| { | |
| "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": 0.14563202857971191, | |
| "logits/rejected": 0.3874575197696686, | |
| "logps/chosen": -454.742919921875, | |
| "logps/rejected": -268.25054931640625, | |
| "loss": 0.3396, | |
| "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": 0.07718739658594131, | |
| "logits/rejected": 0.26869574189186096, | |
| "logps/chosen": -350.5389709472656, | |
| "logps/rejected": -281.45709228515625, | |
| "loss": 0.315, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 2.5761094093468273e-06, | |
| "rewards/margins": -5.979803609079681e-05, | |
| "rewards/rejected": 6.237414345378056e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.051278989762067795, | |
| "logits/rejected": 0.2582473158836365, | |
| "logps/chosen": -354.6767578125, | |
| "logps/rejected": -278.68255615234375, | |
| "loss": 0.3562, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": 0.00019174082262907177, | |
| "rewards/margins": 5.2712995966430753e-05, | |
| "rewards/rejected": 0.00013902783393859863, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.087407685816288, | |
| "logits/rejected": 0.23584076762199402, | |
| "logps/chosen": -341.74859619140625, | |
| "logps/rejected": -292.8842468261719, | |
| "loss": 0.3276, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.00011408107820898294, | |
| "rewards/margins": 0.00017452315660193563, | |
| "rewards/rejected": -6.044206747901626e-05, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.07952137291431427, | |
| "logits/rejected": 0.18448325991630554, | |
| "logps/chosen": -345.06378173828125, | |
| "logps/rejected": -308.3074951171875, | |
| "loss": 0.3338, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": 6.922694592503831e-05, | |
| "rewards/margins": 0.0002068312169285491, | |
| "rewards/rejected": -0.0001376042637275532, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.1114521399140358, | |
| "logits/rejected": 0.2774251103401184, | |
| "logps/chosen": -326.83441162109375, | |
| "logps/rejected": -293.80291748046875, | |
| "loss": 0.3451, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": 0.00041769607923924923, | |
| "rewards/margins": 0.0006514322012662888, | |
| "rewards/rejected": -0.00023373604926746339, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.09482766687870026, | |
| "logits/rejected": 0.27632468938827515, | |
| "logps/chosen": -351.2109680175781, | |
| "logps/rejected": -277.0458068847656, | |
| "loss": 0.339, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.0010114657925441861, | |
| "rewards/margins": 0.0016303223092108965, | |
| "rewards/rejected": -0.0006188566330820322, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.02542046643793583, | |
| "logits/rejected": 0.19503046572208405, | |
| "logps/chosen": -325.17486572265625, | |
| "logps/rejected": -285.79205322265625, | |
| "loss": 0.3265, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": 0.001514864037744701, | |
| "rewards/margins": 0.002718320582062006, | |
| "rewards/rejected": -0.0012034563114866614, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.13358162343502045, | |
| "logits/rejected": 0.31503233313560486, | |
| "logps/chosen": -368.6209716796875, | |
| "logps/rejected": -295.80133056640625, | |
| "loss": 0.3393, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.002275743056088686, | |
| "rewards/margins": 0.005096759647130966, | |
| "rewards/rejected": -0.0028210170567035675, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.08928992599248886, | |
| "logits/rejected": 0.24500127136707306, | |
| "logps/chosen": -346.7084045410156, | |
| "logps/rejected": -290.81646728515625, | |
| "loss": 0.3163, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": 0.004672099836170673, | |
| "rewards/margins": 0.009218016639351845, | |
| "rewards/rejected": -0.004545916803181171, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.17046232521533966, | |
| "logits/rejected": 0.22814074158668518, | |
| "logps/chosen": -332.4991760253906, | |
| "logps/rejected": -293.03314208984375, | |
| "loss": 0.3288, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": 0.0032020125072449446, | |
| "rewards/margins": 0.01331277471035719, | |
| "rewards/rejected": -0.010110762901604176, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.13835321366786957, | |
| "logits/rejected": 0.2707987427711487, | |
| "logps/chosen": -353.07904052734375, | |
| "logps/rejected": -316.55377197265625, | |
| "loss": 0.3145, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.0036005068104714155, | |
| "rewards/margins": 0.02047211304306984, | |
| "rewards/rejected": -0.024072617292404175, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.14788897335529327, | |
| "logits/rejected": 0.2877771258354187, | |
| "logps/chosen": -368.3328857421875, | |
| "logps/rejected": -326.0980224609375, | |
| "loss": 0.3211, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.009556634351611137, | |
| "rewards/margins": 0.031165510416030884, | |
| "rewards/rejected": -0.04072214290499687, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.09604962915182114, | |
| "logits/rejected": 0.29759398102760315, | |
| "logps/chosen": -394.56146240234375, | |
| "logps/rejected": -389.3465270996094, | |
| "loss": 0.2595, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.04469118267297745, | |
| "rewards/margins": 0.059812031686306, | |
| "rewards/rejected": -0.10450321435928345, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.09464170038700104, | |
| "logits/rejected": 0.23635534942150116, | |
| "logps/chosen": -411.34344482421875, | |
| "logps/rejected": -472.7259216308594, | |
| "loss": 0.2809, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.09930221736431122, | |
| "rewards/margins": 0.08144193887710571, | |
| "rewards/rejected": -0.18074414134025574, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.15742775797843933, | |
| "logits/rejected": 0.2725524306297302, | |
| "logps/chosen": -477.2586975097656, | |
| "logps/rejected": -517.0968627929688, | |
| "loss": 0.2592, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.13660530745983124, | |
| "rewards/margins": 0.11236293613910675, | |
| "rewards/rejected": -0.2489682137966156, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.19448170065879822, | |
| "logits/rejected": 0.25956863164901733, | |
| "logps/chosen": -487.07196044921875, | |
| "logps/rejected": -597.939208984375, | |
| "loss": 0.2663, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.15577515959739685, | |
| "rewards/margins": 0.15100589394569397, | |
| "rewards/rejected": -0.3067810833454132, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.18750979006290436, | |
| "logits/rejected": 0.326152503490448, | |
| "logps/chosen": -559.5252685546875, | |
| "logps/rejected": -645.9837036132812, | |
| "loss": 0.2719, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.22856616973876953, | |
| "rewards/margins": 0.12326528877019882, | |
| "rewards/rejected": -0.35183146595954895, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.16625967621803284, | |
| "logits/rejected": 0.29693764448165894, | |
| "logps/chosen": -530.322021484375, | |
| "logps/rejected": -600.0885009765625, | |
| "loss": 0.2955, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.18995243310928345, | |
| "rewards/margins": 0.14340588450431824, | |
| "rewards/rejected": -0.3333582878112793, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.135554239153862, | |
| "logits/rejected": 0.26710349321365356, | |
| "logps/chosen": -582.8619384765625, | |
| "logps/rejected": -616.3886108398438, | |
| "loss": 0.2737, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.21700850129127502, | |
| "rewards/margins": 0.12937961518764496, | |
| "rewards/rejected": -0.3463881015777588, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": 0.10434029996395111, | |
| "logits/rejected": 0.28046977519989014, | |
| "logps/chosen": -585.6417236328125, | |
| "logps/rejected": -614.5296630859375, | |
| "loss": 0.2886, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2205405980348587, | |
| "rewards/margins": 0.12526755034923553, | |
| "rewards/rejected": -0.34580811858177185, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.1413886696100235, | |
| "logits/rejected": 0.27238231897354126, | |
| "logps/chosen": -519.6276245117188, | |
| "logps/rejected": -587.7799072265625, | |
| "loss": 0.2557, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.16974905133247375, | |
| "rewards/margins": 0.1322554051876068, | |
| "rewards/rejected": -0.30200448632240295, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.1467094123363495, | |
| "logits/rejected": 0.2754250466823578, | |
| "logps/chosen": -554.9280395507812, | |
| "logps/rejected": -615.7232666015625, | |
| "loss": 0.2471, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.21137747168540955, | |
| "rewards/margins": 0.1308281123638153, | |
| "rewards/rejected": -0.34220558404922485, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": 0.1171431764960289, | |
| "logits/rejected": 0.2890062928199768, | |
| "logps/chosen": -575.1917724609375, | |
| "logps/rejected": -651.8801879882812, | |
| "loss": 0.244, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.24391531944274902, | |
| "rewards/margins": 0.15722794830799103, | |
| "rewards/rejected": -0.40114325284957886, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": 0.10296036303043365, | |
| "logits/rejected": 0.28510022163391113, | |
| "logps/chosen": -640.1365966796875, | |
| "logps/rejected": -720.7387084960938, | |
| "loss": 0.2331, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.2727048397064209, | |
| "rewards/margins": 0.1723267138004303, | |
| "rewards/rejected": -0.4450315833091736, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": 0.15176770091056824, | |
| "logits/rejected": 0.22689513862133026, | |
| "logps/chosen": -542.8187255859375, | |
| "logps/rejected": -679.3350830078125, | |
| "loss": 0.2444, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.23740482330322266, | |
| "rewards/margins": 0.16511742770671844, | |
| "rewards/rejected": -0.4025222659111023, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": 0.1848079115152359, | |
| "logits/rejected": 0.28918254375457764, | |
| "logps/chosen": -591.1017456054688, | |
| "logps/rejected": -719.2144775390625, | |
| "loss": 0.2388, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.23014459013938904, | |
| "rewards/margins": 0.18921832740306854, | |
| "rewards/rejected": -0.4193629324436188, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": 0.16476555168628693, | |
| "logits/rejected": 0.23265871405601501, | |
| "logps/chosen": -612.1968383789062, | |
| "logps/rejected": -663.5397338867188, | |
| "loss": 0.2918, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.24918679893016815, | |
| "rewards/margins": 0.11979563534259796, | |
| "rewards/rejected": -0.3689824044704437, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": 0.180597186088562, | |
| "logits/rejected": 0.3083464503288269, | |
| "logps/chosen": -560.151123046875, | |
| "logps/rejected": -635.0821533203125, | |
| "loss": 0.2582, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2051992118358612, | |
| "rewards/margins": 0.14893193542957306, | |
| "rewards/rejected": -0.35413113236427307, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": 0.22149360179901123, | |
| "logits/rejected": 0.3726951479911804, | |
| "logps/chosen": -628.3652954101562, | |
| "logps/rejected": -711.3004150390625, | |
| "loss": 0.2145, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.24089030921459198, | |
| "rewards/margins": 0.16530363261699677, | |
| "rewards/rejected": -0.40619391202926636, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": 0.17435990273952484, | |
| "logits/rejected": 0.3360474705696106, | |
| "logps/chosen": -610.1092529296875, | |
| "logps/rejected": -695.5682983398438, | |
| "loss": 0.2356, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.26700595021247864, | |
| "rewards/margins": 0.1659780740737915, | |
| "rewards/rejected": -0.43298402428627014, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": 0.177028626203537, | |
| "logits/rejected": 0.2775452136993408, | |
| "logps/chosen": -601.6631469726562, | |
| "logps/rejected": -679.7595825195312, | |
| "loss": 0.2454, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2515362799167633, | |
| "rewards/margins": 0.15310627222061157, | |
| "rewards/rejected": -0.4046425223350525, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": 0.17227385938167572, | |
| "logits/rejected": 0.2626027464866638, | |
| "logps/chosen": -565.1954956054688, | |
| "logps/rejected": -653.8951416015625, | |
| "loss": 0.2601, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.22992494702339172, | |
| "rewards/margins": 0.14292170107364655, | |
| "rewards/rejected": -0.3728466331958771, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": 0.17768487334251404, | |
| "logits/rejected": 0.33285486698150635, | |
| "logps/chosen": -610.06201171875, | |
| "logps/rejected": -670.5873413085938, | |
| "loss": 0.238, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.23991577327251434, | |
| "rewards/margins": 0.15924175083637238, | |
| "rewards/rejected": -0.39915746450424194, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": 0.12916894257068634, | |
| "logits/rejected": 0.2785950303077698, | |
| "logps/chosen": -571.642578125, | |
| "logps/rejected": -680.286865234375, | |
| "loss": 0.2314, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.2410048544406891, | |
| "rewards/margins": 0.15611061453819275, | |
| "rewards/rejected": -0.39711546897888184, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": 0.1318448781967163, | |
| "logits/rejected": 0.2396216094493866, | |
| "logps/chosen": -577.1714477539062, | |
| "logps/rejected": -694.001220703125, | |
| "loss": 0.23, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.25458115339279175, | |
| "rewards/margins": 0.16680863499641418, | |
| "rewards/rejected": -0.42138975858688354, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": 0.1819082796573639, | |
| "logits/rejected": 0.30339935421943665, | |
| "logps/chosen": -620.7168579101562, | |
| "logps/rejected": -733.3218994140625, | |
| "loss": 0.2306, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.24643263220787048, | |
| "rewards/margins": 0.19664740562438965, | |
| "rewards/rejected": -0.4430800974369049, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": 0.1591918021440506, | |
| "logits/rejected": 0.24382638931274414, | |
| "logps/chosen": -650.5294189453125, | |
| "logps/rejected": -757.6646118164062, | |
| "loss": 0.2606, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.29114747047424316, | |
| "rewards/margins": 0.1727633774280548, | |
| "rewards/rejected": -0.46391087770462036, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": 0.15119919180870056, | |
| "logits/rejected": 0.24405035376548767, | |
| "logps/chosen": -622.4024658203125, | |
| "logps/rejected": -731.4509887695312, | |
| "loss": 0.2127, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.2556970715522766, | |
| "rewards/margins": 0.20232219994068146, | |
| "rewards/rejected": -0.4580192565917969, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": 0.16059614717960358, | |
| "logits/rejected": 0.26955538988113403, | |
| "logps/chosen": -620.3678588867188, | |
| "logps/rejected": -734.6710205078125, | |
| "loss": 0.2319, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.26077157258987427, | |
| "rewards/margins": 0.18923643231391907, | |
| "rewards/rejected": -0.4500080645084381, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": 0.15299072861671448, | |
| "logits/rejected": 0.27529269456863403, | |
| "logps/chosen": -626.7930908203125, | |
| "logps/rejected": -771.9212646484375, | |
| "loss": 0.1765, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.25388702750205994, | |
| "rewards/margins": 0.23886942863464355, | |
| "rewards/rejected": -0.49275651574134827, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": 0.10297727584838867, | |
| "logits/rejected": 0.2494010478258133, | |
| "logps/chosen": -701.2484130859375, | |
| "logps/rejected": -797.5471801757812, | |
| "loss": 0.2255, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.3326496481895447, | |
| "rewards/margins": 0.19009152054786682, | |
| "rewards/rejected": -0.5227411985397339, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": 0.0838174819946289, | |
| "logits/rejected": 0.24609944224357605, | |
| "logps/chosen": -691.7960205078125, | |
| "logps/rejected": -792.8016357421875, | |
| "loss": 0.2473, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3393723964691162, | |
| "rewards/margins": 0.17680588364601135, | |
| "rewards/rejected": -0.5161782503128052, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": 0.13857319951057434, | |
| "logits/rejected": 0.25091537833213806, | |
| "logps/chosen": -635.0294189453125, | |
| "logps/rejected": -770.2885131835938, | |
| "loss": 0.2029, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.28675633668899536, | |
| "rewards/margins": 0.20502133667469025, | |
| "rewards/rejected": -0.4917777180671692, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": 0.09761005640029907, | |
| "logits/rejected": 0.21226266026496887, | |
| "logps/chosen": -669.2428588867188, | |
| "logps/rejected": -749.3931884765625, | |
| "loss": 0.242, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.3128831088542938, | |
| "rewards/margins": 0.17808611690998077, | |
| "rewards/rejected": -0.4909692406654358, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": 0.1553424596786499, | |
| "logits/rejected": 0.25895553827285767, | |
| "logps/chosen": -694.4083862304688, | |
| "logps/rejected": -792.1837158203125, | |
| "loss": 0.2174, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.3040028214454651, | |
| "rewards/margins": 0.2048918753862381, | |
| "rewards/rejected": -0.508894681930542, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": 0.1065811887383461, | |
| "logits/rejected": 0.25707563757896423, | |
| "logps/chosen": -613.1231689453125, | |
| "logps/rejected": -758.7059326171875, | |
| "loss": 0.2107, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.26900747418403625, | |
| "rewards/margins": 0.21452918648719788, | |
| "rewards/rejected": -0.48353663086891174, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": 0.16053031384944916, | |
| "logits/rejected": 0.2852506935596466, | |
| "logps/chosen": -637.3067626953125, | |
| "logps/rejected": -732.3541259765625, | |
| "loss": 0.2118, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2603939175605774, | |
| "rewards/margins": 0.18671852350234985, | |
| "rewards/rejected": -0.44711247086524963, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": 0.1389404833316803, | |
| "logits/rejected": 0.2467120885848999, | |
| "logps/chosen": -638.651123046875, | |
| "logps/rejected": -754.9122314453125, | |
| "loss": 0.2233, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.26695486903190613, | |
| "rewards/margins": 0.20951655507087708, | |
| "rewards/rejected": -0.476471483707428, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": 0.11784114688634872, | |
| "logits/rejected": 0.18245339393615723, | |
| "logps/chosen": -603.6224365234375, | |
| "logps/rejected": -770.082275390625, | |
| "loss": 0.2182, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.28938090801239014, | |
| "rewards/margins": 0.2138332575559616, | |
| "rewards/rejected": -0.5032141804695129, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": 0.11818281561136246, | |
| "logits/rejected": 0.26330405473709106, | |
| "logps/chosen": -623.1153564453125, | |
| "logps/rejected": -809.2369384765625, | |
| "loss": 0.2143, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3059650957584381, | |
| "rewards/margins": 0.23646679520606995, | |
| "rewards/rejected": -0.5424319505691528, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": 0.16755148768424988, | |
| "logits/rejected": 0.28101032972335815, | |
| "logps/chosen": -677.8587646484375, | |
| "logps/rejected": -864.1910400390625, | |
| "loss": 0.2025, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.32487982511520386, | |
| "rewards/margins": 0.2524875998497009, | |
| "rewards/rejected": -0.5773674249649048, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": 0.08524368703365326, | |
| "logits/rejected": 0.21973176300525665, | |
| "logps/chosen": -697.4478759765625, | |
| "logps/rejected": -855.9630737304688, | |
| "loss": 0.22, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.33033424615859985, | |
| "rewards/margins": 0.21664047241210938, | |
| "rewards/rejected": -0.5469747185707092, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": 0.1285276859998703, | |
| "logits/rejected": 0.2174634039402008, | |
| "logps/chosen": -610.0903930664062, | |
| "logps/rejected": -795.0265502929688, | |
| "loss": 0.1928, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.293308824300766, | |
| "rewards/margins": 0.21643824875354767, | |
| "rewards/rejected": -0.5097470879554749, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": 0.14484889805316925, | |
| "logits/rejected": 0.20977553725242615, | |
| "logps/chosen": -650.8162231445312, | |
| "logps/rejected": -814.1326293945312, | |
| "loss": 0.2019, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.2902231812477112, | |
| "rewards/margins": 0.24104826152324677, | |
| "rewards/rejected": -0.5312714576721191, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": 0.11247887462377548, | |
| "logits/rejected": 0.22003412246704102, | |
| "logps/chosen": -649.1162719726562, | |
| "logps/rejected": -767.150634765625, | |
| "loss": 0.2465, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.3138406574726105, | |
| "rewards/margins": 0.182392418384552, | |
| "rewards/rejected": -0.49623313546180725, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": 0.057500071823596954, | |
| "logits/rejected": 0.19665978848934174, | |
| "logps/chosen": -651.9990844726562, | |
| "logps/rejected": -783.0087890625, | |
| "loss": 0.2255, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.3106940984725952, | |
| "rewards/margins": 0.20310230553150177, | |
| "rewards/rejected": -0.5137964487075806, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": 0.11110985279083252, | |
| "logits/rejected": 0.2312009632587433, | |
| "logps/chosen": -621.1902465820312, | |
| "logps/rejected": -777.6902465820312, | |
| "loss": 0.1934, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.25909268856048584, | |
| "rewards/margins": 0.2260764092206955, | |
| "rewards/rejected": -0.48516911268234253, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": 0.14112098515033722, | |
| "logits/rejected": 0.2636359930038452, | |
| "logps/chosen": -635.9463500976562, | |
| "logps/rejected": -735.6661376953125, | |
| "loss": 0.2452, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.266891747713089, | |
| "rewards/margins": 0.19298222661018372, | |
| "rewards/rejected": -0.4598739743232727, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": 0.11148792505264282, | |
| "logits/rejected": 0.25395482778549194, | |
| "logps/chosen": -586.4432373046875, | |
| "logps/rejected": -725.523193359375, | |
| "loss": 0.224, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2619563043117523, | |
| "rewards/margins": 0.19483323395252228, | |
| "rewards/rejected": -0.4567895531654358, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": 0.09673704952001572, | |
| "logits/rejected": 0.20130054652690887, | |
| "logps/chosen": -615.4747314453125, | |
| "logps/rejected": -788.6925659179688, | |
| "loss": 0.2763, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.27522048354148865, | |
| "rewards/margins": 0.21255996823310852, | |
| "rewards/rejected": -0.48778048157691956, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": 0.07239886373281479, | |
| "logits/rejected": 0.18439838290214539, | |
| "logps/chosen": -564.4202880859375, | |
| "logps/rejected": -704.7264404296875, | |
| "loss": 0.2415, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.25343528389930725, | |
| "rewards/margins": 0.1700516790151596, | |
| "rewards/rejected": -0.42348694801330566, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": 0.10070653259754181, | |
| "logits/rejected": 0.19296300411224365, | |
| "logps/chosen": -595.9661865234375, | |
| "logps/rejected": -724.99658203125, | |
| "loss": 0.2326, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.23254573345184326, | |
| "rewards/margins": 0.19659695029258728, | |
| "rewards/rejected": -0.42914265394210815, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": 0.08602897822856903, | |
| "logits/rejected": 0.1648799628019333, | |
| "logps/chosen": -553.3023681640625, | |
| "logps/rejected": -709.7655639648438, | |
| "loss": 0.2185, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.2514170706272125, | |
| "rewards/margins": 0.1915900856256485, | |
| "rewards/rejected": -0.4430071711540222, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": 0.0444771870970726, | |
| "logits/rejected": 0.16466517746448517, | |
| "logps/chosen": -611.7086181640625, | |
| "logps/rejected": -756.202392578125, | |
| "loss": 0.2093, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.27020856738090515, | |
| "rewards/margins": 0.20864149928092957, | |
| "rewards/rejected": -0.4788500666618347, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": 0.08262890577316284, | |
| "logits/rejected": 0.21953897178173065, | |
| "logps/chosen": -657.6599731445312, | |
| "logps/rejected": -753.6175537109375, | |
| "loss": 0.2635, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.30637097358703613, | |
| "rewards/margins": 0.1610550731420517, | |
| "rewards/rejected": -0.46742600202560425, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": 0.08547943085432053, | |
| "logits/rejected": 0.22080206871032715, | |
| "logps/chosen": -639.3602905273438, | |
| "logps/rejected": -765.0423583984375, | |
| "loss": 0.2281, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2679855227470398, | |
| "rewards/margins": 0.2209259271621704, | |
| "rewards/rejected": -0.4889114797115326, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": 0.10255439579486847, | |
| "logits/rejected": 0.17642661929130554, | |
| "logps/chosen": -590.3636474609375, | |
| "logps/rejected": -765.87890625, | |
| "loss": 0.2285, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.2688559293746948, | |
| "rewards/margins": 0.2031112015247345, | |
| "rewards/rejected": -0.47196707129478455, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": 0.09613371640443802, | |
| "logits/rejected": 0.20681801438331604, | |
| "logps/chosen": -639.8070678710938, | |
| "logps/rejected": -783.1290283203125, | |
| "loss": 0.2284, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.28017768263816833, | |
| "rewards/margins": 0.2119220495223999, | |
| "rewards/rejected": -0.4920997619628906, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": 0.10619207471609116, | |
| "logits/rejected": 0.19468547403812408, | |
| "logps/chosen": -697.66162109375, | |
| "logps/rejected": -777.8093872070312, | |
| "loss": 0.2456, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.31186243891716003, | |
| "rewards/margins": 0.1904953420162201, | |
| "rewards/rejected": -0.5023577809333801, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": 0.12791021168231964, | |
| "logits/rejected": 0.18563024699687958, | |
| "logps/chosen": -692.7999267578125, | |
| "logps/rejected": -838.0908203125, | |
| "loss": 0.2424, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.30099743604660034, | |
| "rewards/margins": 0.24136948585510254, | |
| "rewards/rejected": -0.5423668622970581, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": 0.10475460439920425, | |
| "logits/rejected": 0.17532803118228912, | |
| "logps/chosen": -574.7904663085938, | |
| "logps/rejected": -732.5684814453125, | |
| "loss": 0.2116, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.27657538652420044, | |
| "rewards/margins": 0.19332227110862732, | |
| "rewards/rejected": -0.46989765763282776, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": 0.11339239031076431, | |
| "logits/rejected": 0.23391477763652802, | |
| "logps/chosen": -614.44384765625, | |
| "logps/rejected": -739.0775756835938, | |
| "loss": 0.2129, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.26443395018577576, | |
| "rewards/margins": 0.20620107650756836, | |
| "rewards/rejected": -0.47063499689102173, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": 0.11114220321178436, | |
| "logits/rejected": 0.1741330623626709, | |
| "logps/chosen": -649.84423828125, | |
| "logps/rejected": -796.5662841796875, | |
| "loss": 0.249, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.30554836988449097, | |
| "rewards/margins": 0.18621903657913208, | |
| "rewards/rejected": -0.49176740646362305, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": 0.05583947151899338, | |
| "logits/rejected": 0.208854079246521, | |
| "logps/chosen": -677.1410522460938, | |
| "logps/rejected": -806.2962646484375, | |
| "loss": 0.2157, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.3019706606864929, | |
| "rewards/margins": 0.21764162182807922, | |
| "rewards/rejected": -0.5196123123168945, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": 0.11103875935077667, | |
| "logits/rejected": 0.23960928618907928, | |
| "logps/chosen": -651.9210205078125, | |
| "logps/rejected": -767.1956176757812, | |
| "loss": 0.2196, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.28146857023239136, | |
| "rewards/margins": 0.21090731024742126, | |
| "rewards/rejected": -0.492375910282135, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": 0.08522719889879227, | |
| "logits/rejected": 0.18019166588783264, | |
| "logps/chosen": -693.9295043945312, | |
| "logps/rejected": -834.8268432617188, | |
| "loss": 0.2044, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.3021830916404724, | |
| "rewards/margins": 0.22599780559539795, | |
| "rewards/rejected": -0.5281809568405151, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": 0.10727125406265259, | |
| "logits/rejected": 0.2462559938430786, | |
| "logps/chosen": -690.4157104492188, | |
| "logps/rejected": -837.8204956054688, | |
| "loss": 0.2228, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.3111000657081604, | |
| "rewards/margins": 0.22689378261566162, | |
| "rewards/rejected": -0.5379937887191772, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": 0.10468622297048569, | |
| "logits/rejected": 0.21340981125831604, | |
| "logps/chosen": -664.1920166015625, | |
| "logps/rejected": -791.6151123046875, | |
| "loss": 0.2025, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.28642627596855164, | |
| "rewards/margins": 0.2306082546710968, | |
| "rewards/rejected": -0.5170345306396484, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": 0.11628569662570953, | |
| "logits/rejected": 0.23161634802818298, | |
| "logps/chosen": -672.2249755859375, | |
| "logps/rejected": -801.2711181640625, | |
| "loss": 0.2464, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3202977776527405, | |
| "rewards/margins": 0.2009088099002838, | |
| "rewards/rejected": -0.5212065577507019, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": 0.06959172338247299, | |
| "logits/rejected": 0.2281877100467682, | |
| "logps/chosen": -652.8377685546875, | |
| "logps/rejected": -795.1046752929688, | |
| "loss": 0.2284, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.2909380793571472, | |
| "rewards/margins": 0.22871951758861542, | |
| "rewards/rejected": -0.5196576714515686, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": 0.11237098276615143, | |
| "logits/rejected": 0.21487274765968323, | |
| "logps/chosen": -660.4642944335938, | |
| "logps/rejected": -802.35009765625, | |
| "loss": 0.196, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.31749480962753296, | |
| "rewards/margins": 0.21312642097473145, | |
| "rewards/rejected": -0.5306212306022644, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": 0.04496614262461662, | |
| "logits/rejected": 0.25041455030441284, | |
| "logps/chosen": -633.7536010742188, | |
| "logps/rejected": -752.9474487304688, | |
| "loss": 0.2326, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.303285151720047, | |
| "rewards/margins": 0.18882021307945251, | |
| "rewards/rejected": -0.4921053349971771, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": 0.07390692830085754, | |
| "logits/rejected": 0.17046715319156647, | |
| "logps/chosen": -650.1712646484375, | |
| "logps/rejected": -773.189697265625, | |
| "loss": 0.2534, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.32713738083839417, | |
| "rewards/margins": 0.16583597660064697, | |
| "rewards/rejected": -0.4929733872413635, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": 0.08455213159322739, | |
| "logits/rejected": 0.2394600361585617, | |
| "logps/chosen": -624.288818359375, | |
| "logps/rejected": -786.5145263671875, | |
| "loss": 0.2184, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.28965824842453003, | |
| "rewards/margins": 0.20828434824943542, | |
| "rewards/rejected": -0.49794262647628784, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": 0.08053167164325714, | |
| "logits/rejected": 0.1703474223613739, | |
| "logps/chosen": -638.90966796875, | |
| "logps/rejected": -760.4950561523438, | |
| "loss": 0.2568, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.29267439246177673, | |
| "rewards/margins": 0.18851785361766815, | |
| "rewards/rejected": -0.48119229078292847, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": 0.08862993866205215, | |
| "logits/rejected": 0.2355298101902008, | |
| "logps/chosen": -659.1481323242188, | |
| "logps/rejected": -788.7034912109375, | |
| "loss": 0.2149, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.28642982244491577, | |
| "rewards/margins": 0.22406980395317078, | |
| "rewards/rejected": -0.5104995965957642, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": 0.09468914568424225, | |
| "logits/rejected": 0.17423442006111145, | |
| "logps/chosen": -628.0216064453125, | |
| "logps/rejected": -736.0591430664062, | |
| "loss": 0.2635, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.29823777079582214, | |
| "rewards/margins": 0.16002070903778076, | |
| "rewards/rejected": -0.4582585394382477, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": 0.1822114735841751, | |
| "logits/rejected": 0.25673800706863403, | |
| "logps/chosen": -647.0616455078125, | |
| "logps/rejected": -785.24853515625, | |
| "loss": 0.2055, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2835825979709625, | |
| "rewards/margins": 0.21645894646644592, | |
| "rewards/rejected": -0.5000415444374084, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": 0.08051988482475281, | |
| "logits/rejected": 0.22623100876808167, | |
| "logps/chosen": -661.945068359375, | |
| "logps/rejected": -801.7554321289062, | |
| "loss": 0.279, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.3187974691390991, | |
| "rewards/margins": 0.20359888672828674, | |
| "rewards/rejected": -0.5223963260650635, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": 0.09252520650625229, | |
| "logits/rejected": 0.2231016904115677, | |
| "logps/chosen": -594.2275390625, | |
| "logps/rejected": -788.0599975585938, | |
| "loss": 0.2146, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.2893867492675781, | |
| "rewards/margins": 0.22611062228679657, | |
| "rewards/rejected": -0.5154973864555359, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": 0.06350111961364746, | |
| "logits/rejected": 0.20778074860572815, | |
| "logps/chosen": -707.4957275390625, | |
| "logps/rejected": -834.1343994140625, | |
| "loss": 0.2248, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.3360898196697235, | |
| "rewards/margins": 0.18199169635772705, | |
| "rewards/rejected": -0.518081545829773, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": 0.08399736881256104, | |
| "logits/rejected": 0.18887805938720703, | |
| "logps/chosen": -649.8248291015625, | |
| "logps/rejected": -780.5101318359375, | |
| "loss": 0.212, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.2871829569339752, | |
| "rewards/margins": 0.21405375003814697, | |
| "rewards/rejected": -0.5012367367744446, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": 0.10902540385723114, | |
| "logits/rejected": 0.24318072199821472, | |
| "logps/chosen": -606.8221435546875, | |
| "logps/rejected": -784.5928955078125, | |
| "loss": 0.2001, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.284158855676651, | |
| "rewards/margins": 0.23260316252708435, | |
| "rewards/rejected": -0.5167620778083801, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.00839476689712215, | |
| "train_runtime": 334.0046, | |
| "train_samples_per_second": 44.91, | |
| "train_steps_per_second": 2.805 | |
| } | |
| ], | |
| "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 | |
| } | |