| { | |
| "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": 0.1853, | |
| "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": 0.15297521650791168, | |
| "logits/rejected": 0.29175662994384766, | |
| "logps/chosen": -365.80181884765625, | |
| "logps/rejected": -353.0853271484375, | |
| "loss": 0.2099, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.0007080123177729547, | |
| "rewards/margins": -5.8396861277287826e-05, | |
| "rewards/rejected": -0.0006496154237538576, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.11968117952346802, | |
| "logits/rejected": 0.2041483372449875, | |
| "logps/chosen": -340.0993347167969, | |
| "logps/rejected": -348.33087158203125, | |
| "loss": 0.2094, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": -0.000655159296002239, | |
| "rewards/margins": -8.313418220495805e-05, | |
| "rewards/rejected": -0.0005720251356251538, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.2551038861274719, | |
| "logits/rejected": 0.25183868408203125, | |
| "logps/chosen": -383.1521301269531, | |
| "logps/rejected": -364.0672302246094, | |
| "loss": 0.2012, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": -0.0005650260718539357, | |
| "rewards/margins": 0.00010353984544053674, | |
| "rewards/rejected": -0.0006685658590868115, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.1547292321920395, | |
| "logits/rejected": 0.27106207609176636, | |
| "logps/chosen": -401.61614990234375, | |
| "logps/rejected": -385.8863220214844, | |
| "loss": 0.2099, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.0006045111804269254, | |
| "rewards/margins": -5.9384223277447745e-05, | |
| "rewards/rejected": -0.0005451269680634141, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.23326897621154785, | |
| "logits/rejected": 0.27433687448501587, | |
| "logps/chosen": -441.8401794433594, | |
| "logps/rejected": -432.41485595703125, | |
| "loss": 0.2047, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.000930719543248415, | |
| "rewards/margins": 0.000368706532754004, | |
| "rewards/rejected": -0.0012994259595870972, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.17064206302165985, | |
| "logits/rejected": 0.3185887336730957, | |
| "logps/chosen": -410.41473388671875, | |
| "logps/rejected": -414.3666076660156, | |
| "loss": 0.2182, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": -0.001497046323493123, | |
| "rewards/margins": 0.00016530933498870581, | |
| "rewards/rejected": -0.0016623556148260832, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.12393184751272202, | |
| "logits/rejected": 0.2235107123851776, | |
| "logps/chosen": -354.70562744140625, | |
| "logps/rejected": -356.94586181640625, | |
| "loss": 0.2086, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -5.924403740209527e-05, | |
| "rewards/margins": 0.0009310436435043812, | |
| "rewards/rejected": -0.000990287633612752, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.14353762567043304, | |
| "logits/rejected": 0.2516772449016571, | |
| "logps/chosen": -392.6264343261719, | |
| "logps/rejected": -380.66351318359375, | |
| "loss": 0.208, | |
| "rewards/accuracies": 0.40625, | |
| "rewards/chosen": 0.0002285484952153638, | |
| "rewards/margins": 0.001034508110024035, | |
| "rewards/rejected": -0.0008059596875682473, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.24103212356567383, | |
| "logits/rejected": 0.1776101142168045, | |
| "logps/chosen": -393.3184509277344, | |
| "logps/rejected": -416.2762145996094, | |
| "loss": 0.1992, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.0003849788336083293, | |
| "rewards/margins": 0.0017982361605390906, | |
| "rewards/rejected": -0.0021832147613167763, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.16443544626235962, | |
| "logits/rejected": 0.17112873494625092, | |
| "logps/chosen": -416.6537170410156, | |
| "logps/rejected": -411.6963806152344, | |
| "loss": 0.2064, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": 0.0005408526631072164, | |
| "rewards/margins": 0.0026028361171483994, | |
| "rewards/rejected": -0.002061983570456505, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.2384149730205536, | |
| "logits/rejected": 0.1614537537097931, | |
| "logps/chosen": -406.7789306640625, | |
| "logps/rejected": -391.0703430175781, | |
| "loss": 0.2008, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.012324010953307152, | |
| "rewards/margins": 0.0033186424989253283, | |
| "rewards/rejected": 0.009005369618535042, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.08936997503042221, | |
| "logits/rejected": 0.25732293725013733, | |
| "logps/chosen": -397.0160827636719, | |
| "logps/rejected": -431.9867248535156, | |
| "loss": 0.2064, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.01773521490395069, | |
| "rewards/margins": 0.007638473063707352, | |
| "rewards/rejected": 0.01009674184024334, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.14923642575740814, | |
| "logits/rejected": 0.27579236030578613, | |
| "logps/chosen": -407.21258544921875, | |
| "logps/rejected": -401.8697204589844, | |
| "loss": 0.2048, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.005354008637368679, | |
| "rewards/margins": 0.00832393579185009, | |
| "rewards/rejected": -0.0029699269216507673, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.13965365290641785, | |
| "logits/rejected": 0.20428553223609924, | |
| "logps/chosen": -396.0818786621094, | |
| "logps/rejected": -384.4440612792969, | |
| "loss": 0.2016, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": -0.011929613538086414, | |
| "rewards/margins": 0.00782632827758789, | |
| "rewards/rejected": -0.01975594088435173, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.0017524458235129714, | |
| "logits/rejected": 0.048104483634233475, | |
| "logps/chosen": -403.7103576660156, | |
| "logps/rejected": -421.3060607910156, | |
| "loss": 0.1918, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.020950669422745705, | |
| "rewards/margins": 0.01481578964740038, | |
| "rewards/rejected": -0.03576646000146866, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.050279758870601654, | |
| "logits/rejected": 0.12556883692741394, | |
| "logps/chosen": -467.9580993652344, | |
| "logps/rejected": -487.71844482421875, | |
| "loss": 0.1868, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.044767118990421295, | |
| "rewards/margins": 0.03162100166082382, | |
| "rewards/rejected": -0.07638812065124512, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": -0.020991306751966476, | |
| "logits/rejected": 0.15817420184612274, | |
| "logps/chosen": -522.8843383789062, | |
| "logps/rejected": -518.1360473632812, | |
| "loss": 0.2105, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.08218502998352051, | |
| "rewards/margins": 0.02223752811551094, | |
| "rewards/rejected": -0.10442256927490234, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.08929436653852463, | |
| "logits/rejected": 0.09290768206119537, | |
| "logps/chosen": -470.04296875, | |
| "logps/rejected": -501.46661376953125, | |
| "loss": 0.1886, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.08902844041585922, | |
| "rewards/margins": 0.03190689533948898, | |
| "rewards/rejected": -0.12093535810709, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.017135417088866234, | |
| "logits/rejected": 0.09486501663923264, | |
| "logps/chosen": -600.1754760742188, | |
| "logps/rejected": -609.9652709960938, | |
| "loss": 0.1887, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.11600615829229355, | |
| "rewards/margins": 0.041202057152986526, | |
| "rewards/rejected": -0.15720821917057037, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": -0.032647065818309784, | |
| "logits/rejected": 0.09240031987428665, | |
| "logps/chosen": -527.2655029296875, | |
| "logps/rejected": -595.9906005859375, | |
| "loss": 0.1949, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.1231132298707962, | |
| "rewards/margins": 0.051538724452257156, | |
| "rewards/rejected": -0.17465195059776306, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.03914088383316994, | |
| "logits/rejected": 0.028707262128591537, | |
| "logps/chosen": -524.7379760742188, | |
| "logps/rejected": -570.7955322265625, | |
| "loss": 0.1902, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.12313251197338104, | |
| "rewards/margins": 0.04237721115350723, | |
| "rewards/rejected": -0.16550973057746887, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.010909264907240868, | |
| "logits/rejected": -0.024190250784158707, | |
| "logps/chosen": -549.97607421875, | |
| "logps/rejected": -590.9778442382812, | |
| "loss": 0.1843, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.14806947112083435, | |
| "rewards/margins": 0.04638643562793732, | |
| "rewards/rejected": -0.19445592164993286, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": -0.018162641674280167, | |
| "logits/rejected": 0.000972352921962738, | |
| "logps/chosen": -549.0392456054688, | |
| "logps/rejected": -598.3811645507812, | |
| "loss": 0.2025, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.1491001546382904, | |
| "rewards/margins": 0.04847537726163864, | |
| "rewards/rejected": -0.19757553935050964, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": -0.12062356621026993, | |
| "logits/rejected": 0.04868536815047264, | |
| "logps/chosen": -484.58685302734375, | |
| "logps/rejected": -516.2865600585938, | |
| "loss": 0.1908, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.1070113405585289, | |
| "rewards/margins": 0.028431424871087074, | |
| "rewards/rejected": -0.13544276356697083, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": -0.041638366878032684, | |
| "logits/rejected": 0.0221172496676445, | |
| "logps/chosen": -507.4730529785156, | |
| "logps/rejected": -527.9345703125, | |
| "loss": 0.2052, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.09331385791301727, | |
| "rewards/margins": 0.03461749479174614, | |
| "rewards/rejected": -0.1279313564300537, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": -0.05273251608014107, | |
| "logits/rejected": -0.022044766694307327, | |
| "logps/chosen": -538.548583984375, | |
| "logps/rejected": -590.6500854492188, | |
| "loss": 0.182, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.10730306804180145, | |
| "rewards/margins": 0.039316385984420776, | |
| "rewards/rejected": -0.14661946892738342, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": -0.12279339134693146, | |
| "logits/rejected": -0.079288050532341, | |
| "logps/chosen": -560.711181640625, | |
| "logps/rejected": -635.7985229492188, | |
| "loss": 0.1845, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.12682856619358063, | |
| "rewards/margins": 0.0557611808180809, | |
| "rewards/rejected": -0.18258973956108093, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": -0.12978403270244598, | |
| "logits/rejected": -0.05718718096613884, | |
| "logps/chosen": -534.7046508789062, | |
| "logps/rejected": -573.9546508789062, | |
| "loss": 0.1872, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.12655052542686462, | |
| "rewards/margins": 0.05336705967783928, | |
| "rewards/rejected": -0.1799176186323166, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": -0.13656684756278992, | |
| "logits/rejected": -0.07923261821269989, | |
| "logps/chosen": -537.6559448242188, | |
| "logps/rejected": -627.1619873046875, | |
| "loss": 0.1952, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.13470812141895294, | |
| "rewards/margins": 0.05948293209075928, | |
| "rewards/rejected": -0.1941910684108734, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": -0.11429516226053238, | |
| "logits/rejected": -0.12869636714458466, | |
| "logps/chosen": -543.2788696289062, | |
| "logps/rejected": -558.6578369140625, | |
| "loss": 0.1847, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.12002478539943695, | |
| "rewards/margins": 0.050586897879838943, | |
| "rewards/rejected": -0.1706116795539856, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": -0.18651030957698822, | |
| "logits/rejected": -0.052459727972745895, | |
| "logps/chosen": -461.1847229003906, | |
| "logps/rejected": -506.2823181152344, | |
| "loss": 0.1928, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.08158674091100693, | |
| "rewards/margins": 0.045484792441129684, | |
| "rewards/rejected": -0.1270715296268463, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": -0.1330818384885788, | |
| "logits/rejected": -0.09265539795160294, | |
| "logps/chosen": -483.45599365234375, | |
| "logps/rejected": -515.7194213867188, | |
| "loss": 0.1797, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.10115663707256317, | |
| "rewards/margins": 0.04430658370256424, | |
| "rewards/rejected": -0.14546321332454681, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": -0.11196194589138031, | |
| "logits/rejected": -0.09686783701181412, | |
| "logps/chosen": -557.8707885742188, | |
| "logps/rejected": -571.39794921875, | |
| "loss": 0.1958, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.12075225263834, | |
| "rewards/margins": 0.04514995589852333, | |
| "rewards/rejected": -0.16590221226215363, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": -0.08174435794353485, | |
| "logits/rejected": -0.06923134624958038, | |
| "logps/chosen": -531.8570556640625, | |
| "logps/rejected": -605.0074462890625, | |
| "loss": 0.1889, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.11230266094207764, | |
| "rewards/margins": 0.06913084536790848, | |
| "rewards/rejected": -0.18143349885940552, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": -0.16061343252658844, | |
| "logits/rejected": -0.027816006913781166, | |
| "logps/chosen": -534.1594848632812, | |
| "logps/rejected": -573.8477783203125, | |
| "loss": 0.1926, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.0970136970281601, | |
| "rewards/margins": 0.045039448887109756, | |
| "rewards/rejected": -0.14205312728881836, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": -0.24527081847190857, | |
| "logits/rejected": -0.1484527587890625, | |
| "logps/chosen": -544.539306640625, | |
| "logps/rejected": -564.9341430664062, | |
| "loss": 0.1859, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.09417351335287094, | |
| "rewards/margins": 0.056557249277830124, | |
| "rewards/rejected": -0.15073075890541077, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": -0.1722763478755951, | |
| "logits/rejected": -0.17130622267723083, | |
| "logps/chosen": -496.27313232421875, | |
| "logps/rejected": -557.1398315429688, | |
| "loss": 0.1877, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.0965786799788475, | |
| "rewards/margins": 0.057069409638643265, | |
| "rewards/rejected": -0.15364809334278107, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": -0.286950945854187, | |
| "logits/rejected": -0.12660877406597137, | |
| "logps/chosen": -487.8304138183594, | |
| "logps/rejected": -506.80267333984375, | |
| "loss": 0.1784, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.08378596603870392, | |
| "rewards/margins": 0.04325443506240845, | |
| "rewards/rejected": -0.12704041600227356, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": -0.2721463441848755, | |
| "logits/rejected": -0.19791728258132935, | |
| "logps/chosen": -525.049560546875, | |
| "logps/rejected": -564.6629028320312, | |
| "loss": 0.1924, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.1035178154706955, | |
| "rewards/margins": 0.03107512556016445, | |
| "rewards/rejected": -0.1345929503440857, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": -0.17590856552124023, | |
| "logits/rejected": -0.16738948225975037, | |
| "logps/chosen": -520.5001831054688, | |
| "logps/rejected": -564.5961303710938, | |
| "loss": 0.1854, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.09734812378883362, | |
| "rewards/margins": 0.0474289208650589, | |
| "rewards/rejected": -0.14477702975273132, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": -0.17745746672153473, | |
| "logits/rejected": -0.18353696167469025, | |
| "logps/chosen": -508.34637451171875, | |
| "logps/rejected": -577.1897583007812, | |
| "loss": 0.1804, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.0865376815199852, | |
| "rewards/margins": 0.049374908208847046, | |
| "rewards/rejected": -0.13591258227825165, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": -0.17098669707775116, | |
| "logits/rejected": -0.13703958690166473, | |
| "logps/chosen": -539.40087890625, | |
| "logps/rejected": -593.8014526367188, | |
| "loss": 0.1851, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.09274087101221085, | |
| "rewards/margins": 0.05765017122030258, | |
| "rewards/rejected": -0.15039105713367462, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": -0.20213007926940918, | |
| "logits/rejected": -0.20745894312858582, | |
| "logps/chosen": -453.65478515625, | |
| "logps/rejected": -497.0008850097656, | |
| "loss": 0.1824, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.08567062765359879, | |
| "rewards/margins": 0.04529280215501785, | |
| "rewards/rejected": -0.13096341490745544, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": -0.21449732780456543, | |
| "logits/rejected": -0.19523288309574127, | |
| "logps/chosen": -547.9935302734375, | |
| "logps/rejected": -572.2437744140625, | |
| "loss": 0.1901, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.09648506343364716, | |
| "rewards/margins": 0.03453432396054268, | |
| "rewards/rejected": -0.13101938366889954, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": -0.22678379714488983, | |
| "logits/rejected": -0.18668214976787567, | |
| "logps/chosen": -519.3316650390625, | |
| "logps/rejected": -561.0910034179688, | |
| "loss": 0.1896, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.09367823600769043, | |
| "rewards/margins": 0.058413583785295486, | |
| "rewards/rejected": -0.15209180116653442, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": -0.20922398567199707, | |
| "logits/rejected": -0.15190599858760834, | |
| "logps/chosen": -554.7764892578125, | |
| "logps/rejected": -584.9015502929688, | |
| "loss": 0.1777, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.11632993072271347, | |
| "rewards/margins": 0.048789944499731064, | |
| "rewards/rejected": -0.16511985659599304, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": -0.2042142152786255, | |
| "logits/rejected": -0.19644713401794434, | |
| "logps/chosen": -522.6699829101562, | |
| "logps/rejected": -589.4488525390625, | |
| "loss": 0.189, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.10839296877384186, | |
| "rewards/margins": 0.06792866438627243, | |
| "rewards/rejected": -0.17632164061069489, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": -0.29581087827682495, | |
| "logits/rejected": -0.17288121581077576, | |
| "logps/chosen": -492.73297119140625, | |
| "logps/rejected": -565.6483764648438, | |
| "loss": 0.1819, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.10467328131198883, | |
| "rewards/margins": 0.0681912824511528, | |
| "rewards/rejected": -0.17286454141139984, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": -0.1768864393234253, | |
| "logits/rejected": -0.19145308434963226, | |
| "logps/chosen": -481.3701171875, | |
| "logps/rejected": -552.0697021484375, | |
| "loss": 0.1861, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.10367073863744736, | |
| "rewards/margins": 0.0636112317442894, | |
| "rewards/rejected": -0.16728197038173676, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": -0.25040799379348755, | |
| "logits/rejected": -0.2705633044242859, | |
| "logps/chosen": -473.15777587890625, | |
| "logps/rejected": -521.9263916015625, | |
| "loss": 0.1845, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.10329292714595795, | |
| "rewards/margins": 0.04386230558156967, | |
| "rewards/rejected": -0.1471552550792694, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": -0.22839005291461945, | |
| "logits/rejected": -0.18522998690605164, | |
| "logps/chosen": -562.2581176757812, | |
| "logps/rejected": -602.7523193359375, | |
| "loss": 0.1852, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.11718226969242096, | |
| "rewards/margins": 0.054385870695114136, | |
| "rewards/rejected": -0.1715681403875351, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": -0.28231528401374817, | |
| "logits/rejected": -0.1746218502521515, | |
| "logps/chosen": -508.0462951660156, | |
| "logps/rejected": -521.466552734375, | |
| "loss": 0.1887, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.0966549962759018, | |
| "rewards/margins": 0.04605900123715401, | |
| "rewards/rejected": -0.1427139937877655, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": -0.14483687281608582, | |
| "logits/rejected": -0.18159925937652588, | |
| "logps/chosen": -519.6622314453125, | |
| "logps/rejected": -554.4771728515625, | |
| "loss": 0.1881, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.10754968971014023, | |
| "rewards/margins": 0.043608419597148895, | |
| "rewards/rejected": -0.15115809440612793, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.27724790573120117, | |
| "logits/rejected": -0.18303519487380981, | |
| "logps/chosen": -502.09747314453125, | |
| "logps/rejected": -529.7732543945312, | |
| "loss": 0.1902, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.0977000966668129, | |
| "rewards/margins": 0.04301925003528595, | |
| "rewards/rejected": -0.14071933925151825, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": -0.2885403633117676, | |
| "logits/rejected": -0.16289584338665009, | |
| "logps/chosen": -519.3963012695312, | |
| "logps/rejected": -553.4032592773438, | |
| "loss": 0.1871, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.10111876577138901, | |
| "rewards/margins": 0.040404774248600006, | |
| "rewards/rejected": -0.1415235549211502, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": -0.23527821898460388, | |
| "logits/rejected": -0.2247372567653656, | |
| "logps/chosen": -481.41552734375, | |
| "logps/rejected": -552.4132080078125, | |
| "loss": 0.2026, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.10331599414348602, | |
| "rewards/margins": 0.040959432721138, | |
| "rewards/rejected": -0.14427544176578522, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": -0.28348255157470703, | |
| "logits/rejected": -0.26194503903388977, | |
| "logps/chosen": -467.77740478515625, | |
| "logps/rejected": -507.6676330566406, | |
| "loss": 0.1887, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.10089198499917984, | |
| "rewards/margins": 0.0405060276389122, | |
| "rewards/rejected": -0.14139802753925323, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": -0.22403912246227264, | |
| "logits/rejected": -0.19076624512672424, | |
| "logps/chosen": -500.50982666015625, | |
| "logps/rejected": -544.9527587890625, | |
| "loss": 0.1793, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.09000807255506516, | |
| "rewards/margins": 0.05757290869951248, | |
| "rewards/rejected": -0.14758098125457764, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": -0.25425633788108826, | |
| "logits/rejected": -0.13311608135700226, | |
| "logps/chosen": -523.557373046875, | |
| "logps/rejected": -576.8714599609375, | |
| "loss": 0.1667, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.09282426536083221, | |
| "rewards/margins": 0.06447537243366241, | |
| "rewards/rejected": -0.15729963779449463, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": -0.25314217805862427, | |
| "logits/rejected": -0.2236749678850174, | |
| "logps/chosen": -442.9320373535156, | |
| "logps/rejected": -521.5167236328125, | |
| "loss": 0.1816, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.08528304100036621, | |
| "rewards/margins": 0.05453087016940117, | |
| "rewards/rejected": -0.13981391489505768, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": -0.2589682936668396, | |
| "logits/rejected": -0.2686694264411926, | |
| "logps/chosen": -501.1578674316406, | |
| "logps/rejected": -545.9219970703125, | |
| "loss": 0.1949, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -0.09304684400558472, | |
| "rewards/margins": 0.06486930698156357, | |
| "rewards/rejected": -0.1579161435365677, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": -0.15998974442481995, | |
| "logits/rejected": -0.26897841691970825, | |
| "logps/chosen": -483.9093322753906, | |
| "logps/rejected": -574.64990234375, | |
| "loss": 0.1784, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.0897846445441246, | |
| "rewards/margins": 0.059298910200595856, | |
| "rewards/rejected": -0.14908355474472046, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": -0.30697402358055115, | |
| "logits/rejected": -0.2922336459159851, | |
| "logps/chosen": -446.08642578125, | |
| "logps/rejected": -485.38311767578125, | |
| "loss": 0.1835, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.07366035133600235, | |
| "rewards/margins": 0.05683215707540512, | |
| "rewards/rejected": -0.13049249351024628, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": -0.27867692708969116, | |
| "logits/rejected": -0.23723456263542175, | |
| "logps/chosen": -472.146240234375, | |
| "logps/rejected": -522.3912963867188, | |
| "loss": 0.1886, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": -0.08290112018585205, | |
| "rewards/margins": 0.06648631393909454, | |
| "rewards/rejected": -0.1493874490261078, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.27338069677352905, | |
| "logits/rejected": -0.0849432423710823, | |
| "logps/chosen": -505.39404296875, | |
| "logps/rejected": -564.4884643554688, | |
| "loss": 0.1769, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.0819827988743782, | |
| "rewards/margins": 0.07497727125883102, | |
| "rewards/rejected": -0.15696007013320923, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": -0.31306496262550354, | |
| "logits/rejected": -0.15988986194133759, | |
| "logps/chosen": -536.2052001953125, | |
| "logps/rejected": -565.6595458984375, | |
| "loss": 0.1917, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.10776883363723755, | |
| "rewards/margins": 0.04764767736196518, | |
| "rewards/rejected": -0.15541651844978333, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": -0.1312873661518097, | |
| "logits/rejected": -0.1946374773979187, | |
| "logps/chosen": -489.32037353515625, | |
| "logps/rejected": -532.713623046875, | |
| "loss": 0.1923, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.10007087886333466, | |
| "rewards/margins": 0.0466584786772728, | |
| "rewards/rejected": -0.14672937989234924, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": -0.18406830728054047, | |
| "logits/rejected": -0.18264801800251007, | |
| "logps/chosen": -455.0455627441406, | |
| "logps/rejected": -483.98748779296875, | |
| "loss": 0.187, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": -0.08491896092891693, | |
| "rewards/margins": 0.042202599346637726, | |
| "rewards/rejected": -0.12712153792381287, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": -0.2591504454612732, | |
| "logits/rejected": -0.1554795801639557, | |
| "logps/chosen": -488.55328369140625, | |
| "logps/rejected": -532.9073486328125, | |
| "loss": 0.1807, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.09263734519481659, | |
| "rewards/margins": 0.04989578202366829, | |
| "rewards/rejected": -0.1425331085920334, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": -0.28501999378204346, | |
| "logits/rejected": -0.23185932636260986, | |
| "logps/chosen": -485.5430603027344, | |
| "logps/rejected": -541.1561889648438, | |
| "loss": 0.1604, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.09013941138982773, | |
| "rewards/margins": 0.051737189292907715, | |
| "rewards/rejected": -0.14187659323215485, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.25448185205459595, | |
| "logits/rejected": -0.13845598697662354, | |
| "logps/chosen": -504.44085693359375, | |
| "logps/rejected": -557.7171630859375, | |
| "loss": 0.1875, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.09454745799303055, | |
| "rewards/margins": 0.0519348680973053, | |
| "rewards/rejected": -0.14648231863975525, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": -0.20559599995613098, | |
| "logits/rejected": -0.1910923421382904, | |
| "logps/chosen": -518.5410766601562, | |
| "logps/rejected": -559.4114379882812, | |
| "loss": 0.1781, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.09824297577142715, | |
| "rewards/margins": 0.051730893552303314, | |
| "rewards/rejected": -0.14997386932373047, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": -0.20453593134880066, | |
| "logits/rejected": -0.22350621223449707, | |
| "logps/chosen": -517.8855590820312, | |
| "logps/rejected": -554.9312744140625, | |
| "loss": 0.173, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.0931503102183342, | |
| "rewards/margins": 0.059033893048763275, | |
| "rewards/rejected": -0.15218421816825867, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": -0.24915683269500732, | |
| "logits/rejected": -0.14556431770324707, | |
| "logps/chosen": -458.88006591796875, | |
| "logps/rejected": -483.50872802734375, | |
| "loss": 0.1791, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.0776047632098198, | |
| "rewards/margins": 0.055497486144304276, | |
| "rewards/rejected": -0.13310226798057556, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": -0.22538790106773376, | |
| "logits/rejected": -0.16318151354789734, | |
| "logps/chosen": -516.1492309570312, | |
| "logps/rejected": -586.5814208984375, | |
| "loss": 0.1829, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.09301020205020905, | |
| "rewards/margins": 0.06428654491901398, | |
| "rewards/rejected": -0.15729674696922302, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": -0.2806158661842346, | |
| "logits/rejected": -0.066395103931427, | |
| "logps/chosen": -479.8614807128906, | |
| "logps/rejected": -503.1717224121094, | |
| "loss": 0.1733, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.09291915595531464, | |
| "rewards/margins": 0.04909076914191246, | |
| "rewards/rejected": -0.1420099288225174, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": -0.325612872838974, | |
| "logits/rejected": -0.22937624156475067, | |
| "logps/chosen": -526.00341796875, | |
| "logps/rejected": -547.8123168945312, | |
| "loss": 0.1746, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.09224705398082733, | |
| "rewards/margins": 0.054325349628925323, | |
| "rewards/rejected": -0.14657239615917206, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": -0.2622816264629364, | |
| "logits/rejected": -0.15280409157276154, | |
| "logps/chosen": -533.4638061523438, | |
| "logps/rejected": -595.7601318359375, | |
| "loss": 0.1815, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.09688106924295425, | |
| "rewards/margins": 0.05785801261663437, | |
| "rewards/rejected": -0.15473909676074982, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": -0.18345573544502258, | |
| "logits/rejected": -0.16111025214195251, | |
| "logps/chosen": -487.77886962890625, | |
| "logps/rejected": -538.51171875, | |
| "loss": 0.184, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.08573255687952042, | |
| "rewards/margins": 0.0516083724796772, | |
| "rewards/rejected": -0.13734093308448792, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": -0.2816532254219055, | |
| "logits/rejected": -0.15862765908241272, | |
| "logps/chosen": -512.4444580078125, | |
| "logps/rejected": -536.9342651367188, | |
| "loss": 0.1844, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.09420600533485413, | |
| "rewards/margins": 0.05614888668060303, | |
| "rewards/rejected": -0.15035490691661835, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": -0.21904349327087402, | |
| "logits/rejected": -0.258176326751709, | |
| "logps/chosen": -549.0933837890625, | |
| "logps/rejected": -618.8422241210938, | |
| "loss": 0.18, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.10313485562801361, | |
| "rewards/margins": 0.07112576067447662, | |
| "rewards/rejected": -0.17426061630249023, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": -0.3128640353679657, | |
| "logits/rejected": -0.10742131620645523, | |
| "logps/chosen": -525.4251098632812, | |
| "logps/rejected": -553.0595092773438, | |
| "loss": 0.1734, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.08687300980091095, | |
| "rewards/margins": 0.06989692151546478, | |
| "rewards/rejected": -0.15676993131637573, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": -0.1870919167995453, | |
| "logits/rejected": -0.18910066783428192, | |
| "logps/chosen": -514.9647827148438, | |
| "logps/rejected": -548.2384643554688, | |
| "loss": 0.182, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.0961606353521347, | |
| "rewards/margins": 0.05352962762117386, | |
| "rewards/rejected": -0.14969027042388916, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": -0.31641727685928345, | |
| "logits/rejected": -0.21929411590099335, | |
| "logps/chosen": -452.8780212402344, | |
| "logps/rejected": -497.7808532714844, | |
| "loss": 0.1887, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.08406294882297516, | |
| "rewards/margins": 0.06199796125292778, | |
| "rewards/rejected": -0.14606089890003204, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": -0.2824193239212036, | |
| "logits/rejected": -0.2219020426273346, | |
| "logps/chosen": -516.4136962890625, | |
| "logps/rejected": -553.1406860351562, | |
| "loss": 0.1761, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.09720293432474136, | |
| "rewards/margins": 0.046975888311862946, | |
| "rewards/rejected": -0.1441788375377655, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": -0.09483526647090912, | |
| "logits/rejected": -0.24578902125358582, | |
| "logps/chosen": -496.60028076171875, | |
| "logps/rejected": -558.0816650390625, | |
| "loss": 0.175, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.0964212641119957, | |
| "rewards/margins": 0.050969939678907394, | |
| "rewards/rejected": -0.14739122986793518, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": -0.20135729014873505, | |
| "logits/rejected": -0.25128036737442017, | |
| "logps/chosen": -545.1806640625, | |
| "logps/rejected": -568.894775390625, | |
| "loss": 0.1815, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.09586036950349808, | |
| "rewards/margins": 0.04870045185089111, | |
| "rewards/rejected": -0.1445608288049698, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": -0.2596682012081146, | |
| "logits/rejected": -0.22516381740570068, | |
| "logps/chosen": -498.10028076171875, | |
| "logps/rejected": -564.5962524414062, | |
| "loss": 0.1708, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.09488777071237564, | |
| "rewards/margins": 0.06974340975284576, | |
| "rewards/rejected": -0.164631187915802, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": -0.2944473624229431, | |
| "logits/rejected": -0.2125546932220459, | |
| "logps/chosen": -502.841552734375, | |
| "logps/rejected": -546.7924194335938, | |
| "loss": 0.1826, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.08710362762212753, | |
| "rewards/margins": 0.059937745332717896, | |
| "rewards/rejected": -0.14704139530658722, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": -0.2098701447248459, | |
| "logits/rejected": -0.2635635733604431, | |
| "logps/chosen": -481.09478759765625, | |
| "logps/rejected": -569.34326171875, | |
| "loss": 0.1763, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.09881605207920074, | |
| "rewards/margins": 0.06372065842151642, | |
| "rewards/rejected": -0.16253669559955597, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": -0.20950980484485626, | |
| "logits/rejected": -0.2304944545030594, | |
| "logps/chosen": -508.41241455078125, | |
| "logps/rejected": -574.2259521484375, | |
| "loss": 0.1727, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.0780680924654007, | |
| "rewards/margins": 0.06727245450019836, | |
| "rewards/rejected": -0.14534054696559906, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": -0.2829793095588684, | |
| "logits/rejected": -0.2987596392631531, | |
| "logps/chosen": -475.51031494140625, | |
| "logps/rejected": -521.2598876953125, | |
| "loss": 0.1865, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.09187036752700806, | |
| "rewards/margins": 0.06315977871417999, | |
| "rewards/rejected": -0.15503014624118805, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": -0.23163847625255585, | |
| "logits/rejected": -0.24427077174186707, | |
| "logps/chosen": -477.51507568359375, | |
| "logps/rejected": -542.298828125, | |
| "loss": 0.1848, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.0930924192070961, | |
| "rewards/margins": 0.06024498865008354, | |
| "rewards/rejected": -0.15333738923072815, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.1881250925163322, | |
| "train_runtime": 7837.4153, | |
| "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 | |
| } | |