, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5398190021514893, | |
| "rewards/margins": 0.34246671199798584, | |
| "rewards/rejected": -0.8822857141494751, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.782881166649808e-05, | |
| "logits/chosen": -1.9990085363388062, | |
| "logits/rejected": -2.060960292816162, | |
| "logps/chosen": -176.51898193359375, | |
| "logps/rejected": -213.68795776367188, | |
| "loss": 0.5207, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.34073397517204285, | |
| "rewards/margins": 0.4456251859664917, | |
| "rewards/rejected": -0.7863591909408569, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.779177943859413e-05, | |
| "logits/chosen": -1.836498737335205, | |
| "logits/rejected": -1.9420878887176514, | |
| "logps/chosen": -132.9555206298828, | |
| "logps/rejected": -164.89749145507812, | |
| "loss": 0.6992, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.548711895942688, | |
| "rewards/margins": 0.05469922721385956, | |
| "rewards/rejected": -0.6034111976623535, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.775444864626816e-05, | |
| "logits/chosen": -1.9487261772155762, | |
| "logits/rejected": -1.9594744443893433, | |
| "logps/chosen": -158.07406616210938, | |
| "logps/rejected": -211.28558349609375, | |
| "loss": 0.687, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5780103206634521, | |
| "rewards/margins": 0.34122025966644287, | |
| "rewards/rejected": -0.9192306995391846, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7716819778540625e-05, | |
| "logits/chosen": -1.9387106895446777, | |
| "logits/rejected": -1.9535552263259888, | |
| "logps/chosen": -143.24542236328125, | |
| "logps/rejected": -152.4668731689453, | |
| "loss": 0.7511, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.40027323365211487, | |
| "rewards/margins": -0.075428806245327, | |
| "rewards/rejected": -0.32484444975852966, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.767889332833667e-05, | |
| "logits/chosen": -1.8156574964523315, | |
| "logits/rejected": -1.9096050262451172, | |
| "logps/chosen": -166.4335174560547, | |
| "logps/rejected": -194.8555450439453, | |
| "loss": 0.6308, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6348475217819214, | |
| "rewards/margins": 0.36293676495552063, | |
| "rewards/rejected": -0.9977841973304749, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7640669792479676e-05, | |
| "logits/chosen": -2.028461456298828, | |
| "logits/rejected": -2.034698247909546, | |
| "logps/chosen": -197.6614990234375, | |
| "logps/rejected": -226.934326171875, | |
| "loss": 0.6171, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.7803946137428284, | |
| "rewards/margins": 0.23060818016529083, | |
| "rewards/rejected": -1.011002779006958, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.760214967168472e-05, | |
| "logits/chosen": -2.044107675552368, | |
| "logits/rejected": -2.056087017059326, | |
| "logps/chosen": -140.0303497314453, | |
| "logps/rejected": -155.9050750732422, | |
| "loss": 0.7482, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6817924976348877, | |
| "rewards/margins": -0.060957781970500946, | |
| "rewards/rejected": -0.6208346486091614, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.756333347055205e-05, | |
| "logits/chosen": -1.857839584350586, | |
| "logits/rejected": -1.8456567525863647, | |
| "logps/chosen": -159.89170837402344, | |
| "logps/rejected": -171.13980102539062, | |
| "loss": 0.6245, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5193518400192261, | |
| "rewards/margins": 0.17699211835861206, | |
| "rewards/rejected": -0.6963439583778381, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.752422169756048e-05, | |
| "logits/chosen": -1.9076268672943115, | |
| "logits/rejected": -2.0043368339538574, | |
| "logps/chosen": -181.83602905273438, | |
| "logps/rejected": -205.12680053710938, | |
| "loss": 0.769, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.8721389770507812, | |
| "rewards/margins": -0.03039225935935974, | |
| "rewards/rejected": -0.8417466878890991, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.748481486506069e-05, | |
| "logits/chosen": -1.901461124420166, | |
| "logits/rejected": -1.8991363048553467, | |
| "logps/chosen": -158.75704956054688, | |
| "logps/rejected": -162.52723693847656, | |
| "loss": 0.6886, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5035265684127808, | |
| "rewards/margins": 0.1059393435716629, | |
| "rewards/rejected": -0.6094658374786377, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7445113489268544e-05, | |
| "logits/chosen": -1.9314161539077759, | |
| "logits/rejected": -1.937114953994751, | |
| "logps/chosen": -175.22569274902344, | |
| "logps/rejected": -174.74594116210938, | |
| "loss": 1.0026, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -1.054964542388916, | |
| "rewards/margins": -0.28922876715660095, | |
| "rewards/rejected": -0.7657358050346375, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.740511809025833e-05, | |
| "logits/chosen": -2.0365476608276367, | |
| "logits/rejected": -2.0497970581054688, | |
| "logps/chosen": -190.67352294921875, | |
| "logps/rejected": -173.48321533203125, | |
| "loss": 0.7697, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.8780521750450134, | |
| "rewards/margins": -0.05846693366765976, | |
| "rewards/rejected": -0.8195853233337402, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.736482919195593e-05, | |
| "logits/chosen": -1.993350863456726, | |
| "logits/rejected": -1.9442243576049805, | |
| "logps/chosen": -165.4877166748047, | |
| "logps/rejected": -164.430419921875, | |
| "loss": 0.7764, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.5916625261306763, | |
| "rewards/margins": -0.10749640315771103, | |
| "rewards/rejected": -0.4841661751270294, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7324247322131955e-05, | |
| "logits/chosen": -2.0168042182922363, | |
| "logits/rejected": -2.0584840774536133, | |
| "logps/chosen": -167.2913818359375, | |
| "logps/rejected": -182.6678009033203, | |
| "loss": 0.6186, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.4514075219631195, | |
| "rewards/margins": 0.2112567126750946, | |
| "rewards/rejected": -0.6626641750335693, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.728337301239487e-05, | |
| "logits/chosen": -1.883742094039917, | |
| "logits/rejected": -1.867385745048523, | |
| "logps/chosen": -149.0640411376953, | |
| "logps/rejected": -166.1042938232422, | |
| "loss": 0.6005, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.46507394313812256, | |
| "rewards/margins": 0.2587311267852783, | |
| "rewards/rejected": -0.7238050699234009, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7242206798183984e-05, | |
| "logits/chosen": -2.13627290725708, | |
| "logits/rejected": -2.103013753890991, | |
| "logps/chosen": -153.4795684814453, | |
| "logps/rejected": -149.9476776123047, | |
| "loss": 0.742, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5957508683204651, | |
| "rewards/margins": -0.060273557901382446, | |
| "rewards/rejected": -0.5354772806167603, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.720074921876245e-05, | |
| "logits/chosen": -2.079002618789673, | |
| "logits/rejected": -2.1172053813934326, | |
| "logps/chosen": -145.7030029296875, | |
| "logps/rejected": -146.62916564941406, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.4235500395298004, | |
| "rewards/margins": 0.110198013484478, | |
| "rewards/rejected": -0.5337480306625366, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7159000817210205e-05, | |
| "logits/chosen": -1.8756773471832275, | |
| "logits/rejected": -1.8945609331130981, | |
| "logps/chosen": -157.93408203125, | |
| "logps/rejected": -182.92086791992188, | |
| "loss": 0.7563, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.7129658460617065, | |
| "rewards/margins": 0.0887984186410904, | |
| "rewards/rejected": -0.8017643094062805, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.711696214041687e-05, | |
| "logits/chosen": -2.0096912384033203, | |
| "logits/rejected": -1.9164586067199707, | |
| "logps/chosen": -159.55609130859375, | |
| "logps/rejected": -170.66354370117188, | |
| "loss": 0.5354, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2930709719657898, | |
| "rewards/margins": 0.37744176387786865, | |
| "rewards/rejected": -0.6705127358436584, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7074633739074555e-05, | |
| "logits/chosen": -2.0247998237609863, | |
| "logits/rejected": -1.9434438943862915, | |
| "logps/chosen": -181.0531768798828, | |
| "logps/rejected": -178.23292541503906, | |
| "loss": 0.7559, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6423527002334595, | |
| "rewards/margins": -0.11419406533241272, | |
| "rewards/rejected": -0.5281585454940796, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.703201616767067e-05, | |
| "logits/chosen": -1.8810406923294067, | |
| "logits/rejected": -1.9405734539031982, | |
| "logps/chosen": -183.52894592285156, | |
| "logps/rejected": -198.0996551513672, | |
| "loss": 0.6826, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6363824605941772, | |
| "rewards/margins": 0.13868343830108643, | |
| "rewards/rejected": -0.775066077709198, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.6989109984480636e-05, | |
| "logits/chosen": -2.1613376140594482, | |
| "logits/rejected": -2.1215760707855225, | |
| "logps/chosen": -142.79348754882812, | |
| "logps/rejected": -137.41757202148438, | |
| "loss": 0.7831, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6774702668190002, | |
| "rewards/margins": -0.126045823097229, | |
| "rewards/rejected": -0.5514243841171265, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.694591575156061e-05, | |
| "logits/chosen": -2.0644783973693848, | |
| "logits/rejected": -2.104767322540283, | |
| "logps/chosen": -181.3145751953125, | |
| "logps/rejected": -186.40603637695312, | |
| "loss": 0.7309, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5418195128440857, | |
| "rewards/margins": 0.024057641625404358, | |
| "rewards/rejected": -0.5658771991729736, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.690243403474011e-05, | |
| "logits/chosen": -2.0765113830566406, | |
| "logits/rejected": -2.0862977504730225, | |
| "logps/chosen": -152.10926818847656, | |
| "logps/rejected": -182.60452270507812, | |
| "loss": 0.779, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5457407236099243, | |
| "rewards/margins": -0.06785885989665985, | |
| "rewards/rejected": -0.4778818190097809, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.685866540361456e-05, | |
| "logits/chosen": -2.1040709018707275, | |
| "logits/rejected": -2.1037817001342773, | |
| "logps/chosen": -171.85928344726562, | |
| "logps/rejected": -164.0089569091797, | |
| "loss": 0.8034, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6991878747940063, | |
| "rewards/margins": -0.15247105062007904, | |
| "rewards/rejected": -0.5467169284820557, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6814610431537874e-05, | |
| "logits/chosen": -2.147369146347046, | |
| "logits/rejected": -2.2208034992218018, | |
| "logps/chosen": -174.16061401367188, | |
| "logps/rejected": -175.13421630859375, | |
| "loss": 0.82, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6368678212165833, | |
| "rewards/margins": -0.142391175031662, | |
| "rewards/rejected": -0.49447667598724365, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.677026969561494e-05, | |
| "logits/chosen": -2.1030144691467285, | |
| "logits/rejected": -2.033338785171509, | |
| "logps/chosen": -164.89552307128906, | |
| "logps/rejected": -149.99032592773438, | |
| "loss": 0.8334, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.5769500136375427, | |
| "rewards/margins": -0.17100407183170319, | |
| "rewards/rejected": -0.4059459865093231, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6725643776694074e-05, | |
| "logits/chosen": -1.911732792854309, | |
| "logits/rejected": -1.9670432806015015, | |
| "logps/chosen": -132.4580535888672, | |
| "logps/rejected": -138.24024963378906, | |
| "loss": 0.7212, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6518416404724121, | |
| "rewards/margins": -0.02873155102133751, | |
| "rewards/rejected": -0.6231100559234619, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6680733259359346e-05, | |
| "logits/chosen": -2.0536134243011475, | |
| "logits/rejected": -2.1150963306427, | |
| "logps/chosen": -162.5740203857422, | |
| "logps/rejected": -175.80575561523438, | |
| "loss": 0.8083, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6548844575881958, | |
| "rewards/margins": -0.13533146679401398, | |
| "rewards/rejected": -0.519553005695343, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.663553873192299e-05, | |
| "logits/chosen": -2.2090659141540527, | |
| "logits/rejected": -2.1253957748413086, | |
| "logps/chosen": -180.00253295898438, | |
| "logps/rejected": -181.1770782470703, | |
| "loss": 0.7074, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5276303291320801, | |
| "rewards/margins": 0.037765368819236755, | |
| "rewards/rejected": -0.565395712852478, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.659006078641767e-05, | |
| "logits/chosen": -1.9388926029205322, | |
| "logits/rejected": -1.9008660316467285, | |
| "logps/chosen": -164.9468536376953, | |
| "logps/rejected": -165.34970092773438, | |
| "loss": 0.6209, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.4699169993400574, | |
| "rewards/margins": 0.2034609019756317, | |
| "rewards/rejected": -0.6733778715133667, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.654430001858874e-05, | |
| "logits/chosen": -2.0428528785705566, | |
| "logits/rejected": -2.1391069889068604, | |
| "logps/chosen": -133.9488525390625, | |
| "logps/rejected": -146.6528778076172, | |
| "loss": 0.7032, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4486979842185974, | |
| "rewards/margins": 0.06790214776992798, | |
| "rewards/rejected": -0.5166001319885254, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.649825702788643e-05, | |
| "logits/chosen": -2.1632699966430664, | |
| "logits/rejected": -2.0355403423309326, | |
| "logps/chosen": -191.16921997070312, | |
| "logps/rejected": -155.1634521484375, | |
| "loss": 0.7315, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.4651705324649811, | |
| "rewards/margins": 0.024073848500847816, | |
| "rewards/rejected": -0.48924440145492554, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.6451932417457954e-05, | |
| "logits/chosen": -1.9496774673461914, | |
| "logits/rejected": -1.967451810836792, | |
| "logps/chosen": -186.154541015625, | |
| "logps/rejected": -201.9739532470703, | |
| "loss": 0.7148, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.36692532896995544, | |
| "rewards/margins": 0.010707437992095947, | |
| "rewards/rejected": -0.37763282656669617, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.6405326794139696e-05, | |
| "logits/chosen": -2.1893205642700195, | |
| "logits/rejected": -2.2262985706329346, | |
| "logps/chosen": -172.8998565673828, | |
| "logps/rejected": -186.2823028564453, | |
| "loss": 0.7028, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.33371686935424805, | |
| "rewards/margins": 0.029611539095640182, | |
| "rewards/rejected": -0.36332839727401733, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.635844076844919e-05, | |
| "logits/chosen": -2.1050913333892822, | |
| "logits/rejected": -2.1889383792877197, | |
| "logps/chosen": -139.86866760253906, | |
| "logps/rejected": -153.3116912841797, | |
| "loss": 0.8084, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.4279494285583496, | |
| "rewards/margins": -0.18332254886627197, | |
| "rewards/rejected": -0.24462684988975525, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.631127495457713e-05, | |
| "logits/chosen": -2.091176986694336, | |
| "logits/rejected": -2.204737663269043, | |
| "logps/chosen": -158.55857849121094, | |
| "logps/rejected": -183.6383056640625, | |
| "loss": 0.7433, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.34165453910827637, | |
| "rewards/margins": -0.07564406096935272, | |
| "rewards/rejected": -0.26601043343544006, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.626382997037938e-05, | |
| "logits/chosen": -2.1055097579956055, | |
| "logits/rejected": -2.048053503036499, | |
| "logps/chosen": -148.89772033691406, | |
| "logps/rejected": -172.92184448242188, | |
| "loss": 0.7009, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.28506991267204285, | |
| "rewards/margins": 0.059296898543834686, | |
| "rewards/rejected": -0.3443668484687805, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.621610643736878e-05, | |
| "logits/chosen": -1.9385061264038086, | |
| "logits/rejected": -2.0045883655548096, | |
| "logps/chosen": -179.5167999267578, | |
| "logps/rejected": -193.2892608642578, | |
| "loss": 0.7123, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3521873950958252, | |
| "rewards/margins": 0.049636948853731155, | |
| "rewards/rejected": -0.40182438492774963, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.6168104980707107e-05, | |
| "logits/chosen": -1.8955321311950684, | |
| "logits/rejected": -1.9122228622436523, | |
| "logps/chosen": -175.34242248535156, | |
| "logps/rejected": -181.35226440429688, | |
| "loss": 0.6757, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4020342230796814, | |
| "rewards/margins": 0.19264966249465942, | |
| "rewards/rejected": -0.5946838855743408, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.611982622919683e-05, | |
| "logits/chosen": -2.208705425262451, | |
| "logits/rejected": -2.1736044883728027, | |
| "logps/chosen": -205.228759765625, | |
| "logps/rejected": -206.70281982421875, | |
| "loss": 0.6951, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.40116265416145325, | |
| "rewards/margins": 0.1921374797821045, | |
| "rewards/rejected": -0.5933001637458801, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.6071270815272896e-05, | |
| "logits/chosen": -2.0473556518554688, | |
| "logits/rejected": -2.03924560546875, | |
| "logps/chosen": -125.71442413330078, | |
| "logps/rejected": -181.73805236816406, | |
| "loss": 0.7, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3108467757701874, | |
| "rewards/margins": 0.043492816388607025, | |
| "rewards/rejected": -0.3543395698070526, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.6022439374994396e-05, | |
| "logits/chosen": -2.084061622619629, | |
| "logits/rejected": -2.1075713634490967, | |
| "logps/chosen": -170.00535583496094, | |
| "logps/rejected": -173.09616088867188, | |
| "loss": 0.6658, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.33162206411361694, | |
| "rewards/margins": 0.09093569219112396, | |
| "rewards/rejected": -0.4225577712059021, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.597333254803632e-05, | |
| "logits/chosen": -2.064448833465576, | |
| "logits/rejected": -2.136857509613037, | |
| "logps/chosen": -149.3865966796875, | |
| "logps/rejected": -189.15673828125, | |
| "loss": 0.8085, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.4901013970375061, | |
| "rewards/margins": -0.17271549999713898, | |
| "rewards/rejected": -0.31738588213920593, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5923950977681084e-05, | |
| "logits/chosen": -2.1030938625335693, | |
| "logits/rejected": -1.9867299795150757, | |
| "logps/chosen": -187.54412841796875, | |
| "logps/rejected": -161.7672576904297, | |
| "loss": 0.8052, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.48761826753616333, | |
| "rewards/margins": -0.15502014756202698, | |
| "rewards/rejected": -0.33259811997413635, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.587429531081019e-05, | |
| "logits/chosen": -2.0394513607025146, | |
| "logits/rejected": -2.0466253757476807, | |
| "logps/chosen": -171.7015380859375, | |
| "logps/rejected": -159.35626220703125, | |
| "loss": 0.7025, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3231517970561981, | |
| "rewards/margins": -0.0045061372220516205, | |
| "rewards/rejected": -0.3186456859111786, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.582436619789566e-05, | |
| "logits/chosen": -2.129854440689087, | |
| "logits/rejected": -2.1491200923919678, | |
| "logps/chosen": -155.8500518798828, | |
| "logps/rejected": -166.18692016601562, | |
| "loss": 0.6918, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.2755592167377472, | |
| "rewards/margins": 0.11373557150363922, | |
| "rewards/rejected": -0.3892948031425476, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5774164292991625e-05, | |
| "logits/chosen": -2.1740763187408447, | |
| "logits/rejected": -2.259312629699707, | |
| "logps/chosen": -147.3941192626953, | |
| "logps/rejected": -179.2747344970703, | |
| "loss": 0.6813, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2871443033218384, | |
| "rewards/margins": 0.07601044327020645, | |
| "rewards/rejected": -0.3631547689437866, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.572369025372564e-05, | |
| "logits/chosen": -2.1911587715148926, | |
| "logits/rejected": -2.2559661865234375, | |
| "logps/chosen": -163.22071838378906, | |
| "logps/rejected": -169.61575317382812, | |
| "loss": 0.7414, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.521248996257782, | |
| "rewards/margins": -0.01645784080028534, | |
| "rewards/rejected": -0.5047911405563354, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.567294474129015e-05, | |
| "logits/chosen": -2.167720317840576, | |
| "logits/rejected": -2.178614377975464, | |
| "logps/chosen": -150.54217529296875, | |
| "logps/rejected": -159.95115661621094, | |
| "loss": 0.6824, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4768732488155365, | |
| "rewards/margins": 0.061973560601472855, | |
| "rewards/rejected": -0.5388468503952026, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.562192842043381e-05, | |
| "logits/chosen": -2.1215293407440186, | |
| "logits/rejected": -2.142622709274292, | |
| "logps/chosen": -142.4115447998047, | |
| "logps/rejected": -140.78851318359375, | |
| "loss": 0.6628, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.26923662424087524, | |
| "rewards/margins": 0.1439426690340042, | |
| "rewards/rejected": -0.41317927837371826, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.557064195945277e-05, | |
| "logits/chosen": -2.102473020553589, | |
| "logits/rejected": -2.272836446762085, | |
| "logps/chosen": -149.73037719726562, | |
| "logps/rejected": -188.44985961914062, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3324505388736725, | |
| "rewards/margins": 0.1215570718050003, | |
| "rewards/rejected": -0.4540075957775116, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.551908603018191e-05, | |
| "logits/chosen": -2.078631639480591, | |
| "logits/rejected": -2.0118956565856934, | |
| "logps/chosen": -167.2322998046875, | |
| "logps/rejected": -153.5926971435547, | |
| "loss": 0.7748, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3340409994125366, | |
| "rewards/margins": -0.08145566284656525, | |
| "rewards/rejected": -0.25258535146713257, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.546726130798606e-05, | |
| "logits/chosen": -1.9926340579986572, | |
| "logits/rejected": -1.9975799322128296, | |
| "logps/chosen": -153.8031463623047, | |
| "logps/rejected": -149.59437561035156, | |
| "loss": 0.7209, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.39678266644477844, | |
| "rewards/margins": -0.010065795853734016, | |
| "rewards/rejected": -0.38671693205833435, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.541516847175115e-05, | |
| "logits/chosen": -2.0895304679870605, | |
| "logits/rejected": -2.076495885848999, | |
| "logps/chosen": -145.52989196777344, | |
| "logps/rejected": -170.4242706298828, | |
| "loss": 0.7901, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.47353363037109375, | |
| "rewards/margins": -0.07409767806529999, | |
| "rewards/rejected": -0.39943596720695496, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5362808203875295e-05, | |
| "logits/chosen": -2.138859510421753, | |
| "logits/rejected": -2.0678532123565674, | |
| "logps/chosen": -181.20169067382812, | |
| "logps/rejected": -198.79055786132812, | |
| "loss": 0.7287, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5221874713897705, | |
| "rewards/margins": -0.01457640528678894, | |
| "rewards/rejected": -0.507611095905304, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.531018119025989e-05, | |
| "logits/chosen": -2.218453884124756, | |
| "logits/rejected": -2.1501731872558594, | |
| "logps/chosen": -159.3086395263672, | |
| "logps/rejected": -155.74441528320312, | |
| "loss": 0.7366, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.4079001545906067, | |
| "rewards/margins": -0.05238659307360649, | |
| "rewards/rejected": -0.3555135428905487, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.525728812030059e-05, | |
| "logits/chosen": -2.0979692935943604, | |
| "logits/rejected": -2.132126808166504, | |
| "logps/chosen": -169.3253936767578, | |
| "logps/rejected": -174.92539978027344, | |
| "loss": 0.6075, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.30461451411247253, | |
| "rewards/margins": 0.24856308102607727, | |
| "rewards/rejected": -0.5531775951385498, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.520412968687832e-05, | |
| "logits/chosen": -2.1177079677581787, | |
| "logits/rejected": -2.1072580814361572, | |
| "logps/chosen": -169.1343231201172, | |
| "logps/rejected": -179.53546142578125, | |
| "loss": 0.8153, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6010310053825378, | |
| "rewards/margins": -0.13811662793159485, | |
| "rewards/rejected": -0.4629143476486206, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.515070658635013e-05, | |
| "logits/chosen": -2.2024600505828857, | |
| "logits/rejected": -2.2220730781555176, | |
| "logps/chosen": -136.5756378173828, | |
| "logps/rejected": -145.7661590576172, | |
| "loss": 0.7758, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5926961302757263, | |
| "rewards/margins": -0.08495937287807465, | |
| "rewards/rejected": -0.5077367424964905, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.509701951854017e-05, | |
| "logits/chosen": -2.210057258605957, | |
| "logits/rejected": -2.1656928062438965, | |
| "logps/chosen": -172.19387817382812, | |
| "logps/rejected": -177.2809600830078, | |
| "loss": 0.7218, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.4175167679786682, | |
| "rewards/margins": 0.030983783304691315, | |
| "rewards/rejected": -0.4485005736351013, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.504306918673044e-05, | |
| "logits/chosen": -2.147188186645508, | |
| "logits/rejected": -2.2114806175231934, | |
| "logps/chosen": -179.85704040527344, | |
| "logps/rejected": -196.09005737304688, | |
| "loss": 0.6706, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.4910237491130829, | |
| "rewards/margins": 0.14421552419662476, | |
| "rewards/rejected": -0.63523930311203, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.498885629765162e-05, | |
| "logits/chosen": -2.2785000801086426, | |
| "logits/rejected": -2.205921173095703, | |
| "logps/chosen": -221.86685180664062, | |
| "logps/rejected": -187.08416748046875, | |
| "loss": 0.7321, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.4971649944782257, | |
| "rewards/margins": -0.01577048748731613, | |
| "rewards/rejected": -0.4813944697380066, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.4934381561473776e-05, | |
| "logits/chosen": -2.21511173248291, | |
| "logits/rejected": -2.194715976715088, | |
| "logps/chosen": -152.30589294433594, | |
| "logps/rejected": -134.73757934570312, | |
| "loss": 0.7943, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.44001439213752747, | |
| "rewards/margins": -0.11678460240364075, | |
| "rewards/rejected": -0.3232297897338867, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.487964569179711e-05, | |
| "logits/chosen": -2.1763601303100586, | |
| "logits/rejected": -2.206886053085327, | |
| "logps/chosen": -201.2484130859375, | |
| "logps/rejected": -186.90383911132812, | |
| "loss": 0.7077, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.4743629992008209, | |
| "rewards/margins": 0.08780473470687866, | |
| "rewards/rejected": -0.562167763710022, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.482464940564257e-05, | |
| "logits/chosen": -2.2177324295043945, | |
| "logits/rejected": -2.2520463466644287, | |
| "logps/chosen": -187.29617309570312, | |
| "logps/rejected": -183.0156707763672, | |
| "loss": 0.7544, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5175611972808838, | |
| "rewards/margins": -0.00866784155368805, | |
| "rewards/rejected": -0.5088933706283569, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.476939342344246e-05, | |
| "logits/chosen": -2.197991132736206, | |
| "logits/rejected": -2.186018943786621, | |
| "logps/chosen": -150.70626831054688, | |
| "logps/rejected": -135.6775360107422, | |
| "loss": 0.6946, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.15773512423038483, | |
| "rewards/margins": 0.024013396352529526, | |
| "rewards/rejected": -0.18174852430820465, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.471387846903104e-05, | |
| "logits/chosen": -2.1710331439971924, | |
| "logits/rejected": -2.237042188644409, | |
| "logps/chosen": -170.97064208984375, | |
| "logps/rejected": -193.65908813476562, | |
| "loss": 0.7191, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.519289493560791, | |
| "rewards/margins": -0.03059907630085945, | |
| "rewards/rejected": -0.48869046568870544, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.465810526963499e-05, | |
| "logits/chosen": -2.1213455200195312, | |
| "logits/rejected": -2.214155435562134, | |
| "logps/chosen": -142.11195373535156, | |
| "logps/rejected": -176.63465881347656, | |
| "loss": 0.8066, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4935774803161621, | |
| "rewards/margins": -0.11702725291252136, | |
| "rewards/rejected": -0.37655022740364075, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.46020745558639e-05, | |
| "logits/chosen": -2.1311259269714355, | |
| "logits/rejected": -2.1578001976013184, | |
| "logps/chosen": -146.28683471679688, | |
| "logps/rejected": -155.28411865234375, | |
| "loss": 0.773, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6565503478050232, | |
| "rewards/margins": -0.07203666865825653, | |
| "rewards/rejected": -0.5845136046409607, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.454578706170075e-05, | |
| "logits/chosen": -2.3410611152648926, | |
| "logits/rejected": -2.2815985679626465, | |
| "logps/chosen": -167.60433959960938, | |
| "logps/rejected": -166.52891540527344, | |
| "loss": 0.6854, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.4833534061908722, | |
| "rewards/margins": 0.048468299210071564, | |
| "rewards/rejected": -0.5318217277526855, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.448924352449222e-05, | |
| "logits/chosen": -2.137972354888916, | |
| "logits/rejected": -2.164095640182495, | |
| "logps/chosen": -156.0537109375, | |
| "logps/rejected": -171.20938110351562, | |
| "loss": 0.6641, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.42611458897590637, | |
| "rewards/margins": 0.09905887395143509, | |
| "rewards/rejected": -0.5251734256744385, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4432444684939077e-05, | |
| "logits/chosen": -2.206866502761841, | |
| "logits/rejected": -2.218209743499756, | |
| "logps/chosen": -154.92098999023438, | |
| "logps/rejected": -160.64996337890625, | |
| "loss": 0.7018, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.36239269375801086, | |
| "rewards/margins": 0.08214544504880905, | |
| "rewards/rejected": -0.4445381760597229, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.437539128708647e-05, | |
| "logits/chosen": -2.057393789291382, | |
| "logits/rejected": -1.9958287477493286, | |
| "logps/chosen": -154.2297821044922, | |
| "logps/rejected": -139.65081787109375, | |
| "loss": 0.864, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6004846096038818, | |
| "rewards/margins": -0.2311324030160904, | |
| "rewards/rejected": -0.36935219168663025, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.431808407831416e-05, | |
| "logits/chosen": -2.1149046421051025, | |
| "logits/rejected": -2.217886447906494, | |
| "logps/chosen": -147.158203125, | |
| "logps/rejected": -189.3766632080078, | |
| "loss": 0.6049, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.16423264145851135, | |
| "rewards/margins": 0.25643211603164673, | |
| "rewards/rejected": -0.4206647276878357, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.426052380932674e-05, | |
| "logits/chosen": -2.2805943489074707, | |
| "logits/rejected": -2.180037021636963, | |
| "logps/chosen": -166.3658447265625, | |
| "logps/rejected": -154.8529052734375, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5027453899383545, | |
| "rewards/margins": 0.10051669925451279, | |
| "rewards/rejected": -0.6032620668411255, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.420271123414381e-05, | |
| "logits/chosen": -2.244375705718994, | |
| "logits/rejected": -2.2612979412078857, | |
| "logps/chosen": -170.18692016601562, | |
| "logps/rejected": -177.5980987548828, | |
| "loss": 0.6506, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3367692530155182, | |
| "rewards/margins": 0.11379708349704742, | |
| "rewards/rejected": -0.45056629180908203, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4144647110090105e-05, | |
| "logits/chosen": -2.1918513774871826, | |
| "logits/rejected": -2.187420129776001, | |
| "logps/chosen": -171.70425415039062, | |
| "logps/rejected": -162.43191528320312, | |
| "loss": 0.7087, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3047904968261719, | |
| "rewards/margins": 0.06497037410736084, | |
| "rewards/rejected": -0.3697608709335327, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.408633219778555e-05, | |
| "logits/chosen": -2.2267677783966064, | |
| "logits/rejected": -2.2612836360931396, | |
| "logps/chosen": -211.01437377929688, | |
| "logps/rejected": -208.23724365234375, | |
| "loss": 0.5474, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2352912724018097, | |
| "rewards/margins": 0.4508543908596039, | |
| "rewards/rejected": -0.6861456632614136, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.40277672611353e-05, | |
| "logits/chosen": -2.3318138122558594, | |
| "logits/rejected": -2.332906484603882, | |
| "logps/chosen": -185.47195434570312, | |
| "logps/rejected": -174.6714630126953, | |
| "loss": 0.6949, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4576830565929413, | |
| "rewards/margins": 0.10692331194877625, | |
| "rewards/rejected": -0.5646063685417175, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3968953067319777e-05, | |
| "logits/chosen": -2.231781482696533, | |
| "logits/rejected": -2.320117950439453, | |
| "logps/chosen": -142.7481231689453, | |
| "logps/rejected": -169.84375, | |
| "loss": 0.6559, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3941573202610016, | |
| "rewards/margins": 0.2276144176721573, | |
| "rewards/rejected": -0.6217717528343201, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.390989038678455e-05, | |
| "logits/chosen": -2.1980667114257812, | |
| "logits/rejected": -2.0504379272460938, | |
| "logps/chosen": -168.10940551757812, | |
| "logps/rejected": -148.44493103027344, | |
| "loss": 0.7818, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.5723888874053955, | |
| "rewards/margins": -0.1398972123861313, | |
| "rewards/rejected": -0.432491660118103, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3850579993230284e-05, | |
| "logits/chosen": -2.2534234523773193, | |
| "logits/rejected": -2.3104054927825928, | |
| "logps/chosen": -145.6854705810547, | |
| "logps/rejected": -152.75489807128906, | |
| "loss": 0.6966, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3927219808101654, | |
| "rewards/margins": 0.09541331231594086, | |
| "rewards/rejected": -0.48813530802726746, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3791022663602624e-05, | |
| "logits/chosen": -2.245393753051758, | |
| "logits/rejected": -2.2723946571350098, | |
| "logps/chosen": -208.752197265625, | |
| "logps/rejected": -239.25466918945312, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.41935253143310547, | |
| "rewards/margins": 0.23289479315280914, | |
| "rewards/rejected": -0.6522473096847534, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.373121917808196e-05, | |
| "logits/chosen": -2.100860118865967, | |
| "logits/rejected": -2.11472749710083, | |
| "logps/chosen": -172.28466796875, | |
| "logps/rejected": -183.5985107421875, | |
| "loss": 0.6846, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3080265522003174, | |
| "rewards/margins": 0.04853975400328636, | |
| "rewards/rejected": -0.35656630992889404, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.367117032007326e-05, | |
| "logits/chosen": -2.1881682872772217, | |
| "logits/rejected": -2.277254104614258, | |
| "logps/chosen": -145.00588989257812, | |
| "logps/rejected": -148.37994384765625, | |
| "loss": 0.6124, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.10753998160362244, | |
| "rewards/margins": 0.23135872185230255, | |
| "rewards/rejected": -0.3388986885547638, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.361087687619579e-05, | |
| "logits/chosen": -2.1831393241882324, | |
| "logits/rejected": -2.1932125091552734, | |
| "logps/chosen": -129.8612060546875, | |
| "logps/rejected": -142.00599670410156, | |
| "loss": 0.7372, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.625735878944397, | |
| "rewards/margins": -0.01838909275829792, | |
| "rewards/rejected": -0.607346773147583, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3550339636272775e-05, | |
| "logits/chosen": -2.166675329208374, | |
| "logits/rejected": -2.1578290462493896, | |
| "logps/chosen": -175.37667846679688, | |
| "logps/rejected": -163.28125, | |
| "loss": 0.6116, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.3509731590747833, | |
| "rewards/margins": 0.24484336376190186, | |
| "rewards/rejected": -0.5958165526390076, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.348955939332111e-05, | |
| "logits/chosen": -2.218362808227539, | |
| "logits/rejected": -2.2013816833496094, | |
| "logps/chosen": -144.22523498535156, | |
| "logps/rejected": -136.4739990234375, | |
| "loss": 0.6519, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3839653432369232, | |
| "rewards/margins": 0.1404239982366562, | |
| "rewards/rejected": -0.5243893265724182, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.342853694354095e-05, | |
| "logits/chosen": -2.193958044052124, | |
| "logits/rejected": -2.1357054710388184, | |
| "logps/chosen": -179.131591796875, | |
| "logps/rejected": -181.36126708984375, | |
| "loss": 0.8937, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6919696927070618, | |
| "rewards/margins": -0.27157241106033325, | |
| "rewards/rejected": -0.42039719223976135, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.336727308630527e-05, | |
| "logits/chosen": -2.166112184524536, | |
| "logits/rejected": -2.140867233276367, | |
| "logps/chosen": -157.46937561035156, | |
| "logps/rejected": -139.50148010253906, | |
| "loss": 0.7086, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.425024151802063, | |
| "rewards/margins": 0.010826226323843002, | |
| "rewards/rejected": -0.4358503818511963, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.330576862414938e-05, | |
| "logits/chosen": -2.1364035606384277, | |
| "logits/rejected": -2.20801043510437, | |
| "logps/chosen": -224.9893341064453, | |
| "logps/rejected": -179.57144165039062, | |
| "loss": 0.6006, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.18072500824928284, | |
| "rewards/margins": 0.2619746923446655, | |
| "rewards/rejected": -0.44269970059394836, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.324402436276046e-05, | |
| "logits/chosen": -2.157177209854126, | |
| "logits/rejected": -2.151972532272339, | |
| "logps/chosen": -147.28689575195312, | |
| "logps/rejected": -152.43234252929688, | |
| "loss": 0.7044, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.2011633813381195, | |
| "rewards/margins": 0.055140420794487, | |
| "rewards/rejected": -0.2563037872314453, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.318204111096695e-05, | |
| "logits/chosen": -2.361785650253296, | |
| "logits/rejected": -2.332156181335449, | |
| "logps/chosen": -212.84500122070312, | |
| "logps/rejected": -177.1304931640625, | |
| "loss": 0.7486, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4005717933177948, | |
| "rewards/margins": 0.005475823767483234, | |
| "rewards/rejected": -0.40604764223098755, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.3119819680728e-05, | |
| "logits/chosen": -2.354240894317627, | |
| "logits/rejected": -2.3794326782226562, | |
| "logps/chosen": -181.24334716796875, | |
| "logps/rejected": -200.6251983642578, | |
| "loss": 0.773, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.29395994544029236, | |
| "rewards/margins": 0.008357390761375427, | |
| "rewards/rejected": -0.302317351102829, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.305736088712282e-05, | |
| "logits/chosen": -2.40366792678833, | |
| "logits/rejected": -2.402517557144165, | |
| "logps/chosen": -208.20982360839844, | |
| "logps/rejected": -207.42562866210938, | |
| "loss": 0.7212, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.193616583943367, | |
| "rewards/margins": 0.02203945443034172, | |
| "rewards/rejected": -0.21565604209899902, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.299466554833997e-05, | |
| "logits/chosen": -2.0629615783691406, | |
| "logits/rejected": -2.076362371444702, | |
| "logps/chosen": -155.0874481201172, | |
| "logps/rejected": -167.83026123046875, | |
| "loss": 0.633, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2359430491924286, | |
| "rewards/margins": 0.2387825846672058, | |
| "rewards/rejected": -0.4747256338596344, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.293173448566671e-05, | |
| "logits/chosen": -2.1900341510772705, | |
| "logits/rejected": -2.16131329536438, | |
| "logps/chosen": -186.0865478515625, | |
| "logps/rejected": -185.54696655273438, | |
| "loss": 0.7872, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.6240242123603821, | |
| "rewards/margins": -0.11755514144897461, | |
| "rewards/rejected": -0.5064690113067627, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.286856852347816e-05, | |
| "logits/chosen": -2.2081987857818604, | |
| "logits/rejected": -2.1977622509002686, | |
| "logps/chosen": -157.4396514892578, | |
| "logps/rejected": -154.04432678222656, | |
| "loss": 0.6025, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.4773905277252197, | |
| "rewards/margins": 0.26096710562705994, | |
| "rewards/rejected": -0.738357663154602, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.280516848922658e-05, | |
| "logits/chosen": -2.364104747772217, | |
| "logits/rejected": -2.4083893299102783, | |
| "logps/chosen": -182.33734130859375, | |
| "logps/rejected": -185.8186492919922, | |
| "loss": 0.7932, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6770175099372864, | |
| "rewards/margins": -0.1133224219083786, | |
| "rewards/rejected": -0.5636950731277466, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.274153521343046e-05, | |
| "logits/chosen": -2.376795530319214, | |
| "logits/rejected": -2.4445347785949707, | |
| "logps/chosen": -235.20889282226562, | |
| "logps/rejected": -269.0733642578125, | |
| "loss": 0.7455, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5483633279800415, | |
| "rewards/margins": -0.023924335837364197, | |
| "rewards/rejected": -0.5244389772415161, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.267766952966369e-05, | |
| "logits/chosen": -2.3826839923858643, | |
| "logits/rejected": -2.3101706504821777, | |
| "logps/chosen": -192.46347045898438, | |
| "logps/rejected": -210.77610778808594, | |
| "loss": 0.6805, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.35423946380615234, | |
| "rewards/margins": 0.15307310223579407, | |
| "rewards/rejected": -0.507312536239624, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.261357227454463e-05, | |
| "logits/chosen": -2.3622584342956543, | |
| "logits/rejected": -2.1960906982421875, | |
| "logps/chosen": -177.9891815185547, | |
| "logps/rejected": -182.12896728515625, | |
| "loss": 0.6963, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.30851995944976807, | |
| "rewards/margins": 0.08458848297595978, | |
| "rewards/rejected": -0.39310839772224426, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.2549244287725135e-05, | |
| "logits/chosen": -2.0949957370758057, | |
| "logits/rejected": -2.177522659301758, | |
| "logps/chosen": -169.70445251464844, | |
| "logps/rejected": -186.91510009765625, | |
| "loss": 0.6832, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5025824308395386, | |
| "rewards/margins": 0.13031432032585144, | |
| "rewards/rejected": -0.6328967213630676, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.2484686411879554e-05, | |
| "logits/chosen": -2.325777530670166, | |
| "logits/rejected": -2.363949775695801, | |
| "logps/chosen": -169.78359985351562, | |
| "logps/rejected": -188.50552368164062, | |
| "loss": 0.7328, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.5046086311340332, | |
| "rewards/margins": -0.03166092932224274, | |
| "rewards/rejected": -0.4729476869106293, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.2419899492693737e-05, | |
| "logits/chosen": -2.180312156677246, | |
| "logits/rejected": -2.288280725479126, | |
| "logps/chosen": -145.58860778808594, | |
| "logps/rejected": -173.983154296875, | |
| "loss": 0.6565, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4242013692855835, | |
| "rewards/margins": 0.1269078254699707, | |
| "rewards/rejected": -0.5511091947555542, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.235488437885388e-05, | |
| "logits/chosen": -2.2667553424835205, | |
| "logits/rejected": -2.3572263717651367, | |
| "logps/chosen": -171.51913452148438, | |
| "logps/rejected": -191.5248260498047, | |
| "loss": 0.7448, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4607779383659363, | |
| "rewards/margins": -0.03998526930809021, | |
| "rewards/rejected": -0.42079269886016846, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.2289641922035493e-05, | |
| "logits/chosen": -2.08382248878479, | |
| "logits/rejected": -2.0807366371154785, | |
| "logps/chosen": -168.42034912109375, | |
| "logps/rejected": -163.81625366210938, | |
| "loss": 0.6333, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.27064958214759827, | |
| "rewards/margins": 0.14805974066257477, | |
| "rewards/rejected": -0.41870927810668945, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.222417297689217e-05, | |
| "logits/chosen": -1.9381663799285889, | |
| "logits/rejected": -1.8921332359313965, | |
| "logps/chosen": -223.4558868408203, | |
| "logps/rejected": -210.17938232421875, | |
| "loss": 0.7046, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.16921508312225342, | |
| "rewards/margins": 0.15382614731788635, | |
| "rewards/rejected": -0.32304123044013977, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.215847840104442e-05, | |
| "logits/chosen": -2.1920394897460938, | |
| "logits/rejected": -2.1775074005126953, | |
| "logps/chosen": -159.56520080566406, | |
| "logps/rejected": -157.868408203125, | |
| "loss": 0.7058, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.48819100856781006, | |
| "rewards/margins": 0.040394630283117294, | |
| "rewards/rejected": -0.52858567237854, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.209255905506847e-05, | |
| "logits/chosen": -2.223533868789673, | |
| "logits/rejected": -2.2556004524230957, | |
| "logps/chosen": -168.38034057617188, | |
| "logps/rejected": -159.22509765625, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3159790337085724, | |
| "rewards/margins": 0.07181244343519211, | |
| "rewards/rejected": -0.3877914249897003, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.202641580248492e-05, | |
| "logits/chosen": -2.1462109088897705, | |
| "logits/rejected": -2.161759376525879, | |
| "logps/chosen": -202.07864379882812, | |
| "logps/rejected": -202.52467346191406, | |
| "loss": 0.7194, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.32575148344039917, | |
| "rewards/margins": -0.014372788369655609, | |
| "rewards/rejected": -0.31137868762016296, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.19600495097475e-05, | |
| "logits/chosen": -2.3070526123046875, | |
| "logits/rejected": -2.281818389892578, | |
| "logps/chosen": -215.2132110595703, | |
| "logps/rejected": -219.35874938964844, | |
| "loss": 0.5939, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.18687282502651215, | |
| "rewards/margins": 0.36697521805763245, | |
| "rewards/rejected": -0.5538480877876282, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.1893461046231656e-05, | |
| "logits/chosen": -2.2430295944213867, | |
| "logits/rejected": -2.2637746334075928, | |
| "logps/chosen": -169.20651245117188, | |
| "logps/rejected": -161.84983825683594, | |
| "loss": 0.6557, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.4840455651283264, | |
| "rewards/margins": 0.10702849924564362, | |
| "rewards/rejected": -0.5910740494728088, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.182665128422323e-05, | |
| "logits/chosen": -2.1918537616729736, | |
| "logits/rejected": -2.2522811889648438, | |
| "logps/chosen": -159.4562225341797, | |
| "logps/rejected": -157.55862426757812, | |
| "loss": 0.7837, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.272363543510437, | |
| "rewards/margins": -0.07118403166532516, | |
| "rewards/rejected": -0.20117953419685364, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.175962109890696e-05, | |
| "logits/chosen": -2.2962236404418945, | |
| "logits/rejected": -2.294724941253662, | |
| "logps/chosen": -162.7214813232422, | |
| "logps/rejected": -174.26588439941406, | |
| "loss": 0.6253, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.43033263087272644, | |
| "rewards/margins": 0.2920879125595093, | |
| "rewards/rejected": -0.7224205732345581, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.16923713683551e-05, | |
| "logits/chosen": -2.1335949897766113, | |
| "logits/rejected": -2.2332561016082764, | |
| "logps/chosen": -141.69833374023438, | |
| "logps/rejected": -166.99969482421875, | |
| "loss": 0.6806, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4482802748680115, | |
| "rewards/margins": 0.09964673221111298, | |
| "rewards/rejected": -0.5479270219802856, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.162490297351583e-05, | |
| "logits/chosen": -2.2109334468841553, | |
| "logits/rejected": -2.2407150268554688, | |
| "logps/chosen": -147.48382568359375, | |
| "logps/rejected": -170.04962158203125, | |
| "loss": 0.6788, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4719080328941345, | |
| "rewards/margins": 0.10454200953245163, | |
| "rewards/rejected": -0.5764500498771667, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.155721679820176e-05, | |
| "logits/chosen": -2.0820136070251465, | |
| "logits/rejected": -2.1118240356445312, | |
| "logps/chosen": -147.42869567871094, | |
| "logps/rejected": -147.12875366210938, | |
| "loss": 0.7504, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6050678491592407, | |
| "rewards/margins": -0.05039627104997635, | |
| "rewards/rejected": -0.554671585559845, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.1489313729078376e-05, | |
| "logits/chosen": -2.237750768661499, | |
| "logits/rejected": -2.3360304832458496, | |
| "logps/chosen": -160.38446044921875, | |
| "logps/rejected": -181.29734802246094, | |
| "loss": 0.986, | |
| "rewards/accuracies": 0.1875, | |
| "rewards/chosen": -0.6366126537322998, | |
| "rewards/margins": -0.44120755791664124, | |
| "rewards/rejected": -0.19540515542030334, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.142119465565238e-05, | |
| "logits/chosen": -2.1615211963653564, | |
| "logits/rejected": -2.16357421875, | |
| "logps/chosen": -191.34902954101562, | |
| "logps/rejected": -178.8798370361328, | |
| "loss": 0.7931, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.3513278365135193, | |
| "rewards/margins": -0.09741759300231934, | |
| "rewards/rejected": -0.25391024351119995, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.135286047026005e-05, | |
| "logits/chosen": -2.191861391067505, | |
| "logits/rejected": -2.214906930923462, | |
| "logps/chosen": -182.97705078125, | |
| "logps/rejected": -203.17572021484375, | |
| "loss": 0.678, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4434359073638916, | |
| "rewards/margins": 0.14009566605091095, | |
| "rewards/rejected": -0.5835315585136414, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.128431206805557e-05, | |
| "logits/chosen": -2.156165599822998, | |
| "logits/rejected": -2.091118574142456, | |
| "logps/chosen": -163.7310028076172, | |
| "logps/rejected": -149.5469970703125, | |
| "loss": 0.6163, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.4948202669620514, | |
| "rewards/margins": 0.19163364171981812, | |
| "rewards/rejected": -0.6864539384841919, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.121555034699925e-05, | |
| "logits/chosen": -2.149259090423584, | |
| "logits/rejected": -2.2289958000183105, | |
| "logps/chosen": -142.09974670410156, | |
| "logps/rejected": -162.68006896972656, | |
| "loss": 0.694, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.3429271876811981, | |
| "rewards/margins": 0.07812821865081787, | |
| "rewards/rejected": -0.421055406332016, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.114657620784589e-05, | |
| "logits/chosen": -2.2879371643066406, | |
| "logits/rejected": -2.275233268737793, | |
| "logps/chosen": -194.87741088867188, | |
| "logps/rejected": -187.3885955810547, | |
| "loss": 0.6425, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3526943325996399, | |
| "rewards/margins": 0.17375287413597107, | |
| "rewards/rejected": -0.5264471769332886, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.107739055413281e-05, | |
| "logits/chosen": -2.188884973526001, | |
| "logits/rejected": -2.176175594329834, | |
| "logps/chosen": -166.35443115234375, | |
| "logps/rejected": -163.26513671875, | |
| "loss": 0.7033, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5968694090843201, | |
| "rewards/margins": 0.10982798039913177, | |
| "rewards/rejected": -0.7066974639892578, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.1007994292168126e-05, | |
| "logits/chosen": -2.1411023139953613, | |
| "logits/rejected": -2.1539387702941895, | |
| "logps/chosen": -164.981689453125, | |
| "logps/rejected": -184.5115966796875, | |
| "loss": 0.7428, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.49466997385025024, | |
| "rewards/margins": -0.016588065773248672, | |
| "rewards/rejected": -0.4780818819999695, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938388331018864e-05, | |
| "logits/chosen": -2.3704164028167725, | |
| "logits/rejected": -2.2904226779937744, | |
| "logps/chosen": -180.89866638183594, | |
| "logps/rejected": -163.3619384765625, | |
| "loss": 0.7123, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4977205991744995, | |
| "rewards/margins": 0.030645374208688736, | |
| "rewards/rejected": -0.528365969657898, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0868573582499004e-05, | |
| "logits/chosen": -2.067239761352539, | |
| "logits/rejected": -2.026726484298706, | |
| "logps/chosen": -145.52386474609375, | |
| "logps/rejected": -144.47396850585938, | |
| "loss": 0.7085, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5200850963592529, | |
| "rewards/margins": 0.039269786328077316, | |
| "rewards/rejected": -0.559354841709137, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.07985509611576e-05, | |
| "logits/chosen": -2.2816240787506104, | |
| "logits/rejected": -2.248000383377075, | |
| "logps/chosen": -187.68630981445312, | |
| "logps/rejected": -183.52285766601562, | |
| "loss": 0.7173, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.44892194867134094, | |
| "rewards/margins": 0.04539749026298523, | |
| "rewards/rejected": -0.49431943893432617, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.072832138426676e-05, | |
| "logits/chosen": -2.2360525131225586, | |
| "logits/rejected": -2.209181308746338, | |
| "logps/chosen": -194.4606475830078, | |
| "logps/rejected": -161.04115295410156, | |
| "loss": 0.7037, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5675978064537048, | |
| "rewards/margins": 0.11050501465797424, | |
| "rewards/rejected": -0.6781029105186462, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.065788577180962e-05, | |
| "logits/chosen": -2.2562222480773926, | |
| "logits/rejected": -2.2564706802368164, | |
| "logps/chosen": -156.35443115234375, | |
| "logps/rejected": -172.9788055419922, | |
| "loss": 0.6291, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.5148206949234009, | |
| "rewards/margins": 0.19110004603862762, | |
| "rewards/rejected": -0.7059207558631897, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.058724504646834e-05, | |
| "logits/chosen": -2.3457727432250977, | |
| "logits/rejected": -2.4520750045776367, | |
| "logps/chosen": -165.80174255371094, | |
| "logps/rejected": -189.46731567382812, | |
| "loss": 0.6369, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.51256263256073, | |
| "rewards/margins": 0.1984454244375229, | |
| "rewards/rejected": -0.7110080718994141, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0516400133611964e-05, | |
| "logits/chosen": -2.1804251670837402, | |
| "logits/rejected": -2.1718239784240723, | |
| "logps/chosen": -175.29420471191406, | |
| "logps/rejected": -186.0224151611328, | |
| "loss": 0.6086, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6141149997711182, | |
| "rewards/margins": 0.2933781147003174, | |
| "rewards/rejected": -0.9074931144714355, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0445351961284326e-05, | |
| "logits/chosen": -2.364253044128418, | |
| "logits/rejected": -2.419072151184082, | |
| "logps/chosen": -191.72203063964844, | |
| "logps/rejected": -202.510498046875, | |
| "loss": 0.7427, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6713025569915771, | |
| "rewards/margins": -0.016349561512470245, | |
| "rewards/rejected": -0.6549530029296875, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0374101460191895e-05, | |
| "logits/chosen": -2.266031265258789, | |
| "logits/rejected": -2.2433180809020996, | |
| "logps/chosen": -208.40182495117188, | |
| "logps/rejected": -192.03582763671875, | |
| "loss": 0.7206, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5185582041740417, | |
| "rewards/margins": 0.09906256198883057, | |
| "rewards/rejected": -0.6176207661628723, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.030264956369157e-05, | |
| "logits/chosen": -2.292019844055176, | |
| "logits/rejected": -2.3281755447387695, | |
| "logps/chosen": -158.6434783935547, | |
| "logps/rejected": -171.675048828125, | |
| "loss": 0.6101, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.44501081109046936, | |
| "rewards/margins": 0.2512443959712982, | |
| "rewards/rejected": -0.6962552070617676, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.023099720777848e-05, | |
| "logits/chosen": -2.0625979900360107, | |
| "logits/rejected": -2.226754903793335, | |
| "logps/chosen": -167.65057373046875, | |
| "logps/rejected": -217.35430908203125, | |
| "loss": 0.7485, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6113196611404419, | |
| "rewards/margins": 0.010612770915031433, | |
| "rewards/rejected": -0.6219324469566345, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.015914533107367e-05, | |
| "logits/chosen": -2.1320765018463135, | |
| "logits/rejected": -2.168058156967163, | |
| "logps/chosen": -189.58056640625, | |
| "logps/rejected": -182.50790405273438, | |
| "loss": 0.8201, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.690062403678894, | |
| "rewards/margins": -0.02232969179749489, | |
| "rewards/rejected": -0.6677327156066895, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.008709487481187e-05, | |
| "logits/chosen": -2.1868677139282227, | |
| "logits/rejected": -2.1994471549987793, | |
| "logps/chosen": -201.81927490234375, | |
| "logps/rejected": -206.9180908203125, | |
| "loss": 0.6518, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5198031663894653, | |
| "rewards/margins": 0.3545428216457367, | |
| "rewards/rejected": -0.8743458986282349, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.0014846782829104e-05, | |
| "logits/chosen": -2.3197684288024902, | |
| "logits/rejected": -2.280550956726074, | |
| "logps/chosen": -191.38059997558594, | |
| "logps/rejected": -209.5456085205078, | |
| "loss": 0.6311, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.6963108777999878, | |
| "rewards/margins": 0.4248548448085785, | |
| "rewards/rejected": -1.1211657524108887, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.994240200155038e-05, | |
| "logits/chosen": -2.1500096321105957, | |
| "logits/rejected": -2.1376168727874756, | |
| "logps/chosen": -189.9169158935547, | |
| "logps/rejected": -152.22576904296875, | |
| "loss": 0.8067, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5642895102500916, | |
| "rewards/margins": -0.024822479113936424, | |
| "rewards/rejected": -0.5394670963287354, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9869761479977266e-05, | |
| "logits/chosen": -2.169552803039551, | |
| "logits/rejected": -2.2305593490600586, | |
| "logps/chosen": -152.51605224609375, | |
| "logps/rejected": -177.90081787109375, | |
| "loss": 0.6629, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.39782875776290894, | |
| "rewards/margins": 0.18033528327941895, | |
| "rewards/rejected": -0.5781639814376831, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.979692616967543e-05, | |
| "logits/chosen": -2.137956142425537, | |
| "logits/rejected": -2.1780784130096436, | |
| "logps/chosen": -207.0366668701172, | |
| "logps/rejected": -197.46815490722656, | |
| "loss": 0.7157, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.43855610489845276, | |
| "rewards/margins": 0.017738621681928635, | |
| "rewards/rejected": -0.4562947154045105, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9723897024762255e-05, | |
| "logits/chosen": -1.9541856050491333, | |
| "logits/rejected": -2.100348949432373, | |
| "logps/chosen": -163.6127166748047, | |
| "logps/rejected": -188.64373779296875, | |
| "loss": 0.7803, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6812203526496887, | |
| "rewards/margins": 0.0019090883433818817, | |
| "rewards/rejected": -0.6831294298171997, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.965067500189424e-05, | |
| "logits/chosen": -2.1815788745880127, | |
| "logits/rejected": -2.2175307273864746, | |
| "logps/chosen": -168.3467559814453, | |
| "logps/rejected": -206.26513671875, | |
| "loss": 0.7849, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.44792479276657104, | |
| "rewards/margins": -0.1289525330066681, | |
| "rewards/rejected": -0.31897228956222534, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.957726106025455e-05, | |
| "logits/chosen": -2.1104772090911865, | |
| "logits/rejected": -2.1437559127807617, | |
| "logps/chosen": -156.33883666992188, | |
| "logps/rejected": -177.40576171875, | |
| "loss": 0.6855, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5787633657455444, | |
| "rewards/margins": 0.1103430688381195, | |
| "rewards/rejected": -0.6891064047813416, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.950365616154042e-05, | |
| "logits/chosen": -2.0904276371002197, | |
| "logits/rejected": -1.9981458187103271, | |
| "logps/chosen": -153.2527618408203, | |
| "logps/rejected": -130.3262176513672, | |
| "loss": 0.7175, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.4905881881713867, | |
| "rewards/margins": 0.019158679991960526, | |
| "rewards/rejected": -0.5097469091415405, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.942986126995052e-05, | |
| "logits/chosen": -2.0950582027435303, | |
| "logits/rejected": -2.0578324794769287, | |
| "logps/chosen": -187.84454345703125, | |
| "logps/rejected": -182.75221252441406, | |
| "loss": 0.6427, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.49994075298309326, | |
| "rewards/margins": 0.15868376195430756, | |
| "rewards/rejected": -0.658624529838562, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.935587735217242e-05, | |
| "logits/chosen": -2.268324851989746, | |
| "logits/rejected": -2.2720093727111816, | |
| "logps/chosen": -150.13555908203125, | |
| "logps/rejected": -161.95318603515625, | |
| "loss": 0.6464, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.3330093026161194, | |
| "rewards/margins": 0.15659931302070618, | |
| "rewards/rejected": -0.48960861563682556, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.928170537736981e-05, | |
| "logits/chosen": -2.2737841606140137, | |
| "logits/rejected": -2.1588943004608154, | |
| "logps/chosen": -178.9062042236328, | |
| "logps/rejected": -147.05792236328125, | |
| "loss": 0.7555, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5811339616775513, | |
| "rewards/margins": -0.08029365539550781, | |
| "rewards/rejected": -0.5008403658866882, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.920734631716991e-05, | |
| "logits/chosen": -2.042383909225464, | |
| "logits/rejected": -1.9608237743377686, | |
| "logps/chosen": -149.2159423828125, | |
| "logps/rejected": -144.9436492919922, | |
| "loss": 0.6465, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.318014532327652, | |
| "rewards/margins": 0.17784127593040466, | |
| "rewards/rejected": -0.49585577845573425, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.913280114565066e-05, | |
| "logits/chosen": -2.2850501537323, | |
| "logits/rejected": -2.2817530632019043, | |
| "logps/chosen": -186.5043182373047, | |
| "logps/rejected": -209.9068603515625, | |
| "loss": 0.5943, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2195523977279663, | |
| "rewards/margins": 0.2693576216697693, | |
| "rewards/rejected": -0.488910049200058, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.905807083932799e-05, | |
| "logits/chosen": -2.0979740619659424, | |
| "logits/rejected": -2.1025643348693848, | |
| "logps/chosen": -160.7048797607422, | |
| "logps/rejected": -173.43521118164062, | |
| "loss": 0.6972, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3717915713787079, | |
| "rewards/margins": 0.04246264696121216, | |
| "rewards/rejected": -0.41425418853759766, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.898315637714308e-05, | |
| "logits/chosen": -2.081289291381836, | |
| "logits/rejected": -2.132338762283325, | |
| "logps/chosen": -142.55767822265625, | |
| "logps/rejected": -158.46466064453125, | |
| "loss": 0.6854, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.29465389251708984, | |
| "rewards/margins": 0.20354387164115906, | |
| "rewards/rejected": -0.4981977343559265, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.8908058740449436e-05, | |
| "logits/chosen": -2.079378843307495, | |
| "logits/rejected": -2.042525053024292, | |
| "logps/chosen": -206.33206176757812, | |
| "logps/rejected": -229.58287048339844, | |
| "loss": 0.7516, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.552139937877655, | |
| "rewards/margins": 0.021834930405020714, | |
| "rewards/rejected": -0.5739747881889343, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.883277891300011e-05, | |
| "logits/chosen": -2.2758610248565674, | |
| "logits/rejected": -2.3157222270965576, | |
| "logps/chosen": -178.7675323486328, | |
| "logps/rejected": -196.29611206054688, | |
| "loss": 0.6436, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.5158277153968811, | |
| "rewards/margins": 0.2549564838409424, | |
| "rewards/rejected": -0.7707842588424683, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.8757317880934786e-05, | |
| "logits/chosen": -2.155709743499756, | |
| "logits/rejected": -2.213414430618286, | |
| "logps/chosen": -235.17843627929688, | |
| "logps/rejected": -231.29014587402344, | |
| "loss": 0.7146, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.44189226627349854, | |
| "rewards/margins": 0.04725969582796097, | |
| "rewards/rejected": -0.4891519546508789, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.868167663276686e-05, | |
| "logits/chosen": -1.9473401308059692, | |
| "logits/rejected": -2.0311379432678223, | |
| "logps/chosen": -163.64012145996094, | |
| "logps/rejected": -174.59686279296875, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.38134512305259705, | |
| "rewards/margins": 0.09880531579256058, | |
| "rewards/rejected": -0.4801504611968994, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.860585615937051e-05, | |
| "logits/chosen": -2.1732089519500732, | |
| "logits/rejected": -2.123350143432617, | |
| "logps/chosen": -166.0354461669922, | |
| "logps/rejected": -158.03367614746094, | |
| "loss": 0.6836, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.4032551050186157, | |
| "rewards/margins": 0.10109908133745193, | |
| "rewards/rejected": -0.5043542385101318, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.85298574539677e-05, | |
| "logits/chosen": -2.3103952407836914, | |
| "logits/rejected": -2.3746755123138428, | |
| "logps/chosen": -190.70632934570312, | |
| "logps/rejected": -196.38328552246094, | |
| "loss": 0.6636, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5452268123626709, | |
| "rewards/margins": 0.2204725593328476, | |
| "rewards/rejected": -0.7656993865966797, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.845368151211516e-05, | |
| "logits/chosen": -2.1683666706085205, | |
| "logits/rejected": -2.214723825454712, | |
| "logps/chosen": -164.623779296875, | |
| "logps/rejected": -189.20724487304688, | |
| "loss": 0.5786, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.40996867418289185, | |
| "rewards/margins": 0.3506165146827698, | |
| "rewards/rejected": -0.7605851888656616, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.837732933169135e-05, | |
| "logits/chosen": -1.9226962327957153, | |
| "logits/rejected": -2.047916889190674, | |
| "logps/chosen": -135.63719177246094, | |
| "logps/rejected": -167.1735382080078, | |
| "loss": 0.6173, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.28226932883262634, | |
| "rewards/margins": 0.2436113953590393, | |
| "rewards/rejected": -0.525880753993988, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.830080191288342e-05, | |
| "logits/chosen": -2.2397301197052, | |
| "logits/rejected": -2.376453399658203, | |
| "logps/chosen": -137.69940185546875, | |
| "logps/rejected": -161.41018676757812, | |
| "loss": 0.6085, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.2780275046825409, | |
| "rewards/margins": 0.2456223964691162, | |
| "rewards/rejected": -0.5236499309539795, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.822410025817406e-05, | |
| "logits/chosen": -2.126335620880127, | |
| "logits/rejected": -2.159806251525879, | |
| "logps/chosen": -184.33453369140625, | |
| "logps/rejected": -201.81451416015625, | |
| "loss": 0.7145, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4947819411754608, | |
| "rewards/margins": 0.04000832885503769, | |
| "rewards/rejected": -0.5347902774810791, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.8147225372328424e-05, | |
| "logits/chosen": -2.224623441696167, | |
| "logits/rejected": -2.247649908065796, | |
| "logps/chosen": -200.61070251464844, | |
| "logps/rejected": -201.32554626464844, | |
| "loss": 0.7505, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.8298049569129944, | |
| "rewards/margins": -0.05505021661520004, | |
| "rewards/rejected": -0.7747547626495361, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.8070178262380876e-05, | |
| "logits/chosen": -2.1214921474456787, | |
| "logits/rejected": -2.154468536376953, | |
| "logps/chosen": -165.05250549316406, | |
| "logps/rejected": -180.82394409179688, | |
| "loss": 0.7121, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5545100569725037, | |
| "rewards/margins": 0.03025786019861698, | |
| "rewards/rejected": -0.5847679376602173, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.7992959937621896e-05, | |
| "logits/chosen": -2.088984251022339, | |
| "logits/rejected": -2.110299587249756, | |
| "logps/chosen": -170.40736389160156, | |
| "logps/rejected": -202.48419189453125, | |
| "loss": 0.5449, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.07705488055944443, | |
| "rewards/margins": 0.40208369493484497, | |
| "rewards/rejected": -0.47913858294487, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.7915571409584836e-05, | |
| "logits/chosen": -2.269618034362793, | |
| "logits/rejected": -2.4114909172058105, | |
| "logps/chosen": -158.34793090820312, | |
| "logps/rejected": -189.32492065429688, | |
| "loss": 0.599, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3794673979282379, | |
| "rewards/margins": 0.3096137046813965, | |
| "rewards/rejected": -0.689081072807312, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.78380136920326e-05, | |
| "logits/chosen": -2.2543036937713623, | |
| "logits/rejected": -2.237398147583008, | |
| "logps/chosen": -158.48202514648438, | |
| "logps/rejected": -165.46192932128906, | |
| "loss": 0.8152, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.586533784866333, | |
| "rewards/margins": -0.15689069032669067, | |
| "rewards/rejected": -0.4296431541442871, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.776028780094446e-05, | |
| "logits/chosen": -2.1381900310516357, | |
| "logits/rejected": -2.267850160598755, | |
| "logps/chosen": -165.80572509765625, | |
| "logps/rejected": -185.3111114501953, | |
| "loss": 0.6513, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.67560213804245, | |
| "rewards/margins": 0.18039174377918243, | |
| "rewards/rejected": -0.8559938669204712, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.768239475450269e-05, | |
| "logits/chosen": -2.1430163383483887, | |
| "logits/rejected": -2.0726311206817627, | |
| "logps/chosen": -184.41261291503906, | |
| "logps/rejected": -172.13687133789062, | |
| "loss": 0.7117, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.5548739433288574, | |
| "rewards/margins": 0.06595984101295471, | |
| "rewards/rejected": -0.6208338141441345, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.760433557307922e-05, | |
| "logits/chosen": -2.036318063735962, | |
| "logits/rejected": -1.9743261337280273, | |
| "logps/chosen": -153.29840087890625, | |
| "logps/rejected": -148.0943145751953, | |
| "loss": 0.6464, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4892306327819824, | |
| "rewards/margins": 0.15469516813755035, | |
| "rewards/rejected": -0.6439257860183716, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.752611127922232e-05, | |
| "logits/chosen": -1.9810808897018433, | |
| "logits/rejected": -1.9635897874832153, | |
| "logps/chosen": -131.88314819335938, | |
| "logps/rejected": -134.77906799316406, | |
| "loss": 0.6707, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.7301316857337952, | |
| "rewards/margins": 0.0971762016415596, | |
| "rewards/rejected": -0.8273078799247742, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.744772289764316e-05, | |
| "logits/chosen": -2.3128020763397217, | |
| "logits/rejected": -2.197258949279785, | |
| "logps/chosen": -192.55340576171875, | |
| "logps/rejected": -176.83547973632812, | |
| "loss": 0.7391, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.6548212766647339, | |
| "rewards/margins": 0.0352729931473732, | |
| "rewards/rejected": -0.6900942325592041, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.736917145520242e-05, | |
| "logits/chosen": -2.1667685508728027, | |
| "logits/rejected": -2.20652437210083, | |
| "logps/chosen": -164.12649536132812, | |
| "logps/rejected": -167.70028686523438, | |
| "loss": 0.8265, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5798961520195007, | |
| "rewards/margins": -0.11231443285942078, | |
| "rewards/rejected": -0.46758168935775757, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.7290457980896795e-05, | |
| "logits/chosen": -2.2581608295440674, | |
| "logits/rejected": -2.2037389278411865, | |
| "logps/chosen": -222.37803649902344, | |
| "logps/rejected": -228.9108428955078, | |
| "loss": 0.7541, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.6088207960128784, | |
| "rewards/margins": 0.11347416043281555, | |
| "rewards/rejected": -0.7222948670387268, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.721158350584556e-05, | |
| "logits/chosen": -2.240694046020508, | |
| "logits/rejected": -2.2583985328674316, | |
| "logps/chosen": -185.7880859375, | |
| "logps/rejected": -204.0207061767578, | |
| "loss": 0.82, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.915455162525177, | |
| "rewards/margins": -0.15485824644565582, | |
| "rewards/rejected": -0.7605970501899719, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.713254906327703e-05, | |
| "logits/chosen": -2.2125234603881836, | |
| "logits/rejected": -2.260956287384033, | |
| "logps/chosen": -149.2694091796875, | |
| "logps/rejected": -172.06015014648438, | |
| "loss": 0.7258, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6628731489181519, | |
| "rewards/margins": 0.03825627267360687, | |
| "rewards/rejected": -0.7011294364929199, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.705335568851506e-05, | |
| "logits/chosen": -2.180601119995117, | |
| "logits/rejected": -2.2720632553100586, | |
| "logps/chosen": -139.24539184570312, | |
| "logps/rejected": -142.7550811767578, | |
| "loss": 0.6549, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.501410722732544, | |
| "rewards/margins": 0.1533694863319397, | |
| "rewards/rejected": -0.6547802686691284, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.697400441896543e-05, | |
| "logits/chosen": -2.1138863563537598, | |
| "logits/rejected": -2.1623191833496094, | |
| "logps/chosen": -160.19261169433594, | |
| "logps/rejected": -176.69573974609375, | |
| "loss": 0.9557, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.7389885187149048, | |
| "rewards/margins": -0.19968615472316742, | |
| "rewards/rejected": -0.5393024682998657, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.689449629410234e-05, | |
| "logits/chosen": -2.2931628227233887, | |
| "logits/rejected": -2.288308620452881, | |
| "logps/chosen": -236.5543670654297, | |
| "logps/rejected": -224.42071533203125, | |
| "loss": 0.8062, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.63106769323349, | |
| "rewards/margins": -0.12141348421573639, | |
| "rewards/rejected": -0.5096542835235596, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.681483235545468e-05, | |
| "logits/chosen": -2.1109299659729004, | |
| "logits/rejected": -2.1560161113739014, | |
| "logps/chosen": -156.2798614501953, | |
| "logps/rejected": -159.78579711914062, | |
| "loss": 0.6784, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.47323617339134216, | |
| "rewards/margins": 0.09092390537261963, | |
| "rewards/rejected": -0.564160168170929, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6735013646592475e-05, | |
| "logits/chosen": -2.210202693939209, | |
| "logits/rejected": -2.252317190170288, | |
| "logps/chosen": -169.9296112060547, | |
| "logps/rejected": -176.38490295410156, | |
| "loss": 0.7048, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4336109757423401, | |
| "rewards/margins": 0.14937293529510498, | |
| "rewards/rejected": -0.5829839110374451, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6655041213113184e-05, | |
| "logits/chosen": -2.198397159576416, | |
| "logits/rejected": -2.211723566055298, | |
| "logps/chosen": -211.31134033203125, | |
| "logps/rejected": -191.85545349121094, | |
| "loss": 0.7174, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.8685683608055115, | |
| "rewards/margins": 0.12455281615257263, | |
| "rewards/rejected": -0.9931210875511169, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.657491610262802e-05, | |
| "logits/chosen": -2.2646853923797607, | |
| "logits/rejected": -2.2568039894104004, | |
| "logps/chosen": -144.8211669921875, | |
| "logps/rejected": -129.66143798828125, | |
| "loss": 0.788, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4251525402069092, | |
| "rewards/margins": -0.13967978954315186, | |
| "rewards/rejected": -0.2854727506637573, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6494639364748184e-05, | |
| "logits/chosen": -2.1162049770355225, | |
| "logits/rejected": -2.1026415824890137, | |
| "logps/chosen": -199.32569885253906, | |
| "logps/rejected": -194.08029174804688, | |
| "loss": 0.6697, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.8081960082054138, | |
| "rewards/margins": 0.15443959832191467, | |
| "rewards/rejected": -0.9626356363296509, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.641421205107116e-05, | |
| "logits/chosen": -2.3222200870513916, | |
| "logits/rejected": -2.2925901412963867, | |
| "logps/chosen": -162.95602416992188, | |
| "logps/rejected": -149.4652557373047, | |
| "loss": 0.6895, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5014199614524841, | |
| "rewards/margins": 0.12428833544254303, | |
| "rewards/rejected": -0.625708281993866, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.633363521516693e-05, | |
| "logits/chosen": -2.1190714836120605, | |
| "logits/rejected": -2.1951043605804443, | |
| "logps/chosen": -174.7755126953125, | |
| "logps/rejected": -179.20071411132812, | |
| "loss": 0.699, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.42257675528526306, | |
| "rewards/margins": 0.17960743606090546, | |
| "rewards/rejected": -0.6021841764450073, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.625290991256414e-05, | |
| "logits/chosen": -2.0900583267211914, | |
| "logits/rejected": -2.1080591678619385, | |
| "logps/chosen": -135.29946899414062, | |
| "logps/rejected": -141.16790771484375, | |
| "loss": 0.7975, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.7247196435928345, | |
| "rewards/margins": -0.14072172343730927, | |
| "rewards/rejected": -0.5839979648590088, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6172037200736325e-05, | |
| "logits/chosen": -2.2358274459838867, | |
| "logits/rejected": -2.232405424118042, | |
| "logps/chosen": -161.24636840820312, | |
| "logps/rejected": -166.5486602783203, | |
| "loss": 0.7395, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.6730926632881165, | |
| "rewards/margins": 0.026289351284503937, | |
| "rewards/rejected": -0.699381947517395, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.609101813908801e-05, | |
| "logits/chosen": -2.3274168968200684, | |
| "logits/rejected": -2.331890344619751, | |
| "logps/chosen": -195.30133056640625, | |
| "logps/rejected": -219.84608459472656, | |
| "loss": 0.5521, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.5582520961761475, | |
| "rewards/margins": 0.3810853660106659, | |
| "rewards/rejected": -0.939337432384491, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.600985378894086e-05, | |
| "logits/chosen": -2.26960825920105, | |
| "logits/rejected": -2.260411024093628, | |
| "logps/chosen": -166.19061279296875, | |
| "logps/rejected": -156.9241180419922, | |
| "loss": 0.6529, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.43417248129844666, | |
| "rewards/margins": 0.15785039961338043, | |
| "rewards/rejected": -0.5920228362083435, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.5928545213519784e-05, | |
| "logits/chosen": -2.35903263092041, | |
| "logits/rejected": -2.344247579574585, | |
| "logps/chosen": -166.66845703125, | |
| "logps/rejected": -182.43173217773438, | |
| "loss": 0.7179, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6532277464866638, | |
| "rewards/margins": 0.04449939727783203, | |
| "rewards/rejected": -0.6977271437644958, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.5847093477938956e-05, | |
| "logits/chosen": -2.2892184257507324, | |
| "logits/rejected": -2.289012908935547, | |
| "logps/chosen": -159.4162139892578, | |
| "logps/rejected": -170.12847900390625, | |
| "loss": 0.7077, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5636630654335022, | |
| "rewards/margins": 0.03781846538186073, | |
| "rewards/rejected": -0.601481556892395, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.576549964918794e-05, | |
| "logits/chosen": -2.432997226715088, | |
| "logits/rejected": -2.489370346069336, | |
| "logps/chosen": -219.2733154296875, | |
| "logps/rejected": -211.38180541992188, | |
| "loss": 0.707, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -1.058005690574646, | |
| "rewards/margins": 0.009618997573852539, | |
| "rewards/rejected": -1.0676246881484985, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5683764796117634e-05, | |
| "logits/chosen": -2.390814781188965, | |
| "logits/rejected": -2.3141655921936035, | |
| "logps/chosen": -180.41212463378906, | |
| "logps/rejected": -175.58995056152344, | |
| "loss": 0.8505, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.7344385385513306, | |
| "rewards/margins": -0.22211739420890808, | |
| "rewards/rejected": -0.5123211145401001, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.560188998942634e-05, | |
| "logits/chosen": -2.296417474746704, | |
| "logits/rejected": -2.3206100463867188, | |
| "logps/chosen": -177.83306884765625, | |
| "logps/rejected": -176.24867248535156, | |
| "loss": 0.6957, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.7862380743026733, | |
| "rewards/margins": 0.04238337650895119, | |
| "rewards/rejected": -0.828621506690979, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5519876301645684e-05, | |
| "logits/chosen": -2.2782628536224365, | |
| "logits/rejected": -2.3132076263427734, | |
| "logps/chosen": -195.24978637695312, | |
| "logps/rejected": -190.56097412109375, | |
| "loss": 0.963, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.9706141948699951, | |
| "rewards/margins": -0.22657813131809235, | |
| "rewards/rejected": -0.7440361976623535, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543772480712658e-05, | |
| "logits/chosen": -2.2542788982391357, | |
| "logits/rejected": -2.2529568672180176, | |
| "logps/chosen": -187.711669921875, | |
| "logps/rejected": -195.0843505859375, | |
| "loss": 0.7652, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.38882195949554443, | |
| "rewards/margins": -0.07953725755214691, | |
| "rewards/rejected": -0.3092847466468811, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.535543658202518e-05, | |
| "logits/chosen": -2.358016014099121, | |
| "logits/rejected": -2.395139694213867, | |
| "logps/chosen": -183.68569946289062, | |
| "logps/rejected": -180.41876220703125, | |
| "loss": 0.762, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5203931927680969, | |
| "rewards/margins": -0.0901247039437294, | |
| "rewards/rejected": -0.43026843667030334, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5273012704288745e-05, | |
| "logits/chosen": -2.331780195236206, | |
| "logits/rejected": -2.328648805618286, | |
| "logps/chosen": -172.1591796875, | |
| "logps/rejected": -173.70399475097656, | |
| "loss": 0.6944, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.739228367805481, | |
| "rewards/margins": 0.05720563977956772, | |
| "rewards/rejected": -0.7964339256286621, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.519045425364156e-05, | |
| "logits/chosen": -2.370368480682373, | |
| "logits/rejected": -2.483001947402954, | |
| "logps/chosen": -176.0197296142578, | |
| "logps/rejected": -192.3662567138672, | |
| "loss": 0.6453, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6221442222595215, | |
| "rewards/margins": 0.18113647401332855, | |
| "rewards/rejected": -0.8032807111740112, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5107762311570735e-05, | |
| "logits/chosen": -2.0812158584594727, | |
| "logits/rejected": -2.133009195327759, | |
| "logps/chosen": -168.70896911621094, | |
| "logps/rejected": -162.06129455566406, | |
| "loss": 0.7665, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.7016276121139526, | |
| "rewards/margins": -0.06663860380649567, | |
| "rewards/rejected": -0.6349890232086182, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.50249379613121e-05, | |
| "logits/chosen": -2.4355461597442627, | |
| "logits/rejected": -2.370302438735962, | |
| "logps/chosen": -192.8936767578125, | |
| "logps/rejected": -197.0657196044922, | |
| "loss": 0.6898, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.6052863597869873, | |
| "rewards/margins": 0.21437309682369232, | |
| "rewards/rejected": -0.819659411907196, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.4941982287835974e-05, | |
| "logits/chosen": -2.306861639022827, | |
| "logits/rejected": -2.360962390899658, | |
| "logps/chosen": -155.9933319091797, | |
| "logps/rejected": -171.38294982910156, | |
| "loss": 0.5591, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.46738651394844055, | |
| "rewards/margins": 0.3353773355484009, | |
| "rewards/rejected": -0.802763819694519, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4858896377832966e-05, | |
| "logits/chosen": -2.302903890609741, | |
| "logits/rejected": -2.251715898513794, | |
| "logps/chosen": -150.3655242919922, | |
| "logps/rejected": -138.7833251953125, | |
| "loss": 0.8571, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.7627382278442383, | |
| "rewards/margins": -0.25699174404144287, | |
| "rewards/rejected": -0.5057464838027954, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4775681319699746e-05, | |
| "logits/chosen": -2.1884803771972656, | |
| "logits/rejected": -2.1626532077789307, | |
| "logps/chosen": -168.5961456298828, | |
| "logps/rejected": -159.1962890625, | |
| "loss": 0.8819, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.9429460763931274, | |
| "rewards/margins": -0.2255764901638031, | |
| "rewards/rejected": -0.7173696160316467, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.469233820352477e-05, | |
| "logits/chosen": -2.30151629447937, | |
| "logits/rejected": -2.3949134349823, | |
| "logps/chosen": -167.24282836914062, | |
| "logps/rejected": -211.11749267578125, | |
| "loss": 0.6071, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.587899923324585, | |
| "rewards/margins": 0.233824223279953, | |
| "rewards/rejected": -0.8217241764068604, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4608868121074e-05, | |
| "logits/chosen": -2.2129104137420654, | |
| "logits/rejected": -2.2608776092529297, | |
| "logps/chosen": -146.16940307617188, | |
| "logps/rejected": -162.7283477783203, | |
| "loss": 0.8571, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.529353678226471, | |
| "rewards/margins": -0.18388289213180542, | |
| "rewards/rejected": -0.3454708158969879, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.452527216577665e-05, | |
| "logits/chosen": -2.2917697429656982, | |
| "logits/rejected": -2.2677531242370605, | |
| "logps/chosen": -140.55955505371094, | |
| "logps/rejected": -153.37045288085938, | |
| "loss": 0.7399, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5034148693084717, | |
| "rewards/margins": 0.03199050575494766, | |
| "rewards/rejected": -0.5354053378105164, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4441551432710784e-05, | |
| "logits/chosen": -2.2404325008392334, | |
| "logits/rejected": -2.228639602661133, | |
| "logps/chosen": -177.03250122070312, | |
| "logps/rejected": -174.8275146484375, | |
| "loss": 0.7158, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.9018635749816895, | |
| "rewards/margins": 0.07460644841194153, | |
| "rewards/rejected": -0.9764699339866638, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4357707018589036e-05, | |
| "logits/chosen": -2.318934679031372, | |
| "logits/rejected": -2.3687174320220947, | |
| "logps/chosen": -153.05734252929688, | |
| "logps/rejected": -178.81900024414062, | |
| "loss": 0.8638, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6083782315254211, | |
| "rewards/margins": -0.1766158491373062, | |
| "rewards/rejected": -0.43176233768463135, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.42737400217442e-05, | |
| "logits/chosen": -2.183048963546753, | |
| "logits/rejected": -2.3311784267425537, | |
| "logps/chosen": -166.5866241455078, | |
| "logps/rejected": -211.22116088867188, | |
| "loss": 0.6709, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.43151307106018066, | |
| "rewards/margins": 0.28348180651664734, | |
| "rewards/rejected": -0.7149949073791504, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4189651542114884e-05, | |
| "logits/chosen": -2.173567771911621, | |
| "logits/rejected": -2.2484710216522217, | |
| "logps/chosen": -209.43142700195312, | |
| "logps/rejected": -208.07943725585938, | |
| "loss": 0.6661, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5889863967895508, | |
| "rewards/margins": 0.12892867624759674, | |
| "rewards/rejected": -0.7179149985313416, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.410544268123106e-05, | |
| "logits/chosen": -2.267691135406494, | |
| "logits/rejected": -2.2806975841522217, | |
| "logps/chosen": -158.83993530273438, | |
| "logps/rejected": -164.43789672851562, | |
| "loss": 0.7507, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4183788001537323, | |
| "rewards/margins": -0.05176548659801483, | |
| "rewards/rejected": -0.3666132986545563, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.4021114542199664e-05, | |
| "logits/chosen": -2.186677932739258, | |
| "logits/rejected": -2.1767873764038086, | |
| "logps/chosen": -157.10214233398438, | |
| "logps/rejected": -161.1920928955078, | |
| "loss": 0.7073, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.364505410194397, | |
| "rewards/margins": 0.058986373245716095, | |
| "rewards/rejected": -0.42349180579185486, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.393666822969012e-05, | |
| "logits/chosen": -2.2252414226531982, | |
| "logits/rejected": -2.2977192401885986, | |
| "logps/chosen": -182.1011962890625, | |
| "logps/rejected": -182.70895385742188, | |
| "loss": 0.6249, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.47798287868499756, | |
| "rewards/margins": 0.19156885147094727, | |
| "rewards/rejected": -0.6695517897605896, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3852104849919905e-05, | |
| "logits/chosen": -2.3564376831054688, | |
| "logits/rejected": -2.347639322280884, | |
| "logps/chosen": -221.4008331298828, | |
| "logps/rejected": -230.65008544921875, | |
| "loss": 0.7011, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.682982861995697, | |
| "rewards/margins": 0.05443616956472397, | |
| "rewards/rejected": -0.737419068813324, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3767425510640026e-05, | |
| "logits/chosen": -2.242929458618164, | |
| "logits/rejected": -2.2702651023864746, | |
| "logps/chosen": -257.7521057128906, | |
| "logps/rejected": -235.80870056152344, | |
| "loss": 0.8796, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.8189445734024048, | |
| "rewards/margins": -0.18073782324790955, | |
| "rewards/rejected": -0.6382068395614624, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3682631321120504e-05, | |
| "logits/chosen": -2.17533802986145, | |
| "logits/rejected": -2.1521806716918945, | |
| "logps/chosen": -181.9336395263672, | |
| "logps/rejected": -189.3466339111328, | |
| "loss": 0.7851, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.8867939710617065, | |
| "rewards/margins": -0.09112608432769775, | |
| "rewards/rejected": -0.7956678867340088, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.35977233921359e-05, | |
| "logits/chosen": -2.3076705932617188, | |
| "logits/rejected": -2.3349483013153076, | |
| "logps/chosen": -209.75247192382812, | |
| "logps/rejected": -202.7659912109375, | |
| "loss": 0.7735, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.7706497311592102, | |
| "rewards/margins": -0.07641672343015671, | |
| "rewards/rejected": -0.6942330002784729, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.351270283595066e-05, | |
| "logits/chosen": -2.349508285522461, | |
| "logits/rejected": -2.343428611755371, | |
| "logps/chosen": -206.8453826904297, | |
| "logps/rejected": -203.99423217773438, | |
| "loss": 0.7483, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.6214013695716858, | |
| "rewards/margins": -0.06958682835102081, | |
| "rewards/rejected": -0.5518144965171814, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.342757076630467e-05, | |
| "logits/chosen": -2.241558313369751, | |
| "logits/rejected": -2.2577505111694336, | |
| "logps/chosen": -151.2669677734375, | |
| "logps/rejected": -175.61398315429688, | |
| "loss": 0.6901, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.8200474381446838, | |
| "rewards/margins": 0.10837754607200623, | |
| "rewards/rejected": -0.9284249544143677, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3342328298398565e-05, | |
| "logits/chosen": -2.346818447113037, | |
| "logits/rejected": -2.319014072418213, | |
| "logps/chosen": -232.11334228515625, | |
| "logps/rejected": -231.2914276123047, | |
| "loss": 0.8118, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.7047826647758484, | |
| "rewards/margins": -0.15696799755096436, | |
| "rewards/rejected": -0.547814667224884, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.3256976548879184e-05, | |
| "logits/chosen": -2.21856689453125, | |
| "logits/rejected": -2.2464771270751953, | |
| "logps/chosen": -170.10299682617188, | |
| "logps/rejected": -203.31494140625, | |
| "loss": 0.6017, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.6384906768798828, | |
| "rewards/margins": 0.3270528316497803, | |
| "rewards/rejected": -0.9655435085296631, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.317151663582488e-05, | |
| "logits/chosen": -2.074263095855713, | |
| "logits/rejected": -2.1142096519470215, | |
| "logps/chosen": -150.205322265625, | |
| "logps/rejected": -170.9268035888672, | |
| "loss": 0.6907, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6000096201896667, | |
| "rewards/margins": 0.10498014837503433, | |
| "rewards/rejected": -0.7049897909164429, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.308594967873095e-05, | |
| "logits/chosen": -2.1494641304016113, | |
| "logits/rejected": -2.1588127613067627, | |
| "logps/chosen": -156.4353790283203, | |
| "logps/rejected": -182.8409881591797, | |
| "loss": 0.5753, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.4203645884990692, | |
| "rewards/margins": 0.34836992621421814, | |
| "rewards/rejected": -0.7687345147132874, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.300027679849492e-05, | |
| "logits/chosen": -2.0936717987060547, | |
| "logits/rejected": -2.1512672901153564, | |
| "logps/chosen": -147.28402709960938, | |
| "logps/rejected": -154.80926513671875, | |
| "loss": 0.6875, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.6472322940826416, | |
| "rewards/margins": 0.17809806764125824, | |
| "rewards/rejected": -0.825330376625061, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2914499117401865e-05, | |
| "logits/chosen": -2.4345755577087402, | |
| "logits/rejected": -2.317566394805908, | |
| "logps/chosen": -254.519287109375, | |
| "logps/rejected": -226.63009643554688, | |
| "loss": 0.7925, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.68197101354599, | |
| "rewards/margins": -0.08323682844638824, | |
| "rewards/rejected": -0.5987342000007629, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2828617759109714e-05, | |
| "logits/chosen": -2.1829962730407715, | |
| "logits/rejected": -2.186575412750244, | |
| "logps/chosen": -160.66737365722656, | |
| "logps/rejected": -164.59449768066406, | |
| "loss": 0.6558, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.502325177192688, | |
| "rewards/margins": 0.1352367401123047, | |
| "rewards/rejected": -0.6375619173049927, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.274263384863453e-05, | |
| "logits/chosen": -2.490436315536499, | |
| "logits/rejected": -2.4755001068115234, | |
| "logps/chosen": -183.80218505859375, | |
| "logps/rejected": -168.38766479492188, | |
| "loss": 0.8063, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5363278388977051, | |
| "rewards/margins": -0.12316776812076569, | |
| "rewards/rejected": -0.4131600558757782, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.265654851233579e-05, | |
| "logits/chosen": -2.363406181335449, | |
| "logits/rejected": -2.375720500946045, | |
| "logps/chosen": -172.26904296875, | |
| "logps/rejected": -158.56695556640625, | |
| "loss": 0.7663, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.6086016297340393, | |
| "rewards/margins": -0.03708178177475929, | |
| "rewards/rejected": -0.5715197920799255, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2570362877901605e-05, | |
| "logits/chosen": -2.2647805213928223, | |
| "logits/rejected": -2.3734748363494873, | |
| "logps/chosen": -162.90496826171875, | |
| "logps/rejected": -196.47779846191406, | |
| "loss": 0.6566, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4291108548641205, | |
| "rewards/margins": 0.1305677592754364, | |
| "rewards/rejected": -0.5596785545349121, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2484078074333954e-05, | |
| "logits/chosen": -2.1634037494659424, | |
| "logits/rejected": -2.255723237991333, | |
| "logps/chosen": -139.2611083984375, | |
| "logps/rejected": -139.0371551513672, | |
| "loss": 0.7789, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3354373574256897, | |
| "rewards/margins": -0.11600707471370697, | |
| "rewards/rejected": -0.21943029761314392, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.2397695231933894e-05, | |
| "logits/chosen": -2.270599365234375, | |
| "logits/rejected": -2.3111140727996826, | |
| "logps/chosen": -156.4451141357422, | |
| "logps/rejected": -180.2852783203125, | |
| "loss": 0.8026, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.5979058742523193, | |
| "rewards/margins": -0.14397534728050232, | |
| "rewards/rejected": -0.4539305567741394, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.231121548228676e-05, | |
| "logits/chosen": -2.267456531524658, | |
| "logits/rejected": -2.276815414428711, | |
| "logps/chosen": -182.44110107421875, | |
| "logps/rejected": -203.32937622070312, | |
| "loss": 0.6073, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.3719044029712677, | |
| "rewards/margins": 0.380204975605011, | |
| "rewards/rejected": -0.7521094083786011, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.2224639958247346e-05, | |
| "logits/chosen": -2.419814348220825, | |
| "logits/rejected": -2.473423480987549, | |
| "logps/chosen": -183.04537963867188, | |
| "logps/rejected": -196.78546142578125, | |
| "loss": 0.7777, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.40696901082992554, | |
| "rewards/margins": -0.05508175492286682, | |
| "rewards/rejected": -0.3518872857093811, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.213796979392505e-05, | |
| "logits/chosen": -2.109714984893799, | |
| "logits/rejected": -2.1827077865600586, | |
| "logps/chosen": -203.89097595214844, | |
| "logps/rejected": -194.51148986816406, | |
| "loss": 0.9216, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.8011529445648193, | |
| "rewards/margins": -0.2627783715724945, | |
| "rewards/rejected": -0.5383745431900024, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.205120612466904e-05, | |
| "logits/chosen": -2.0772898197174072, | |
| "logits/rejected": -2.1390609741210938, | |
| "logps/chosen": -165.24966430664062, | |
| "logps/rejected": -185.6719970703125, | |
| "loss": 0.6722, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.3127191364765167, | |
| "rewards/margins": 0.09625725448131561, | |
| "rewards/rejected": -0.40897637605667114, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.196435008705332e-05, | |
| "logits/chosen": -2.3135664463043213, | |
| "logits/rejected": -2.2848143577575684, | |
| "logps/chosen": -166.76080322265625, | |
| "logps/rejected": -177.1256866455078, | |
| "loss": 0.7196, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3662790358066559, | |
| "rewards/margins": 0.0703655257821083, | |
| "rewards/rejected": -0.4366445541381836, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.187740281886195e-05, | |
| "logits/chosen": -2.239975929260254, | |
| "logits/rejected": -2.302727460861206, | |
| "logps/chosen": -169.4385986328125, | |
| "logps/rejected": -211.87564086914062, | |
| "loss": 0.5544, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.4735289514064789, | |
| "rewards/margins": 0.43947097659111023, | |
| "rewards/rejected": -0.9129999279975891, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.179036545907405e-05, | |
| "logits/chosen": -2.1537723541259766, | |
| "logits/rejected": -2.12878680229187, | |
| "logps/chosen": -179.65725708007812, | |
| "logps/rejected": -185.7678680419922, | |
| "loss": 0.7027, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.6164141893386841, | |
| "rewards/margins": 0.10950362682342529, | |
| "rewards/rejected": -0.7259179353713989, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.170323914784889e-05, | |
| "logits/chosen": -2.279991626739502, | |
| "logits/rejected": -2.3006529808044434, | |
| "logps/chosen": -186.97596740722656, | |
| "logps/rejected": -177.9309844970703, | |
| "loss": 0.7124, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.36138916015625, | |
| "rewards/margins": 0.08198119699954987, | |
| "rewards/rejected": -0.44337037205696106, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.161602502651099e-05, | |
| "logits/chosen": -2.066502094268799, | |
| "logits/rejected": -2.1805591583251953, | |
| "logps/chosen": -137.30715942382812, | |
| "logps/rejected": -165.40679931640625, | |
| "loss": 0.7314, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.33508479595184326, | |
| "rewards/margins": -0.01991906762123108, | |
| "rewards/rejected": -0.31516575813293457, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.1528724237535165e-05, | |
| "logits/chosen": -2.1526942253112793, | |
| "logits/rejected": -2.1386966705322266, | |
| "logps/chosen": -194.58193969726562, | |
| "logps/rejected": -182.65084838867188, | |
| "loss": 0.6389, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.4528476297855377, | |
| "rewards/margins": 0.17260009050369263, | |
| "rewards/rejected": -0.6254477500915527, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.144133792453154e-05, | |
| "logits/chosen": -2.1543819904327393, | |
| "logits/rejected": -2.119481325149536, | |
| "logps/chosen": -184.62754821777344, | |
| "logps/rejected": -180.76568603515625, | |
| "loss": 0.7137, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5128884315490723, | |
| "rewards/margins": 0.0639130175113678, | |
| "rewards/rejected": -0.5768015384674072, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.1353867232230564e-05, | |
| "logits/chosen": -2.213226795196533, | |
| "logits/rejected": -2.2723488807678223, | |
| "logps/chosen": -186.5143585205078, | |
| "logps/rejected": -195.0329132080078, | |
| "loss": 0.6662, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.4016525447368622, | |
| "rewards/margins": 0.18189382553100586, | |
| "rewards/rejected": -0.5835464000701904, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.126631330646802e-05, | |
| "logits/chosen": -2.284442901611328, | |
| "logits/rejected": -2.293401002883911, | |
| "logps/chosen": -153.34088134765625, | |
| "logps/rejected": -156.06895446777344, | |
| "loss": 0.67, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.3650515377521515, | |
| "rewards/margins": 0.11144375056028366, | |
| "rewards/rejected": -0.47649532556533813, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.117867729417004e-05, | |
| "logits/chosen": -2.1648364067077637, | |
| "logits/rejected": -2.1554629802703857, | |
| "logps/chosen": -179.8097686767578, | |
| "logps/rejected": -194.11749267578125, | |
| "loss": 0.6739, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5912140607833862, | |
| "rewards/margins": 0.15043725073337555, | |
| "rewards/rejected": -0.7416512966156006, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.109096034333805e-05, | |
| "logits/chosen": -2.4058139324188232, | |
| "logits/rejected": -2.4260308742523193, | |
| "logps/chosen": -213.5891571044922, | |
| "logps/rejected": -187.69091796875, | |
| "loss": 0.7764, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.28837403655052185, | |
| "rewards/margins": -0.10906901955604553, | |
| "rewards/rejected": -0.17930501699447632, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.100316360303376e-05, | |
| "logits/chosen": -2.2765913009643555, | |
| "logits/rejected": -2.2620866298675537, | |
| "logps/chosen": -171.04461669921875, | |
| "logps/rejected": -203.96115112304688, | |
| "loss": 0.6852, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4230179190635681, | |
| "rewards/margins": 0.22685571014881134, | |
| "rewards/rejected": -0.649873673915863, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.091528822336405e-05, | |
| "logits/chosen": -2.064196825027466, | |
| "logits/rejected": -2.0691938400268555, | |
| "logps/chosen": -120.39249420166016, | |
| "logps/rejected": -145.6290283203125, | |
| "loss": 0.6413, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.29033100605010986, | |
| "rewards/margins": 0.18924906849861145, | |
| "rewards/rejected": -0.4795801043510437, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.082733535546601e-05, | |
| "logits/chosen": -2.163872241973877, | |
| "logits/rejected": -2.082183837890625, | |
| "logps/chosen": -161.35110473632812, | |
| "logps/rejected": -148.28184509277344, | |
| "loss": 0.8127, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.40609651803970337, | |
| "rewards/margins": -0.16133582592010498, | |
| "rewards/rejected": -0.24476072192192078, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.073930615149174e-05, | |
| "logits/chosen": -2.113492965698242, | |
| "logits/rejected": -2.0525457859039307, | |
| "logps/chosen": -201.8779754638672, | |
| "logps/rejected": -208.43814086914062, | |
| "loss": 0.7315, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5157983303070068, | |
| "rewards/margins": 0.011869803071022034, | |
| "rewards/rejected": -0.5276680588722229, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.065120176459338e-05, | |
| "logits/chosen": -2.392458438873291, | |
| "logits/rejected": -2.4764301776885986, | |
| "logps/chosen": -184.01644897460938, | |
| "logps/rejected": -201.86964416503906, | |
| "loss": 0.7516, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6866721510887146, | |
| "rewards/margins": -0.07206161320209503, | |
| "rewards/rejected": -0.6146105527877808, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.056302334890786e-05, | |
| "logits/chosen": -2.18422794342041, | |
| "logits/rejected": -2.1438355445861816, | |
| "logps/chosen": -170.0500030517578, | |
| "logps/rejected": -165.754150390625, | |
| "loss": 0.6645, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4873097538948059, | |
| "rewards/margins": 0.21327584981918335, | |
| "rewards/rejected": -0.7005856037139893, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0474772059541935e-05, | |
| "logits/chosen": -2.0524024963378906, | |
| "logits/rejected": -2.108396530151367, | |
| "logps/chosen": -150.6943359375, | |
| "logps/rejected": -150.51748657226562, | |
| "loss": 0.7992, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.42782896757125854, | |
| "rewards/margins": -0.07703240215778351, | |
| "rewards/rejected": -0.35079655051231384, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0386449052556943e-05, | |
| "logits/chosen": -2.0368001461029053, | |
| "logits/rejected": -2.184194803237915, | |
| "logps/chosen": -138.47592163085938, | |
| "logps/rejected": -183.95521545410156, | |
| "loss": 0.5623, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.270087331533432, | |
| "rewards/margins": 0.3816663920879364, | |
| "rewards/rejected": -0.6517537236213684, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.029805548495371e-05, | |
| "logits/chosen": -2.0560860633850098, | |
| "logits/rejected": -1.949674367904663, | |
| "logps/chosen": -203.41712951660156, | |
| "logps/rejected": -180.80010986328125, | |
| "loss": 0.6875, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.5640463829040527, | |
| "rewards/margins": 0.12875407934188843, | |
| "rewards/rejected": -0.6928005218505859, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0209592514657365e-05, | |
| "logits/chosen": -2.1493146419525146, | |
| "logits/rejected": -2.1944329738616943, | |
| "logps/chosen": -168.15786743164062, | |
| "logps/rejected": -182.1389617919922, | |
| "loss": 0.9768, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4937537908554077, | |
| "rewards/margins": -0.398692786693573, | |
| "rewards/rejected": -0.09506092220544815, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0121061300502213e-05, | |
| "logits/chosen": -2.2222800254821777, | |
| "logits/rejected": -2.1859827041625977, | |
| "logps/chosen": -175.8221893310547, | |
| "logps/rejected": -168.3649139404297, | |
| "loss": 0.7054, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4064522981643677, | |
| "rewards/margins": 0.054148584604263306, | |
| "rewards/rejected": -0.46060091257095337, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0032463002216505e-05, | |
| "logits/chosen": -2.137861967086792, | |
| "logits/rejected": -2.1612131595611572, | |
| "logps/chosen": -193.35569763183594, | |
| "logps/rejected": -207.08152770996094, | |
| "loss": 0.818, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.35235530138015747, | |
| "rewards/margins": -0.1532372683286667, | |
| "rewards/rejected": -0.1991180181503296, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.9943798780407288e-05, | |
| "logits/chosen": -2.2527735233306885, | |
| "logits/rejected": -2.2115044593811035, | |
| "logps/chosen": -213.3223114013672, | |
| "logps/rejected": -219.9688720703125, | |
| "loss": 0.7714, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.6288760900497437, | |
| "rewards/margins": -0.022522225975990295, | |
| "rewards/rejected": -0.6063538789749146, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9855069796545186e-05, | |
| "logits/chosen": -2.100188732147217, | |
| "logits/rejected": -2.1022632122039795, | |
| "logps/chosen": -155.2397003173828, | |
| "logps/rejected": -164.0891571044922, | |
| "loss": 0.6691, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.33306682109832764, | |
| "rewards/margins": 0.1229613795876503, | |
| "rewards/rejected": -0.4560282230377197, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9766277212949172e-05, | |
| "logits/chosen": -2.2691118717193604, | |
| "logits/rejected": -2.3205344676971436, | |
| "logps/chosen": -198.963623046875, | |
| "logps/rejected": -213.0981903076172, | |
| "loss": 0.8591, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.6136840581893921, | |
| "rewards/margins": -0.21531665325164795, | |
| "rewards/rejected": -0.39836740493774414, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9677422192771365e-05, | |
| "logits/chosen": -2.1816768646240234, | |
| "logits/rejected": -2.1574275493621826, | |
| "logps/chosen": -158.79043579101562, | |
| "logps/rejected": -171.4162139892578, | |
| "loss": 0.6841, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.46722474694252014, | |
| "rewards/margins": 0.09585972875356674, | |
| "rewards/rejected": -0.5630844831466675, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9588505899981756e-05, | |
| "logits/chosen": -2.0480222702026367, | |
| "logits/rejected": -2.047271490097046, | |
| "logps/chosen": -129.3850555419922, | |
| "logps/rejected": -154.4320068359375, | |
| "loss": 0.524, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.15025204420089722, | |
| "rewards/margins": 0.4296504855155945, | |
| "rewards/rejected": -0.5799025297164917, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9499529499353024e-05, | |
| "logits/chosen": -2.1488020420074463, | |
| "logits/rejected": -2.096888780593872, | |
| "logps/chosen": -208.1055145263672, | |
| "logps/rejected": -179.47232055664062, | |
| "loss": 0.7535, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4541819393634796, | |
| "rewards/margins": -0.024926690384745598, | |
| "rewards/rejected": -0.42925530672073364, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9410494156445216e-05, | |
| "logits/chosen": -1.9598826169967651, | |
| "logits/rejected": -1.9990031719207764, | |
| "logps/chosen": -150.49160766601562, | |
| "logps/rejected": -166.03346252441406, | |
| "loss": 0.6391, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.43558382987976074, | |
| "rewards/margins": 0.15237781405448914, | |
| "rewards/rejected": -0.5879616737365723, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9321401037590502e-05, | |
| "logits/chosen": -2.167135238647461, | |
| "logits/rejected": -2.114692211151123, | |
| "logps/chosen": -234.1712188720703, | |
| "logps/rejected": -230.25656127929688, | |
| "loss": 0.7597, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.40359535813331604, | |
| "rewards/margins": -0.039054736495018005, | |
| "rewards/rejected": -0.36454063653945923, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.923225130987791e-05, | |
| "logits/chosen": -2.0620713233947754, | |
| "logits/rejected": -2.0893771648406982, | |
| "logps/chosen": -145.77232360839844, | |
| "logps/rejected": -163.47850036621094, | |
| "loss": 0.5631, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3432729244232178, | |
| "rewards/margins": 0.3810776472091675, | |
| "rewards/rejected": -0.7243505716323853, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9143046141138015e-05, | |
| "logits/chosen": -2.12490177154541, | |
| "logits/rejected": -2.1528639793395996, | |
| "logps/chosen": -146.39840698242188, | |
| "logps/rejected": -183.64056396484375, | |
| "loss": 0.6113, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.24956199526786804, | |
| "rewards/margins": 0.3226763904094696, | |
| "rewards/rejected": -0.5722384452819824, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.9053786699927642e-05, | |
| "logits/chosen": -2.079617738723755, | |
| "logits/rejected": -2.130861759185791, | |
| "logps/chosen": -173.88525390625, | |
| "logps/rejected": -163.13998413085938, | |
| "loss": 0.7147, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2581624984741211, | |
| "rewards/margins": 0.087901271879673, | |
| "rewards/rejected": -0.3460637629032135, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8964474155514588e-05, | |
| "logits/chosen": -1.984022617340088, | |
| "logits/rejected": -2.0337073802948, | |
| "logps/chosen": -159.10443115234375, | |
| "logps/rejected": -184.33538818359375, | |
| "loss": 0.6007, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.34940189123153687, | |
| "rewards/margins": 0.2434595674276352, | |
| "rewards/rejected": -0.5928614139556885, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8875109677862272e-05, | |
| "logits/chosen": -2.155977487564087, | |
| "logits/rejected": -2.2557568550109863, | |
| "logps/chosen": -181.8987579345703, | |
| "logps/rejected": -184.79747009277344, | |
| "loss": 0.7536, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.5606122016906738, | |
| "rewards/margins": -0.04494345188140869, | |
| "rewards/rejected": -0.5156688094139099, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.878569443761442e-05, | |
| "logits/chosen": -2.160471200942993, | |
| "logits/rejected": -2.194584846496582, | |
| "logps/chosen": -159.53195190429688, | |
| "logps/rejected": -159.7945556640625, | |
| "loss": 0.6753, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.46481114625930786, | |
| "rewards/margins": 0.08550146222114563, | |
| "rewards/rejected": -0.5503125786781311, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8696229606079722e-05, | |
| "logits/chosen": -2.0540730953216553, | |
| "logits/rejected": -2.092503070831299, | |
| "logps/chosen": -153.29518127441406, | |
| "logps/rejected": -164.7957305908203, | |
| "loss": 0.6701, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2695135772228241, | |
| "rewards/margins": 0.10369019210338593, | |
| "rewards/rejected": -0.37320375442504883, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8606716355216523e-05, | |
| "logits/chosen": -2.26692795753479, | |
| "logits/rejected": -2.271298408508301, | |
| "logps/chosen": -146.24447631835938, | |
| "logps/rejected": -152.31927490234375, | |
| "loss": 0.6907, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.16566777229309082, | |
| "rewards/margins": 0.08077192306518555, | |
| "rewards/rejected": -0.24643969535827637, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8517155857617405e-05, | |
| "logits/chosen": -2.247234344482422, | |
| "logits/rejected": -2.256042242050171, | |
| "logps/chosen": -169.71026611328125, | |
| "logps/rejected": -180.32542419433594, | |
| "loss": 0.6593, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.2998695969581604, | |
| "rewards/margins": 0.12919357419013977, | |
| "rewards/rejected": -0.4290631413459778, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8427549286493904e-05, | |
| "logits/chosen": -2.1715965270996094, | |
| "logits/rejected": -2.1411590576171875, | |
| "logps/chosen": -177.94264221191406, | |
| "logps/rejected": -186.37371826171875, | |
| "loss": 0.6029, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.3689126968383789, | |
| "rewards/margins": 0.34154629707336426, | |
| "rewards/rejected": -0.7104589939117432, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.833789781566109e-05, | |
| "logits/chosen": -2.0405919551849365, | |
| "logits/rejected": -2.1091623306274414, | |
| "logps/chosen": -153.8590545654297, | |
| "logps/rejected": -166.40802001953125, | |
| "loss": 0.8301, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.4923684298992157, | |
| "rewards/margins": -0.17652876675128937, | |
| "rewards/rejected": -0.3158397078514099, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8248202619522192e-05, | |
| "logits/chosen": -2.1246650218963623, | |
| "logits/rejected": -2.1084625720977783, | |
| "logps/chosen": -209.72300720214844, | |
| "logps/rejected": -196.4936065673828, | |
| "loss": 0.7741, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.5271788835525513, | |
| "rewards/margins": -0.10752873122692108, | |
| "rewards/rejected": -0.419650137424469, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.8158464873053237e-05, | |
| "logits/chosen": -2.069751024246216, | |
| "logits/rejected": -2.1570510864257812, | |
| "logps/chosen": -146.4033966064453, | |
| "logps/rejected": -171.73431396484375, | |
| "loss": 0.5828, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.20372886955738068, | |
| "rewards/margins": 0.34476152062416077, | |
| "rewards/rejected": -0.5484904050827026, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.8068685751787636e-05, | |
| "logits/chosen": -2.2205162048339844, | |
| "logits/rejected": -2.2368240356445312, | |
| "logps/chosen": -188.92645263671875, | |
| "logps/rejected": -199.38587951660156, | |
| "loss": 0.694, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.4027389585971832, | |
| "rewards/margins": 0.1017821878194809, | |
| "rewards/rejected": -0.5045211315155029, | |
| "step": 500 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 968, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |