| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9994666666666666, | |
| "eval_steps": 500, | |
| "global_step": 937, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.319148936170213e-08, | |
| "logits/chosen": 0.4053989052772522, | |
| "logits/rejected": 0.1312936246395111, | |
| "logps/chosen": -434.00537109375, | |
| "logps/rejected": -516.5983276367188, | |
| "loss": 0.309, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.319148936170213e-07, | |
| "logits/chosen": 0.15412023663520813, | |
| "logits/rejected": 0.2926546633243561, | |
| "logps/chosen": -365.67779541015625, | |
| "logps/rejected": -353.0097961425781, | |
| "loss": 0.3148, | |
| "rewards/accuracies": 0.2777777910232544, | |
| "rewards/chosen": -0.0005839762161485851, | |
| "rewards/margins": -9.90507487586001e-06, | |
| "rewards/rejected": -0.0005740711931139231, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.12003284692764282, | |
| "logits/rejected": 0.2044968158006668, | |
| "logps/chosen": -340.22833251953125, | |
| "logps/rejected": -348.37213134765625, | |
| "loss": 0.3073, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": -0.0007841844344511628, | |
| "rewards/margins": -0.00017091934569180012, | |
| "rewards/rejected": -0.0006132650305517018, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.2551342844963074, | |
| "logits/rejected": 0.2518610954284668, | |
| "logps/chosen": -383.241455078125, | |
| "logps/rejected": -364.0563049316406, | |
| "loss": 0.2998, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.0006543128984048963, | |
| "rewards/margins": 3.304009396742913e-06, | |
| "rewards/rejected": -0.0006576168816536665, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.15533123910427094, | |
| "logits/rejected": 0.27131038904190063, | |
| "logps/chosen": -402.34442138671875, | |
| "logps/rejected": -386.5889587402344, | |
| "loss": 0.3147, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.001332725747488439, | |
| "rewards/margins": -8.493573113810271e-05, | |
| "rewards/rejected": -0.0012477899435907602, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.2343147248029709, | |
| "logits/rejected": 0.27498680353164673, | |
| "logps/chosen": -442.1997985839844, | |
| "logps/rejected": -432.7005920410156, | |
| "loss": 0.3086, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -0.0012903345050290227, | |
| "rewards/margins": 0.0002948477049358189, | |
| "rewards/rejected": -0.0015851823845878243, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.1715986281633377, | |
| "logits/rejected": 0.3192042112350464, | |
| "logps/chosen": -409.2780456542969, | |
| "logps/rejected": -413.3976135253906, | |
| "loss": 0.311, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.00036035641096532345, | |
| "rewards/margins": 0.0003330351901240647, | |
| "rewards/rejected": -0.0006933916010893881, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.1239662617444992, | |
| "logits/rejected": 0.22276969254016876, | |
| "logps/chosen": -356.0749206542969, | |
| "logps/rejected": -358.4507751464844, | |
| "loss": 0.3075, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.0014285400975495577, | |
| "rewards/margins": 0.0010666290763765574, | |
| "rewards/rejected": -0.002495169173926115, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.14190419018268585, | |
| "logits/rejected": 0.2491893321275711, | |
| "logps/chosen": -399.1524963378906, | |
| "logps/rejected": -387.684326171875, | |
| "loss": 0.3052, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.00629749009385705, | |
| "rewards/margins": 0.0015292944153770804, | |
| "rewards/rejected": -0.007826784625649452, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.23625846207141876, | |
| "logits/rejected": 0.17152468860149384, | |
| "logps/chosen": -408.34222412109375, | |
| "logps/rejected": -431.87689208984375, | |
| "loss": 0.3018, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.01540874969214201, | |
| "rewards/margins": 0.0023751407861709595, | |
| "rewards/rejected": -0.017783891409635544, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.15845081210136414, | |
| "logits/rejected": 0.16451677680015564, | |
| "logps/chosen": -451.0520935058594, | |
| "logps/rejected": -446.60052490234375, | |
| "loss": 0.314, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.03385753557085991, | |
| "rewards/margins": 0.003108574775978923, | |
| "rewards/rejected": -0.03696610778570175, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.2290416657924652, | |
| "logits/rejected": 0.15046942234039307, | |
| "logps/chosen": -470.06536865234375, | |
| "logps/rejected": -457.4624938964844, | |
| "loss": 0.3177, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.05096244812011719, | |
| "rewards/margins": 0.006424424704164267, | |
| "rewards/rejected": -0.057386863976716995, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.0766684040427208, | |
| "logits/rejected": 0.24015113711357117, | |
| "logps/chosen": -494.6429748535156, | |
| "logps/rejected": -534.6770629882812, | |
| "loss": 0.2842, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.07989172637462616, | |
| "rewards/margins": 0.012701849453151226, | |
| "rewards/rejected": -0.09259358048439026, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.1341143548488617, | |
| "logits/rejected": 0.257051020860672, | |
| "logps/chosen": -523.8453369140625, | |
| "logps/rejected": -525.6238403320312, | |
| "loss": 0.297, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.11127883195877075, | |
| "rewards/margins": 0.015445197932422161, | |
| "rewards/rejected": -0.1267240345478058, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.1251862347126007, | |
| "logits/rejected": 0.18638013303279877, | |
| "logps/chosen": -528.09912109375, | |
| "logps/rejected": -526.5968627929688, | |
| "loss": 0.2885, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": -0.1439468413591385, | |
| "rewards/margins": 0.01796194352209568, | |
| "rewards/rejected": -0.16190877556800842, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.004607650451362133, | |
| "logits/rejected": 0.046932995319366455, | |
| "logps/chosen": -519.865478515625, | |
| "logps/rejected": -557.3140869140625, | |
| "loss": 0.2817, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.1371057778596878, | |
| "rewards/margins": 0.034668728709220886, | |
| "rewards/rejected": -0.1717745065689087, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.07374308258295059, | |
| "logits/rejected": 0.1448010951280594, | |
| "logps/chosen": -570.7451171875, | |
| "logps/rejected": -618.2618408203125, | |
| "loss": 0.2727, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.14755412936210632, | |
| "rewards/margins": 0.05937732011079788, | |
| "rewards/rejected": -0.2069314420223236, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.008384035900235176, | |
| "logits/rejected": 0.18068811297416687, | |
| "logps/chosen": -607.10693359375, | |
| "logps/rejected": -628.9288330078125, | |
| "loss": 0.273, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.16640765964984894, | |
| "rewards/margins": 0.04880762845277786, | |
| "rewards/rejected": -0.2152152955532074, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.11068884283304214, | |
| "logits/rejected": 0.1132630854845047, | |
| "logps/chosen": -507.98583984375, | |
| "logps/rejected": -551.5580444335938, | |
| "loss": 0.2848, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.12697137892246246, | |
| "rewards/margins": 0.04405537620186806, | |
| "rewards/rejected": -0.17102673649787903, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.033597201108932495, | |
| "logits/rejected": 0.10967175662517548, | |
| "logps/chosen": -624.1072998046875, | |
| "logps/rejected": -641.2291870117188, | |
| "loss": 0.2779, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.13993802666664124, | |
| "rewards/margins": 0.0485340878367424, | |
| "rewards/rejected": -0.18847212195396423, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": -0.03184026479721069, | |
| "logits/rejected": 0.09308318793773651, | |
| "logps/chosen": -545.9130249023438, | |
| "logps/rejected": -619.6661376953125, | |
| "loss": 0.277, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.14176079630851746, | |
| "rewards/margins": 0.05656677484512329, | |
| "rewards/rejected": -0.19832755625247955, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.009775908663868904, | |
| "logits/rejected": -0.003572815330699086, | |
| "logps/chosen": -554.5719604492188, | |
| "logps/rejected": -616.5828857421875, | |
| "loss": 0.261, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.15296652913093567, | |
| "rewards/margins": 0.05833062529563904, | |
| "rewards/rejected": -0.2112971544265747, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": -0.05359144136309624, | |
| "logits/rejected": -0.09271421283483505, | |
| "logps/chosen": -588.1046142578125, | |
| "logps/rejected": -641.3451538085938, | |
| "loss": 0.2567, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.18619795143604279, | |
| "rewards/margins": 0.058625251054763794, | |
| "rewards/rejected": -0.24482321739196777, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": -0.09103509038686752, | |
| "logits/rejected": -0.07337954640388489, | |
| "logps/chosen": -569.4376220703125, | |
| "logps/rejected": -632.4283447265625, | |
| "loss": 0.3067, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.1694985032081604, | |
| "rewards/margins": 0.062124162912368774, | |
| "rewards/rejected": -0.23162266612052917, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": -0.18483319878578186, | |
| "logits/rejected": -0.019147302955389023, | |
| "logps/chosen": -496.79150390625, | |
| "logps/rejected": -542.9866333007812, | |
| "loss": 0.2758, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.1192159652709961, | |
| "rewards/margins": 0.04292690008878708, | |
| "rewards/rejected": -0.16214287281036377, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": -0.11058203876018524, | |
| "logits/rejected": -0.04781871661543846, | |
| "logps/chosen": -514.0642700195312, | |
| "logps/rejected": -546.6673583984375, | |
| "loss": 0.2877, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.09990507364273071, | |
| "rewards/margins": 0.04675906524062157, | |
| "rewards/rejected": -0.14666414260864258, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": -0.08988751471042633, | |
| "logits/rejected": -0.06392081081867218, | |
| "logps/chosen": -541.5545654296875, | |
| "logps/rejected": -603.2689819335938, | |
| "loss": 0.2804, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.11030902713537216, | |
| "rewards/margins": 0.0489293709397316, | |
| "rewards/rejected": -0.15923838317394257, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": -0.14757739007472992, | |
| "logits/rejected": -0.11037661880254745, | |
| "logps/chosen": -569.8583984375, | |
| "logps/rejected": -653.552734375, | |
| "loss": 0.2831, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.13597574830055237, | |
| "rewards/margins": 0.06436828523874283, | |
| "rewards/rejected": -0.2003440111875534, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": -0.15325765311717987, | |
| "logits/rejected": -0.08648681640625, | |
| "logps/chosen": -532.4110107421875, | |
| "logps/rejected": -580.8661499023438, | |
| "loss": 0.2633, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.1242569088935852, | |
| "rewards/margins": 0.06257221847772598, | |
| "rewards/rejected": -0.18682913482189178, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": -0.159153550863266, | |
| "logits/rejected": -0.1052144318819046, | |
| "logps/chosen": -548.8265380859375, | |
| "logps/rejected": -647.3924560546875, | |
| "loss": 0.2877, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.14587877690792084, | |
| "rewards/margins": 0.06854279339313507, | |
| "rewards/rejected": -0.2144215852022171, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": -0.13682904839515686, | |
| "logits/rejected": -0.15670131146907806, | |
| "logps/chosen": -573.5653076171875, | |
| "logps/rejected": -609.18896484375, | |
| "loss": 0.2689, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.15031126141548157, | |
| "rewards/margins": 0.07083144038915634, | |
| "rewards/rejected": -0.2211426943540573, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": -0.20464687049388885, | |
| "logits/rejected": -0.07677153497934341, | |
| "logps/chosen": -494.4937438964844, | |
| "logps/rejected": -564.7810668945312, | |
| "loss": 0.2712, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.11489574611186981, | |
| "rewards/margins": 0.07067451626062393, | |
| "rewards/rejected": -0.18557025492191315, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": -0.15810438990592957, | |
| "logits/rejected": -0.12246428430080414, | |
| "logps/chosen": -530.6107788085938, | |
| "logps/rejected": -583.3417358398438, | |
| "loss": 0.2944, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.14831142127513885, | |
| "rewards/margins": 0.06477416306734085, | |
| "rewards/rejected": -0.2130856066942215, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": -0.13243547081947327, | |
| "logits/rejected": -0.12164945900440216, | |
| "logps/chosen": -611.0227661132812, | |
| "logps/rejected": -638.0067138671875, | |
| "loss": 0.2708, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.173904150724411, | |
| "rewards/margins": 0.058606814593076706, | |
| "rewards/rejected": -0.2325109988451004, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": -0.09038461744785309, | |
| "logits/rejected": -0.08104994148015976, | |
| "logps/chosen": -577.1747436523438, | |
| "logps/rejected": -659.236328125, | |
| "loss": 0.2662, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.15762028098106384, | |
| "rewards/margins": 0.07804208993911743, | |
| "rewards/rejected": -0.23566241562366486, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": -0.15925371646881104, | |
| "logits/rejected": -0.02752949297428131, | |
| "logps/chosen": -556.4603881835938, | |
| "logps/rejected": -596.9464111328125, | |
| "loss": 0.2753, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.11931456625461578, | |
| "rewards/margins": 0.045837145298719406, | |
| "rewards/rejected": -0.1651517152786255, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": -0.2364901602268219, | |
| "logits/rejected": -0.1447964608669281, | |
| "logps/chosen": -554.9327392578125, | |
| "logps/rejected": -575.6502075195312, | |
| "loss": 0.2698, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.10456685721874237, | |
| "rewards/margins": 0.05687998980283737, | |
| "rewards/rejected": -0.16144683957099915, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": -0.1706862449645996, | |
| "logits/rejected": -0.1711297631263733, | |
| "logps/chosen": -512.93115234375, | |
| "logps/rejected": -578.0069580078125, | |
| "loss": 0.2742, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.11323670297861099, | |
| "rewards/margins": 0.06127854436635971, | |
| "rewards/rejected": -0.1745152473449707, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": -0.2998970150947571, | |
| "logits/rejected": -0.14411643147468567, | |
| "logps/chosen": -507.00347900390625, | |
| "logps/rejected": -539.485595703125, | |
| "loss": 0.2566, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.10295897722244263, | |
| "rewards/margins": 0.05676443502306938, | |
| "rewards/rejected": -0.1597234159708023, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": -0.2854919731616974, | |
| "logits/rejected": -0.21910324692726135, | |
| "logps/chosen": -544.7410888671875, | |
| "logps/rejected": -592.5999145507812, | |
| "loss": 0.2625, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.12320928275585175, | |
| "rewards/margins": 0.03932071477174759, | |
| "rewards/rejected": -0.16252999007701874, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": -0.17541547119617462, | |
| "logits/rejected": -0.17069751024246216, | |
| "logps/chosen": -524.9605102539062, | |
| "logps/rejected": -575.748046875, | |
| "loss": 0.2735, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.10180839151144028, | |
| "rewards/margins": 0.05412058159708977, | |
| "rewards/rejected": -0.15592898428440094, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": -0.1704898327589035, | |
| "logits/rejected": -0.17921459674835205, | |
| "logps/chosen": -513.3040161132812, | |
| "logps/rejected": -590.165771484375, | |
| "loss": 0.2485, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.09149526059627533, | |
| "rewards/margins": 0.05739337205886841, | |
| "rewards/rejected": -0.14888863265514374, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": -0.18199101090431213, | |
| "logits/rejected": -0.1548750400543213, | |
| "logps/chosen": -554.3550415039062, | |
| "logps/rejected": -620.1253662109375, | |
| "loss": 0.2814, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.10769498348236084, | |
| "rewards/margins": 0.06901998817920685, | |
| "rewards/rejected": -0.1767149716615677, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": -0.2242196500301361, | |
| "logits/rejected": -0.23621216416358948, | |
| "logps/chosen": -472.6395568847656, | |
| "logps/rejected": -534.2669677734375, | |
| "loss": 0.2554, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.10465536266565323, | |
| "rewards/margins": 0.06357412785291672, | |
| "rewards/rejected": -0.16822949051856995, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": -0.24027328193187714, | |
| "logits/rejected": -0.22386594116687775, | |
| "logps/chosen": -576.7644653320312, | |
| "logps/rejected": -606.2843627929688, | |
| "loss": 0.2728, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.1252559870481491, | |
| "rewards/margins": 0.03980395942926407, | |
| "rewards/rejected": -0.1650599241256714, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": -0.24792134761810303, | |
| "logits/rejected": -0.20381967723369598, | |
| "logps/chosen": -540.4932250976562, | |
| "logps/rejected": -590.7255249023438, | |
| "loss": 0.2625, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.11483976989984512, | |
| "rewards/margins": 0.06688660383224487, | |
| "rewards/rejected": -0.1817263662815094, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": -0.22240272164344788, | |
| "logits/rejected": -0.16780824959278107, | |
| "logps/chosen": -575.5987548828125, | |
| "logps/rejected": -615.41259765625, | |
| "loss": 0.274, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.1371522694826126, | |
| "rewards/margins": 0.05847863107919693, | |
| "rewards/rejected": -0.19563090801239014, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": -0.22179889678955078, | |
| "logits/rejected": -0.2187281847000122, | |
| "logps/chosen": -551.1289672851562, | |
| "logps/rejected": -633.8525390625, | |
| "loss": 0.2638, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.13685192167758942, | |
| "rewards/margins": 0.08387337625026703, | |
| "rewards/rejected": -0.22072529792785645, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": -0.31543129682540894, | |
| "logits/rejected": -0.1996282935142517, | |
| "logps/chosen": -528.4293212890625, | |
| "logps/rejected": -616.8591918945312, | |
| "loss": 0.2639, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.14036959409713745, | |
| "rewards/margins": 0.08370579779148102, | |
| "rewards/rejected": -0.22407540678977966, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": -0.1916017085313797, | |
| "logits/rejected": -0.20610889792442322, | |
| "logps/chosen": -513.9475708007812, | |
| "logps/rejected": -600.5337524414062, | |
| "loss": 0.2788, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.13624821603298187, | |
| "rewards/margins": 0.07949782907962799, | |
| "rewards/rejected": -0.21574604511260986, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": -0.25276193022727966, | |
| "logits/rejected": -0.27581551671028137, | |
| "logps/chosen": -503.4148864746094, | |
| "logps/rejected": -561.3239135742188, | |
| "loss": 0.2674, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.13355006277561188, | |
| "rewards/margins": 0.05300268530845642, | |
| "rewards/rejected": -0.1865527629852295, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": -0.24862711131572723, | |
| "logits/rejected": -0.2061503380537033, | |
| "logps/chosen": -596.1258544921875, | |
| "logps/rejected": -645.7933959960938, | |
| "loss": 0.2702, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.15105004608631134, | |
| "rewards/margins": 0.06355919688940048, | |
| "rewards/rejected": -0.21460923552513123, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": -0.29833105206489563, | |
| "logits/rejected": -0.1915448009967804, | |
| "logps/chosen": -537.9877319335938, | |
| "logps/rejected": -557.24169921875, | |
| "loss": 0.2662, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.12659652531147003, | |
| "rewards/margins": 0.051892559975385666, | |
| "rewards/rejected": -0.178489089012146, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": -0.15996907651424408, | |
| "logits/rejected": -0.19643227756023407, | |
| "logps/chosen": -550.2649536132812, | |
| "logps/rejected": -597.158203125, | |
| "loss": 0.2658, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.13815242052078247, | |
| "rewards/margins": 0.05568670108914375, | |
| "rewards/rejected": -0.19383911788463593, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.2864486575126648, | |
| "logits/rejected": -0.20076756179332733, | |
| "logps/chosen": -528.1815185546875, | |
| "logps/rejected": -566.8001098632812, | |
| "loss": 0.2851, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -0.12378416955471039, | |
| "rewards/margins": 0.05396197363734245, | |
| "rewards/rejected": -0.17774614691734314, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": -0.3104463219642639, | |
| "logits/rejected": -0.18759891390800476, | |
| "logps/chosen": -548.5665283203125, | |
| "logps/rejected": -590.8157958984375, | |
| "loss": 0.262, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.1302889883518219, | |
| "rewards/margins": 0.04864712059497833, | |
| "rewards/rejected": -0.17893610894680023, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": -0.2759536802768707, | |
| "logits/rejected": -0.2688484787940979, | |
| "logps/chosen": -513.609375, | |
| "logps/rejected": -596.7303466796875, | |
| "loss": 0.2862, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.13550987839698792, | |
| "rewards/margins": 0.053082652390003204, | |
| "rewards/rejected": -0.18859252333641052, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": -0.3110070824623108, | |
| "logits/rejected": -0.294691264629364, | |
| "logps/chosen": -499.83721923828125, | |
| "logps/rejected": -553.9511108398438, | |
| "loss": 0.2786, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.1329517960548401, | |
| "rewards/margins": 0.05472971126437187, | |
| "rewards/rejected": -0.18768151104450226, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": -0.2652963697910309, | |
| "logits/rejected": -0.2378832995891571, | |
| "logps/chosen": -536.1483154296875, | |
| "logps/rejected": -593.2440795898438, | |
| "loss": 0.269, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.1256466507911682, | |
| "rewards/margins": 0.07022566348314285, | |
| "rewards/rejected": -0.19587230682373047, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": -0.2916659712791443, | |
| "logits/rejected": -0.17712509632110596, | |
| "logps/chosen": -560.8721923828125, | |
| "logps/rejected": -634.2991943359375, | |
| "loss": 0.2612, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.1301390826702118, | |
| "rewards/margins": 0.08458835631608963, | |
| "rewards/rejected": -0.21472744643688202, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": -0.2919694781303406, | |
| "logits/rejected": -0.2693978548049927, | |
| "logps/chosen": -478.8960876464844, | |
| "logps/rejected": -579.1296997070312, | |
| "loss": 0.2587, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.1212470531463623, | |
| "rewards/margins": 0.07617990672588348, | |
| "rewards/rejected": -0.1974269449710846, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": -0.2872418761253357, | |
| "logits/rejected": -0.3003498911857605, | |
| "logps/chosen": -541.9686279296875, | |
| "logps/rejected": -597.8888549804688, | |
| "loss": 0.2804, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.13385756313800812, | |
| "rewards/margins": 0.07602544128894806, | |
| "rewards/rejected": -0.20988301932811737, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": -0.19217848777770996, | |
| "logits/rejected": -0.3079163134098053, | |
| "logps/chosen": -527.4349365234375, | |
| "logps/rejected": -640.5628051757812, | |
| "loss": 0.2827, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.1333102285861969, | |
| "rewards/margins": 0.08168620616197586, | |
| "rewards/rejected": -0.21499641239643097, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": -0.3484002649784088, | |
| "logits/rejected": -0.33647680282592773, | |
| "logps/chosen": -478.8099670410156, | |
| "logps/rejected": -542.4144287109375, | |
| "loss": 0.2699, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.10638388246297836, | |
| "rewards/margins": 0.0811399295926094, | |
| "rewards/rejected": -0.18752381205558777, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": -0.3240329623222351, | |
| "logits/rejected": -0.28249433636665344, | |
| "logps/chosen": -507.2118225097656, | |
| "logps/rejected": -577.4991455078125, | |
| "loss": 0.2603, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.11796662956476212, | |
| "rewards/margins": 0.08652862906455994, | |
| "rewards/rejected": -0.20449523627758026, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.31510046124458313, | |
| "logits/rejected": -0.13614080846309662, | |
| "logps/chosen": -538.2633056640625, | |
| "logps/rejected": -618.5576782226562, | |
| "loss": 0.2772, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.11485198885202408, | |
| "rewards/margins": 0.09617733955383301, | |
| "rewards/rejected": -0.2110293209552765, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": -0.3489423096179962, | |
| "logits/rejected": -0.20355625450611115, | |
| "logps/chosen": -573.0088500976562, | |
| "logps/rejected": -612.1022338867188, | |
| "loss": 0.2873, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.14457251131534576, | |
| "rewards/margins": 0.057286668568849564, | |
| "rewards/rejected": -0.20185919106006622, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": -0.171179860830307, | |
| "logits/rejected": -0.2416304349899292, | |
| "logps/chosen": -521.5185546875, | |
| "logps/rejected": -576.9376220703125, | |
| "loss": 0.276, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.13226917386054993, | |
| "rewards/margins": 0.05868411809206009, | |
| "rewards/rejected": -0.19095326960086823, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": -0.22147266566753387, | |
| "logits/rejected": -0.22640736401081085, | |
| "logps/chosen": -479.4178161621094, | |
| "logps/rejected": -519.9364013671875, | |
| "loss": 0.2594, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.10929121822118759, | |
| "rewards/margins": 0.0537792444229126, | |
| "rewards/rejected": -0.1630704551935196, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": -0.28457021713256836, | |
| "logits/rejected": -0.1854095160961151, | |
| "logps/chosen": -514.6656494140625, | |
| "logps/rejected": -569.9134521484375, | |
| "loss": 0.2667, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.11874970048666, | |
| "rewards/margins": 0.060789503157138824, | |
| "rewards/rejected": -0.17953920364379883, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": -0.30244481563568115, | |
| "logits/rejected": -0.25662168860435486, | |
| "logps/chosen": -507.129150390625, | |
| "logps/rejected": -575.9667358398438, | |
| "loss": 0.2562, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.11172547191381454, | |
| "rewards/margins": 0.06496167182922363, | |
| "rewards/rejected": -0.17668713629245758, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.280193954706192, | |
| "logits/rejected": -0.17224326729774475, | |
| "logps/chosen": -535.2366943359375, | |
| "logps/rejected": -601.5757446289062, | |
| "loss": 0.2549, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.12534330785274506, | |
| "rewards/margins": 0.06499765813350677, | |
| "rewards/rejected": -0.19034096598625183, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": -0.228287011384964, | |
| "logits/rejected": -0.22289884090423584, | |
| "logps/chosen": -550.7459106445312, | |
| "logps/rejected": -601.8172607421875, | |
| "loss": 0.2557, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.1304478794336319, | |
| "rewards/margins": 0.06193193793296814, | |
| "rewards/rejected": -0.19237980246543884, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": -0.23304042220115662, | |
| "logits/rejected": -0.2573367953300476, | |
| "logps/chosen": -546.1334838867188, | |
| "logps/rejected": -603.4042358398438, | |
| "loss": 0.253, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.12139827013015747, | |
| "rewards/margins": 0.07925884425640106, | |
| "rewards/rejected": -0.20065709948539734, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": -0.28443416953086853, | |
| "logits/rejected": -0.18750058114528656, | |
| "logps/chosen": -487.604248046875, | |
| "logps/rejected": -528.2503051757812, | |
| "loss": 0.2834, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.10632891952991486, | |
| "rewards/margins": 0.07151490449905396, | |
| "rewards/rejected": -0.17784383893013, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": -0.26037511229515076, | |
| "logits/rejected": -0.2036275863647461, | |
| "logps/chosen": -544.8526611328125, | |
| "logps/rejected": -635.510009765625, | |
| "loss": 0.2576, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.12171368300914764, | |
| "rewards/margins": 0.08451158553361893, | |
| "rewards/rejected": -0.20622527599334717, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": -0.3115929961204529, | |
| "logits/rejected": -0.104661725461483, | |
| "logps/chosen": -510.882080078125, | |
| "logps/rejected": -545.64501953125, | |
| "loss": 0.2533, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.12393975257873535, | |
| "rewards/margins": 0.060543425381183624, | |
| "rewards/rejected": -0.18448317050933838, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": -0.354599267244339, | |
| "logits/rejected": -0.2644585967063904, | |
| "logps/chosen": -556.0206909179688, | |
| "logps/rejected": -593.3365478515625, | |
| "loss": 0.2578, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.12226442992687225, | |
| "rewards/margins": 0.06983228772878647, | |
| "rewards/rejected": -0.1920967400074005, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": -0.29683440923690796, | |
| "logits/rejected": -0.1917039006948471, | |
| "logps/chosen": -565.5325927734375, | |
| "logps/rejected": -639.9913330078125, | |
| "loss": 0.2528, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.128949835896492, | |
| "rewards/margins": 0.07002034783363342, | |
| "rewards/rejected": -0.19897018373012543, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": -0.20871396362781525, | |
| "logits/rejected": -0.18840913474559784, | |
| "logps/chosen": -515.9395751953125, | |
| "logps/rejected": -580.9120483398438, | |
| "loss": 0.2695, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.1138933077454567, | |
| "rewards/margins": 0.06584799289703369, | |
| "rewards/rejected": -0.1797412931919098, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": -0.31950509548187256, | |
| "logits/rejected": -0.20166322588920593, | |
| "logps/chosen": -542.8575439453125, | |
| "logps/rejected": -579.9568481445312, | |
| "loss": 0.2798, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.12461917102336884, | |
| "rewards/margins": 0.0687582939863205, | |
| "rewards/rejected": -0.19337746500968933, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": -0.24710452556610107, | |
| "logits/rejected": -0.285795658826828, | |
| "logps/chosen": -584.72412109375, | |
| "logps/rejected": -667.7575073242188, | |
| "loss": 0.2646, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.13876564800739288, | |
| "rewards/margins": 0.08441019803285599, | |
| "rewards/rejected": -0.22317583858966827, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": -0.34350934624671936, | |
| "logits/rejected": -0.1507629156112671, | |
| "logps/chosen": -555.4998779296875, | |
| "logps/rejected": -603.321533203125, | |
| "loss": 0.2657, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.11694788932800293, | |
| "rewards/margins": 0.09008406102657318, | |
| "rewards/rejected": -0.20703193545341492, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": -0.2246185541152954, | |
| "logits/rejected": -0.23372571170330048, | |
| "logps/chosen": -545.0233154296875, | |
| "logps/rejected": -591.8538208007812, | |
| "loss": 0.2813, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.12621915340423584, | |
| "rewards/margins": 0.06708641350269318, | |
| "rewards/rejected": -0.19330556690692902, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": -0.364227294921875, | |
| "logits/rejected": -0.26786571741104126, | |
| "logps/chosen": -481.79718017578125, | |
| "logps/rejected": -542.220947265625, | |
| "loss": 0.2508, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.11298207938671112, | |
| "rewards/margins": 0.07751892507076263, | |
| "rewards/rejected": -0.19050098955631256, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": -0.33576229214668274, | |
| "logits/rejected": -0.281321257352829, | |
| "logps/chosen": -549.2854614257812, | |
| "logps/rejected": -602.2276611328125, | |
| "loss": 0.2735, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.1300746351480484, | |
| "rewards/margins": 0.06319116055965424, | |
| "rewards/rejected": -0.19326578080654144, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": -0.1330328732728958, | |
| "logits/rejected": -0.2882300913333893, | |
| "logps/chosen": -529.1005859375, | |
| "logps/rejected": -605.9659423828125, | |
| "loss": 0.2781, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.12892162799835205, | |
| "rewards/margins": 0.0663539245724678, | |
| "rewards/rejected": -0.19527556002140045, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": -0.24367213249206543, | |
| "logits/rejected": -0.2970259189605713, | |
| "logps/chosen": -579.6039428710938, | |
| "logps/rejected": -616.4037475585938, | |
| "loss": 0.2589, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.13028362393379211, | |
| "rewards/margins": 0.061786215752363205, | |
| "rewards/rejected": -0.19206984341144562, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": -0.3084169030189514, | |
| "logits/rejected": -0.2781293988227844, | |
| "logps/chosen": -531.304931640625, | |
| "logps/rejected": -618.241943359375, | |
| "loss": 0.2481, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.12809237837791443, | |
| "rewards/margins": 0.09018449485301971, | |
| "rewards/rejected": -0.21827688813209534, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": -0.3312014043331146, | |
| "logits/rejected": -0.25709718465805054, | |
| "logps/chosen": -530.1060791015625, | |
| "logps/rejected": -596.2859497070312, | |
| "loss": 0.2425, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.11436812579631805, | |
| "rewards/margins": 0.08216674625873566, | |
| "rewards/rejected": -0.1965348720550537, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": -0.24755005538463593, | |
| "logits/rejected": -0.3069338798522949, | |
| "logps/chosen": -514.0535888671875, | |
| "logps/rejected": -623.8255615234375, | |
| "loss": 0.2624, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.13177482783794403, | |
| "rewards/margins": 0.08524416387081146, | |
| "rewards/rejected": -0.2170190066099167, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": -0.24237994849681854, | |
| "logits/rejected": -0.26542186737060547, | |
| "logps/chosen": -535.3384399414062, | |
| "logps/rejected": -621.1607055664062, | |
| "loss": 0.2496, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.10499405860900879, | |
| "rewards/margins": 0.08728118240833282, | |
| "rewards/rejected": -0.1922752410173416, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": -0.3247453570365906, | |
| "logits/rejected": -0.34661737084388733, | |
| "logps/chosen": -501.97705078125, | |
| "logps/rejected": -568.5880737304688, | |
| "loss": 0.2714, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.11833710968494415, | |
| "rewards/margins": 0.08402119576931, | |
| "rewards/rejected": -0.20235827565193176, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": -0.27017658948898315, | |
| "logits/rejected": -0.2859312891960144, | |
| "logps/chosen": -512.9287109375, | |
| "logps/rejected": -591.2950439453125, | |
| "loss": 0.2695, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.12850606441497803, | |
| "rewards/margins": 0.07382752746343613, | |
| "rewards/rejected": -0.20233359932899475, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.274457171320788, | |
| "train_runtime": 7836.9626, | |
| "train_samples_per_second": 3.828, | |
| "train_steps_per_second": 0.12 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 937, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |