| { | |
| "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": 0.03003755956888199, | |
| "logits/rejected": 0.14685480296611786, | |
| "logps/chosen": -314.14739990234375, | |
| "logps/rejected": -285.09228515625, | |
| "loss": 0.379, | |
| "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": 0.24652427434921265, | |
| "logits/rejected": 0.310183584690094, | |
| "logps/chosen": -415.2710876464844, | |
| "logps/rejected": -420.6697692871094, | |
| "loss": 0.3666, | |
| "rewards/accuracies": 0.3055555522441864, | |
| "rewards/chosen": -0.0005877503426745534, | |
| "rewards/margins": -1.95145039469935e-05, | |
| "rewards/rejected": -0.000568235875107348, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.2645895481109619, | |
| "logits/rejected": 0.19589334726333618, | |
| "logps/chosen": -365.053466796875, | |
| "logps/rejected": -364.3542785644531, | |
| "loss": 0.3639, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": -0.0009052521781995893, | |
| "rewards/margins": 2.238732122350484e-05, | |
| "rewards/rejected": -0.000927639368455857, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.20655973255634308, | |
| "logits/rejected": 0.2799147963523865, | |
| "logps/chosen": -425.14288330078125, | |
| "logps/rejected": -414.9308166503906, | |
| "loss": 0.3657, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.0010341120650991797, | |
| "rewards/margins": 0.00020249195222277194, | |
| "rewards/rejected": -0.0012366041773930192, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.1586448699235916, | |
| "logits/rejected": 0.3028702735900879, | |
| "logps/chosen": -435.25311279296875, | |
| "logps/rejected": -412.69769287109375, | |
| "loss": 0.3732, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": -0.000818841508589685, | |
| "rewards/margins": 0.0002176587877329439, | |
| "rewards/rejected": -0.0010365003254264593, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.30979618430137634, | |
| "logits/rejected": 0.18890634179115295, | |
| "logps/chosen": -376.4929504394531, | |
| "logps/rejected": -388.2627868652344, | |
| "loss": 0.3709, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.0009732651524245739, | |
| "rewards/margins": 0.00020587551989592612, | |
| "rewards/rejected": -0.0011791407596319914, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.25725987553596497, | |
| "logits/rejected": 0.27524492144584656, | |
| "logps/chosen": -435.30682373046875, | |
| "logps/rejected": -389.1386413574219, | |
| "loss": 0.38, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.0007843401399441063, | |
| "rewards/margins": 0.0004105453845113516, | |
| "rewards/rejected": 0.00037379475543275476, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.19644694030284882, | |
| "logits/rejected": 0.2764621675014496, | |
| "logps/chosen": -445.50042724609375, | |
| "logps/rejected": -425.63714599609375, | |
| "loss": 0.3673, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": 0.004284649156033993, | |
| "rewards/margins": 0.0006289929151535034, | |
| "rewards/rejected": 0.003655656473711133, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.17820295691490173, | |
| "logits/rejected": 0.14415612816810608, | |
| "logps/chosen": -420.62811279296875, | |
| "logps/rejected": -431.13250732421875, | |
| "loss": 0.3782, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.011514987796545029, | |
| "rewards/margins": 0.0015228450065478683, | |
| "rewards/rejected": 0.009992143139243126, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.1115020141005516, | |
| "logits/rejected": 0.2590204179286957, | |
| "logps/chosen": -431.19622802734375, | |
| "logps/rejected": -393.2071228027344, | |
| "loss": 0.3752, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": 0.013366015627980232, | |
| "rewards/margins": 0.0019467566162347794, | |
| "rewards/rejected": 0.011419258080422878, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.17629960179328918, | |
| "logits/rejected": 0.18853217363357544, | |
| "logps/chosen": -391.21258544921875, | |
| "logps/rejected": -405.5911560058594, | |
| "loss": 0.3634, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": 0.006202789954841137, | |
| "rewards/margins": 0.0038450881838798523, | |
| "rewards/rejected": 0.0023577010724693537, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.19157476723194122, | |
| "logits/rejected": 0.2552744746208191, | |
| "logps/chosen": -388.6427001953125, | |
| "logps/rejected": -401.7698974609375, | |
| "loss": 0.371, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.011058819480240345, | |
| "rewards/margins": 0.004346723668277264, | |
| "rewards/rejected": -0.015405543148517609, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.1481851041316986, | |
| "logits/rejected": 0.15830105543136597, | |
| "logps/chosen": -485.09942626953125, | |
| "logps/rejected": -512.1583251953125, | |
| "loss": 0.3596, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.04192991554737091, | |
| "rewards/margins": 0.01182691939175129, | |
| "rewards/rejected": -0.05375684052705765, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.06818593293428421, | |
| "logits/rejected": 0.1832435131072998, | |
| "logps/chosen": -470.56787109375, | |
| "logps/rejected": -495.54296875, | |
| "loss": 0.3416, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.07512780278921127, | |
| "rewards/margins": 0.020879648625850677, | |
| "rewards/rejected": -0.09600745886564255, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.1163097396492958, | |
| "logits/rejected": 0.10711435973644257, | |
| "logps/chosen": -498.22589111328125, | |
| "logps/rejected": -498.1195373535156, | |
| "loss": 0.3582, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.10014933347702026, | |
| "rewards/margins": 0.015763022005558014, | |
| "rewards/rejected": -0.11591235548257828, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.019728729501366615, | |
| "logits/rejected": 0.078994520008564, | |
| "logps/chosen": -495.50775146484375, | |
| "logps/rejected": -522.085693359375, | |
| "loss": 0.3351, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.10014746338129044, | |
| "rewards/margins": 0.027558928355574608, | |
| "rewards/rejected": -0.1277063935995102, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.03191109746694565, | |
| "logits/rejected": 0.026030873879790306, | |
| "logps/chosen": -531.992919921875, | |
| "logps/rejected": -575.7620239257812, | |
| "loss": 0.3317, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.12353821098804474, | |
| "rewards/margins": 0.04307270795106888, | |
| "rewards/rejected": -0.16661091148853302, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.045843735337257385, | |
| "logits/rejected": -0.08834481984376907, | |
| "logps/chosen": -522.8604736328125, | |
| "logps/rejected": -612.8219604492188, | |
| "loss": 0.3325, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.12141416221857071, | |
| "rewards/margins": 0.06917469203472137, | |
| "rewards/rejected": -0.1905888319015503, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.002990189241245389, | |
| "logits/rejected": 0.019473141059279442, | |
| "logps/chosen": -489.2099609375, | |
| "logps/rejected": -574.8321533203125, | |
| "loss": 0.3276, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.1273360550403595, | |
| "rewards/margins": 0.07243254780769348, | |
| "rewards/rejected": -0.19976858794689178, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": -0.11638545989990234, | |
| "logits/rejected": -0.009681841358542442, | |
| "logps/chosen": -484.964111328125, | |
| "logps/rejected": -536.099609375, | |
| "loss": 0.3322, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.12863071262836456, | |
| "rewards/margins": 0.03380769491195679, | |
| "rewards/rejected": -0.16243840754032135, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": 0.027831071987748146, | |
| "logits/rejected": -0.09354544430971146, | |
| "logps/chosen": -500.5443420410156, | |
| "logps/rejected": -592.7569580078125, | |
| "loss": 0.3133, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.13087543845176697, | |
| "rewards/margins": 0.05950712040066719, | |
| "rewards/rejected": -0.19038255512714386, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": -0.0018000513082370162, | |
| "logits/rejected": -0.0076873586513102055, | |
| "logps/chosen": -555.552734375, | |
| "logps/rejected": -638.7350463867188, | |
| "loss": 0.3334, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.15001370012760162, | |
| "rewards/margins": 0.07444901764392853, | |
| "rewards/rejected": -0.22446270287036896, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": -0.06452230364084244, | |
| "logits/rejected": -0.07058578729629517, | |
| "logps/chosen": -568.3099365234375, | |
| "logps/rejected": -612.906005859375, | |
| "loss": 0.3192, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.16391663253307343, | |
| "rewards/margins": 0.06138613820075989, | |
| "rewards/rejected": -0.22530274093151093, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": -0.08687802404165268, | |
| "logits/rejected": -0.076780766248703, | |
| "logps/chosen": -612.529296875, | |
| "logps/rejected": -664.7831420898438, | |
| "loss": 0.3315, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.188650444149971, | |
| "rewards/margins": 0.07060692459344864, | |
| "rewards/rejected": -0.25925734639167786, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": -0.1167052760720253, | |
| "logits/rejected": -0.08061762899160385, | |
| "logps/chosen": -604.2794189453125, | |
| "logps/rejected": -630.2823486328125, | |
| "loss": 0.329, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.16320253908634186, | |
| "rewards/margins": 0.0545579195022583, | |
| "rewards/rejected": -0.21776047348976135, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": -0.16316112875938416, | |
| "logits/rejected": -0.14865779876708984, | |
| "logps/chosen": -552.4512939453125, | |
| "logps/rejected": -629.3975830078125, | |
| "loss": 0.3366, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.14637981355190277, | |
| "rewards/margins": 0.06526754051446915, | |
| "rewards/rejected": -0.21164734661579132, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": -0.030054206028580666, | |
| "logits/rejected": -0.08666323125362396, | |
| "logps/chosen": -507.67291259765625, | |
| "logps/rejected": -623.1397094726562, | |
| "loss": 0.3152, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.12688292562961578, | |
| "rewards/margins": 0.09281791746616364, | |
| "rewards/rejected": -0.21970084309577942, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": -0.16756266355514526, | |
| "logits/rejected": -0.08279438316822052, | |
| "logps/chosen": -546.107666015625, | |
| "logps/rejected": -583.4124755859375, | |
| "loss": 0.3663, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.14095798134803772, | |
| "rewards/margins": 0.05275345966219902, | |
| "rewards/rejected": -0.19371144473552704, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": -0.07078931480646133, | |
| "logits/rejected": -0.024411968886852264, | |
| "logps/chosen": -521.5887451171875, | |
| "logps/rejected": -594.6179809570312, | |
| "loss": 0.3399, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.12093210220336914, | |
| "rewards/margins": 0.07108847051858902, | |
| "rewards/rejected": -0.19202056527137756, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": -0.11468502134084702, | |
| "logits/rejected": -0.13895608484745026, | |
| "logps/chosen": -504.01617431640625, | |
| "logps/rejected": -566.6771240234375, | |
| "loss": 0.3359, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.09359009563922882, | |
| "rewards/margins": 0.06398390233516693, | |
| "rewards/rejected": -0.15757399797439575, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": -0.15202876925468445, | |
| "logits/rejected": -0.06361649930477142, | |
| "logps/chosen": -508.5723571777344, | |
| "logps/rejected": -572.959716796875, | |
| "loss": 0.3201, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.10353565216064453, | |
| "rewards/margins": 0.06490160524845123, | |
| "rewards/rejected": -0.16843725740909576, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": -0.13953888416290283, | |
| "logits/rejected": -0.08700491487979889, | |
| "logps/chosen": -540.6795654296875, | |
| "logps/rejected": -607.9012451171875, | |
| "loss": 0.3162, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.12709401547908783, | |
| "rewards/margins": 0.06415100395679474, | |
| "rewards/rejected": -0.19124500453472137, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": -0.16462412476539612, | |
| "logits/rejected": -0.05515863746404648, | |
| "logps/chosen": -506.59637451171875, | |
| "logps/rejected": -600.5057373046875, | |
| "loss": 0.314, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.12504397332668304, | |
| "rewards/margins": 0.0829906165599823, | |
| "rewards/rejected": -0.20803460478782654, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": -0.08193415403366089, | |
| "logits/rejected": -0.05177025869488716, | |
| "logps/chosen": -479.7059631347656, | |
| "logps/rejected": -517.7894287109375, | |
| "loss": 0.3372, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.10423684120178223, | |
| "rewards/margins": 0.046939365565776825, | |
| "rewards/rejected": -0.15117622911930084, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": -0.15444053709506989, | |
| "logits/rejected": -0.02473391965031624, | |
| "logps/chosen": -556.2892456054688, | |
| "logps/rejected": -601.0501708984375, | |
| "loss": 0.3136, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.0967629998922348, | |
| "rewards/margins": 0.06504927575588226, | |
| "rewards/rejected": -0.16181227564811707, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": 0.011976124718785286, | |
| "logits/rejected": 0.010825025849044323, | |
| "logps/chosen": -520.3416137695312, | |
| "logps/rejected": -617.6292114257812, | |
| "loss": 0.3083, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.09249953180551529, | |
| "rewards/margins": 0.09974247217178345, | |
| "rewards/rejected": -0.19224199652671814, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": 0.0032506571151316166, | |
| "logits/rejected": -0.05111609026789665, | |
| "logps/chosen": -534.9469604492188, | |
| "logps/rejected": -627.3300170898438, | |
| "loss": 0.3096, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.11961764097213745, | |
| "rewards/margins": 0.09712241590023041, | |
| "rewards/rejected": -0.21674005687236786, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": -0.04654575139284134, | |
| "logits/rejected": 0.02392629161477089, | |
| "logps/chosen": -564.7318115234375, | |
| "logps/rejected": -613.5872802734375, | |
| "loss": 0.326, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.13059517741203308, | |
| "rewards/margins": 0.08819568157196045, | |
| "rewards/rejected": -0.21879085898399353, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": -0.061980921775102615, | |
| "logits/rejected": 0.08209049701690674, | |
| "logps/chosen": -578.1087646484375, | |
| "logps/rejected": -638.3121948242188, | |
| "loss": 0.3223, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.13412974774837494, | |
| "rewards/margins": 0.07588179409503937, | |
| "rewards/rejected": -0.2100115567445755, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": -0.05568457394838333, | |
| "logits/rejected": -0.08035759627819061, | |
| "logps/chosen": -512.9743041992188, | |
| "logps/rejected": -560.0599365234375, | |
| "loss": 0.346, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -0.1260611116886139, | |
| "rewards/margins": 0.04897795990109444, | |
| "rewards/rejected": -0.17503906786441803, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": -0.05415787175297737, | |
| "logits/rejected": 0.043633949011564255, | |
| "logps/chosen": -507.5973205566406, | |
| "logps/rejected": -553.030517578125, | |
| "loss": 0.326, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.10723169893026352, | |
| "rewards/margins": 0.06125740334391594, | |
| "rewards/rejected": -0.16848909854888916, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": -0.07862399518489838, | |
| "logits/rejected": -0.04382779076695442, | |
| "logps/chosen": -506.9168395996094, | |
| "logps/rejected": -584.4060668945312, | |
| "loss": 0.3222, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.09048938751220703, | |
| "rewards/margins": 0.07321049273014069, | |
| "rewards/rejected": -0.16369988024234772, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": -0.046413786709308624, | |
| "logits/rejected": -0.0055254013277590275, | |
| "logps/chosen": -521.91259765625, | |
| "logps/rejected": -581.7081298828125, | |
| "loss": 0.3205, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.10815830528736115, | |
| "rewards/margins": 0.06724079698324203, | |
| "rewards/rejected": -0.1753990799188614, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": -0.06368018686771393, | |
| "logits/rejected": -0.005307002458721399, | |
| "logps/chosen": -505.17950439453125, | |
| "logps/rejected": -562.3865966796875, | |
| "loss": 0.3294, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.11235632002353668, | |
| "rewards/margins": 0.0673181563615799, | |
| "rewards/rejected": -0.17967447638511658, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": -0.12101604044437408, | |
| "logits/rejected": -0.049035437405109406, | |
| "logps/chosen": -559.6871337890625, | |
| "logps/rejected": -634.565673828125, | |
| "loss": 0.3143, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.14436963200569153, | |
| "rewards/margins": 0.07392090559005737, | |
| "rewards/rejected": -0.2182905375957489, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": 0.005751132965087891, | |
| "logits/rejected": 0.00010851323895622045, | |
| "logps/chosen": -529.8284301757812, | |
| "logps/rejected": -600.77783203125, | |
| "loss": 0.332, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.12567661702632904, | |
| "rewards/margins": 0.07157091796398163, | |
| "rewards/rejected": -0.19724753499031067, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": 0.012865634635090828, | |
| "logits/rejected": -0.07809716463088989, | |
| "logps/chosen": -546.5116577148438, | |
| "logps/rejected": -665.4036865234375, | |
| "loss": 0.3226, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.14242392778396606, | |
| "rewards/margins": 0.08929804712533951, | |
| "rewards/rejected": -0.23172196745872498, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": -0.08698035031557083, | |
| "logits/rejected": 0.01734771579504013, | |
| "logps/chosen": -509.57977294921875, | |
| "logps/rejected": -574.0706176757812, | |
| "loss": 0.3039, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.13136714696884155, | |
| "rewards/margins": 0.09488403052091599, | |
| "rewards/rejected": -0.22625115513801575, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": -0.11155049502849579, | |
| "logits/rejected": -0.028806675225496292, | |
| "logps/chosen": -595.2827758789062, | |
| "logps/rejected": -679.2177734375, | |
| "loss": 0.3197, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.16865819692611694, | |
| "rewards/margins": 0.08433203399181366, | |
| "rewards/rejected": -0.2529902160167694, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": -0.09512178599834442, | |
| "logits/rejected": 0.04600674659013748, | |
| "logps/chosen": -571.9730224609375, | |
| "logps/rejected": -646.7930297851562, | |
| "loss": 0.3138, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.17400047183036804, | |
| "rewards/margins": 0.07587233185768127, | |
| "rewards/rejected": -0.24987280368804932, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": -0.04382805526256561, | |
| "logits/rejected": -0.0209367536008358, | |
| "logps/chosen": -529.54931640625, | |
| "logps/rejected": -601.68701171875, | |
| "loss": 0.332, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.15199477970600128, | |
| "rewards/margins": 0.06770177185535431, | |
| "rewards/rejected": -0.2196965217590332, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": -0.06805320829153061, | |
| "logits/rejected": -0.0435967817902565, | |
| "logps/chosen": -574.57568359375, | |
| "logps/rejected": -631.7911987304688, | |
| "loss": 0.3189, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.13725146651268005, | |
| "rewards/margins": 0.07907568663358688, | |
| "rewards/rejected": -0.21632714569568634, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": -0.053729694336652756, | |
| "logits/rejected": -0.009868663735687733, | |
| "logps/chosen": -575.8187866210938, | |
| "logps/rejected": -617.7347412109375, | |
| "loss": 0.3371, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.14825518429279327, | |
| "rewards/margins": 0.07731777429580688, | |
| "rewards/rejected": -0.22557297348976135, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": -0.1181708350777626, | |
| "logits/rejected": -0.051215410232543945, | |
| "logps/chosen": -563.2030639648438, | |
| "logps/rejected": -628.1510009765625, | |
| "loss": 0.3196, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.1552611142396927, | |
| "rewards/margins": 0.07067932933568954, | |
| "rewards/rejected": -0.22594043612480164, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.0351797491312027, | |
| "logits/rejected": -0.09967297315597534, | |
| "logps/chosen": -557.1263427734375, | |
| "logps/rejected": -605.281982421875, | |
| "loss": 0.3204, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.14646664261817932, | |
| "rewards/margins": 0.07007946074008942, | |
| "rewards/rejected": -0.21654610335826874, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": -0.1466439813375473, | |
| "logits/rejected": -0.06507374346256256, | |
| "logps/chosen": -511.141845703125, | |
| "logps/rejected": -615.8265380859375, | |
| "loss": 0.3076, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.15620598196983337, | |
| "rewards/margins": 0.08635568618774414, | |
| "rewards/rejected": -0.24256166815757751, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": -0.01115155965089798, | |
| "logits/rejected": -0.14947594702243805, | |
| "logps/chosen": -603.1889038085938, | |
| "logps/rejected": -690.7286376953125, | |
| "loss": 0.3239, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.17474761605262756, | |
| "rewards/margins": 0.09008537232875824, | |
| "rewards/rejected": -0.2648329734802246, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": -0.05594000220298767, | |
| "logits/rejected": 0.006943766959011555, | |
| "logps/chosen": -565.42724609375, | |
| "logps/rejected": -629.1002197265625, | |
| "loss": 0.3094, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.15515407919883728, | |
| "rewards/margins": 0.07448241859674454, | |
| "rewards/rejected": -0.22963650524616241, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": -0.13735847175121307, | |
| "logits/rejected": 0.04537354037165642, | |
| "logps/chosen": -552.429931640625, | |
| "logps/rejected": -570.6002807617188, | |
| "loss": 0.3396, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": -0.15960142016410828, | |
| "rewards/margins": 0.04359527677297592, | |
| "rewards/rejected": -0.2031966894865036, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": -0.07736693322658539, | |
| "logits/rejected": -0.05264727398753166, | |
| "logps/chosen": -544.4369506835938, | |
| "logps/rejected": -598.355712890625, | |
| "loss": 0.3445, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.15497347712516785, | |
| "rewards/margins": 0.06363853067159653, | |
| "rewards/rejected": -0.21861203014850616, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": -0.07268525660037994, | |
| "logits/rejected": -0.053814847022295, | |
| "logps/chosen": -544.3436279296875, | |
| "logps/rejected": -624.2950439453125, | |
| "loss": 0.3343, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.146144837141037, | |
| "rewards/margins": 0.07093065977096558, | |
| "rewards/rejected": -0.21707551181316376, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": -0.07499315589666367, | |
| "logits/rejected": -0.029876768589019775, | |
| "logps/chosen": -565.154296875, | |
| "logps/rejected": -655.4544067382812, | |
| "loss": 0.2842, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.13710975646972656, | |
| "rewards/margins": 0.07628355920314789, | |
| "rewards/rejected": -0.21339330077171326, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": -0.10578916221857071, | |
| "logits/rejected": 0.04064566642045975, | |
| "logps/chosen": -581.9578247070312, | |
| "logps/rejected": -640.3221435546875, | |
| "loss": 0.3135, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.13465943932533264, | |
| "rewards/margins": 0.0698099285364151, | |
| "rewards/rejected": -0.20446936786174774, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": -0.08297138661146164, | |
| "logits/rejected": -0.07259473949670792, | |
| "logps/chosen": -585.0145263671875, | |
| "logps/rejected": -616.9873657226562, | |
| "loss": 0.3218, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.1459634006023407, | |
| "rewards/margins": 0.07179699838161469, | |
| "rewards/rejected": -0.21776041388511658, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": -0.0852777510881424, | |
| "logits/rejected": -0.1009194403886795, | |
| "logps/chosen": -514.9520263671875, | |
| "logps/rejected": -590.4235229492188, | |
| "loss": 0.3253, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.12546613812446594, | |
| "rewards/margins": 0.07676034420728683, | |
| "rewards/rejected": -0.20222647488117218, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.05136794596910477, | |
| "logits/rejected": -0.047903820872306824, | |
| "logps/chosen": -556.219970703125, | |
| "logps/rejected": -654.3678588867188, | |
| "loss": 0.3159, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.1294935643672943, | |
| "rewards/margins": 0.09803324192762375, | |
| "rewards/rejected": -0.22752681374549866, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": -0.09225879609584808, | |
| "logits/rejected": -0.1522790640592575, | |
| "logps/chosen": -538.4688110351562, | |
| "logps/rejected": -610.7028198242188, | |
| "loss": 0.3291, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.1511029750108719, | |
| "rewards/margins": 0.06751099228858948, | |
| "rewards/rejected": -0.21861395239830017, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": -0.07817684859037399, | |
| "logits/rejected": -0.11500729620456696, | |
| "logps/chosen": -518.6499633789062, | |
| "logps/rejected": -615.0601806640625, | |
| "loss": 0.3069, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.1302158087491989, | |
| "rewards/margins": 0.09146422147750854, | |
| "rewards/rejected": -0.22168001532554626, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": -0.04123290255665779, | |
| "logits/rejected": -0.0999784916639328, | |
| "logps/chosen": -584.451171875, | |
| "logps/rejected": -703.47998046875, | |
| "loss": 0.3058, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.15831688046455383, | |
| "rewards/margins": 0.11426403373479843, | |
| "rewards/rejected": -0.27258092164993286, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": -0.07548685371875763, | |
| "logits/rejected": -0.11166922748088837, | |
| "logps/chosen": -550.4864501953125, | |
| "logps/rejected": -603.9669189453125, | |
| "loss": 0.3336, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.15913598239421844, | |
| "rewards/margins": 0.06622796505689621, | |
| "rewards/rejected": -0.22536394000053406, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": -0.07478149980306625, | |
| "logits/rejected": -0.011451124213635921, | |
| "logps/chosen": -555.3692626953125, | |
| "logps/rejected": -622.7389526367188, | |
| "loss": 0.3177, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.14817936718463898, | |
| "rewards/margins": 0.07494769245386124, | |
| "rewards/rejected": -0.22312703728675842, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.15149171650409698, | |
| "logits/rejected": -0.008707734756171703, | |
| "logps/chosen": -537.5607299804688, | |
| "logps/rejected": -632.0250244140625, | |
| "loss": 0.3153, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.14303314685821533, | |
| "rewards/margins": 0.1020839661359787, | |
| "rewards/rejected": -0.24511709809303284, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": -0.08291809260845184, | |
| "logits/rejected": -0.06536273658275604, | |
| "logps/chosen": -605.5488891601562, | |
| "logps/rejected": -701.1112060546875, | |
| "loss": 0.3347, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.16533637046813965, | |
| "rewards/margins": 0.09125199168920517, | |
| "rewards/rejected": -0.2565883696079254, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": -0.13041993975639343, | |
| "logits/rejected": -0.04777718335390091, | |
| "logps/chosen": -541.6893310546875, | |
| "logps/rejected": -630.17138671875, | |
| "loss": 0.3055, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.14917008578777313, | |
| "rewards/margins": 0.09115210920572281, | |
| "rewards/rejected": -0.24032218754291534, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": -0.09324527531862259, | |
| "logits/rejected": -0.11122927814722061, | |
| "logps/chosen": -541.8341064453125, | |
| "logps/rejected": -629.8985595703125, | |
| "loss": 0.3379, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.16076061129570007, | |
| "rewards/margins": 0.07641086727380753, | |
| "rewards/rejected": -0.2371714562177658, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": -0.1291544884443283, | |
| "logits/rejected": -0.02545195445418358, | |
| "logps/chosen": -582.502197265625, | |
| "logps/rejected": -676.6446533203125, | |
| "loss": 0.3138, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.15107361972332, | |
| "rewards/margins": 0.09025880694389343, | |
| "rewards/rejected": -0.24133244156837463, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": -0.1935300976037979, | |
| "logits/rejected": -0.0874740406870842, | |
| "logps/chosen": -634.2234497070312, | |
| "logps/rejected": -651.4421997070312, | |
| "loss": 0.3293, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.17711934447288513, | |
| "rewards/margins": 0.060172189027071, | |
| "rewards/rejected": -0.23729152977466583, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": -0.09884585440158844, | |
| "logits/rejected": -0.052547551691532135, | |
| "logps/chosen": -597.0643310546875, | |
| "logps/rejected": -668.3179931640625, | |
| "loss": 0.3121, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.15433833003044128, | |
| "rewards/margins": 0.07509264349937439, | |
| "rewards/rejected": -0.22943095862865448, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": -0.17812196910381317, | |
| "logits/rejected": -0.04409525915980339, | |
| "logps/chosen": -529.5570068359375, | |
| "logps/rejected": -592.1903076171875, | |
| "loss": 0.3069, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.12470044940710068, | |
| "rewards/margins": 0.08706951141357422, | |
| "rewards/rejected": -0.2117699682712555, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": -0.17632423341274261, | |
| "logits/rejected": 0.020423654466867447, | |
| "logps/chosen": -539.355224609375, | |
| "logps/rejected": -550.2349853515625, | |
| "loss": 0.3477, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.14735418558120728, | |
| "rewards/margins": 0.04963860288262367, | |
| "rewards/rejected": -0.19699278473854065, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": -0.11399457603693008, | |
| "logits/rejected": -0.008361217565834522, | |
| "logps/chosen": -550.0650634765625, | |
| "logps/rejected": -629.9093017578125, | |
| "loss": 0.342, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.14632944762706757, | |
| "rewards/margins": 0.09393130242824554, | |
| "rewards/rejected": -0.2402607500553131, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": -0.1393519937992096, | |
| "logits/rejected": -0.07454421371221542, | |
| "logps/chosen": -484.50213623046875, | |
| "logps/rejected": -572.2808837890625, | |
| "loss": 0.3571, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.12875160574913025, | |
| "rewards/margins": 0.08377497643232346, | |
| "rewards/rejected": -0.21252655982971191, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": -0.0922452062368393, | |
| "logits/rejected": -0.11933571100234985, | |
| "logps/chosen": -504.70806884765625, | |
| "logps/rejected": -641.5916748046875, | |
| "loss": 0.3064, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.13076017796993256, | |
| "rewards/margins": 0.10374923050403595, | |
| "rewards/rejected": -0.2345094233751297, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": -0.08234205842018127, | |
| "logits/rejected": 0.009737101383507252, | |
| "logps/chosen": -560.4942016601562, | |
| "logps/rejected": -647.23095703125, | |
| "loss": 0.3171, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.1362009346485138, | |
| "rewards/margins": 0.08297161757946014, | |
| "rewards/rejected": -0.21917252242565155, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": -0.010607835836708546, | |
| "logits/rejected": -0.1094130128622055, | |
| "logps/chosen": -571.977783203125, | |
| "logps/rejected": -655.345703125, | |
| "loss": 0.3039, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.1528376042842865, | |
| "rewards/margins": 0.06883414089679718, | |
| "rewards/rejected": -0.22167173027992249, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": -0.10850746929645538, | |
| "logits/rejected": -0.10889410972595215, | |
| "logps/chosen": -419.2569885253906, | |
| "logps/rejected": -512.2922973632812, | |
| "loss": 0.3312, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.11631663888692856, | |
| "rewards/margins": 0.07554365694522858, | |
| "rewards/rejected": -0.19186028838157654, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": -0.16420094668865204, | |
| "logits/rejected": -0.08157550543546677, | |
| "logps/chosen": -544.343994140625, | |
| "logps/rejected": -583.8645629882812, | |
| "loss": 0.318, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.1383691132068634, | |
| "rewards/margins": 0.07660222053527832, | |
| "rewards/rejected": -0.21497133374214172, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": -0.13259460031986237, | |
| "logits/rejected": -0.15146221220493317, | |
| "logps/chosen": -514.0303955078125, | |
| "logps/rejected": -571.2809448242188, | |
| "loss": 0.3349, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.13906028866767883, | |
| "rewards/margins": 0.06452740728855133, | |
| "rewards/rejected": -0.20358769595623016, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": -0.0921165943145752, | |
| "logits/rejected": -0.084059938788414, | |
| "logps/chosen": -513.1318969726562, | |
| "logps/rejected": -630.0098876953125, | |
| "loss": 0.309, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.13389384746551514, | |
| "rewards/margins": 0.10617604106664658, | |
| "rewards/rejected": -0.24006986618041992, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": -0.1827644407749176, | |
| "logits/rejected": -0.017491687089204788, | |
| "logps/chosen": -513.20703125, | |
| "logps/rejected": -602.4353637695312, | |
| "loss": 0.3324, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.12371834367513657, | |
| "rewards/margins": 0.09411574900150299, | |
| "rewards/rejected": -0.21783408522605896, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": -0.10702376067638397, | |
| "logits/rejected": -0.09792954474687576, | |
| "logps/chosen": -554.7036743164062, | |
| "logps/rejected": -612.7283935546875, | |
| "loss": 0.3209, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.1507497876882553, | |
| "rewards/margins": 0.05762406066060066, | |
| "rewards/rejected": -0.20837385952472687, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": -0.14019733667373657, | |
| "logits/rejected": -0.11023645102977753, | |
| "logps/chosen": -621.0760498046875, | |
| "logps/rejected": -737.30078125, | |
| "loss": 0.3071, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": -0.15968570113182068, | |
| "rewards/margins": 0.11614030599594116, | |
| "rewards/rejected": -0.27582600712776184, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": -0.09468534588813782, | |
| "logits/rejected": -0.021795030683279037, | |
| "logps/chosen": -583.4420166015625, | |
| "logps/rejected": -647.2615356445312, | |
| "loss": 0.3225, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.14064806699752808, | |
| "rewards/margins": 0.08540509641170502, | |
| "rewards/rejected": -0.22605319321155548, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": -0.09509189426898956, | |
| "logits/rejected": -0.08425020426511765, | |
| "logps/chosen": -551.0325317382812, | |
| "logps/rejected": -589.43017578125, | |
| "loss": 0.3333, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.1517583280801773, | |
| "rewards/margins": 0.05821775645017624, | |
| "rewards/rejected": -0.20997610688209534, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.3301706374836579, | |
| "train_runtime": 6917.8099, | |
| "train_samples_per_second": 4.337, | |
| "train_steps_per_second": 0.135 | |
| } | |
| ], | |
| "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 | |
| } | |