| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1875, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.6595744680851065e-08, | |
| "logits/chosen": 0.09666420519351959, | |
| "logits/rejected": 0.3057119846343994, | |
| "logps/chosen": -431.574951171875, | |
| "logps/rejected": -283.99554443359375, | |
| "loss": 0.3643, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.6595744680851066e-07, | |
| "logits/chosen": 0.06902924180030823, | |
| "logits/rejected": 0.19579637050628662, | |
| "logps/chosen": -345.7933654785156, | |
| "logps/rejected": -284.0511779785156, | |
| "loss": 0.3431, | |
| "rewards/accuracies": 0.4027777910232544, | |
| "rewards/chosen": -5.811791925225407e-05, | |
| "rewards/margins": -0.00014306220691651106, | |
| "rewards/rejected": 8.494428766425699e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.319148936170213e-07, | |
| "logits/chosen": 0.07102914154529572, | |
| "logits/rejected": 0.24960489571094513, | |
| "logps/chosen": -328.02667236328125, | |
| "logps/rejected": -273.9169006347656, | |
| "loss": 0.3699, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -2.8161617592559196e-05, | |
| "rewards/margins": -6.676381599390879e-05, | |
| "rewards/rejected": 3.860218566842377e-05, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.97872340425532e-07, | |
| "logits/chosen": 0.10551998764276505, | |
| "logits/rejected": 0.2433021515607834, | |
| "logps/chosen": -354.22650146484375, | |
| "logps/rejected": -284.6506042480469, | |
| "loss": 0.3402, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": 0.00011991789506282657, | |
| "rewards/margins": 2.2664942207484273e-06, | |
| "rewards/rejected": 0.00011765138333430514, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.12176469713449478, | |
| "logits/rejected": 0.3288267254829407, | |
| "logps/chosen": -371.8106384277344, | |
| "logps/rejected": -293.2636413574219, | |
| "loss": 0.3463, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": 0.0001882923679659143, | |
| "rewards/margins": 0.00011801043001469225, | |
| "rewards/rejected": 7.028193067526445e-05, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3297872340425533e-06, | |
| "logits/chosen": 0.1286633312702179, | |
| "logits/rejected": 0.13883765041828156, | |
| "logps/chosen": -318.04205322265625, | |
| "logps/rejected": -258.83013916015625, | |
| "loss": 0.3691, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": 0.0002136583934770897, | |
| "rewards/margins": 6.237901106942445e-05, | |
| "rewards/rejected": 0.0001512793533038348, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.11209549754858017, | |
| "logits/rejected": 0.25248101353645325, | |
| "logps/chosen": -362.73858642578125, | |
| "logps/rejected": -315.79681396484375, | |
| "loss": 0.3559, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.00025600154185667634, | |
| "rewards/margins": 0.00027534778928384185, | |
| "rewards/rejected": -1.9346229237271473e-05, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.8617021276595745e-06, | |
| "logits/chosen": 0.11582938581705093, | |
| "logits/rejected": 0.23768457770347595, | |
| "logps/chosen": -365.4197998046875, | |
| "logps/rejected": -307.07269287109375, | |
| "loss": 0.3196, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.0007645456935279071, | |
| "rewards/margins": 0.000753869186155498, | |
| "rewards/rejected": 1.0676437341317069e-05, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.09230409562587738, | |
| "logits/rejected": 0.23419447243213654, | |
| "logps/chosen": -309.068603515625, | |
| "logps/rejected": -264.81463623046875, | |
| "loss": 0.3248, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.0010737198172137141, | |
| "rewards/margins": 0.0011370908468961716, | |
| "rewards/rejected": -6.337100057862699e-05, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.393617021276596e-06, | |
| "logits/chosen": 0.10440310090780258, | |
| "logits/rejected": 0.23364253342151642, | |
| "logps/chosen": -349.4715881347656, | |
| "logps/rejected": -306.0603332519531, | |
| "loss": 0.3274, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.0017630963120609522, | |
| "rewards/margins": 0.0020842214580625296, | |
| "rewards/rejected": -0.00032112523331306875, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.11755403131246567, | |
| "logits/rejected": 0.20094576478004456, | |
| "logps/chosen": -346.9091796875, | |
| "logps/rejected": -281.81396484375, | |
| "loss": 0.3318, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": 0.002608776092529297, | |
| "rewards/margins": 0.0030841627158224583, | |
| "rewards/rejected": -0.00047538644867017865, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.9255319148936174e-06, | |
| "logits/chosen": 0.0515226311981678, | |
| "logits/rejected": 0.22483673691749573, | |
| "logps/chosen": -371.9663391113281, | |
| "logps/rejected": -313.88739013671875, | |
| "loss": 0.3365, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.0030859257094562054, | |
| "rewards/margins": 0.003779615741223097, | |
| "rewards/rejected": -0.0006936899153515697, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.07034258544445038, | |
| "logits/rejected": 0.16451817750930786, | |
| "logps/chosen": -297.6949768066406, | |
| "logps/rejected": -275.7294616699219, | |
| "loss": 0.3283, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.0033937774132937193, | |
| "rewards/margins": 0.004883460234850645, | |
| "rewards/rejected": -0.0014896828215569258, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.457446808510639e-06, | |
| "logits/chosen": 0.08784133940935135, | |
| "logits/rejected": 0.23549315333366394, | |
| "logps/chosen": -326.3392639160156, | |
| "logps/rejected": -306.0093994140625, | |
| "loss": 0.3241, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.0068435766734182835, | |
| "rewards/margins": 0.011338241398334503, | |
| "rewards/rejected": -0.00449466472491622, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.03845987841486931, | |
| "logits/rejected": 0.23602671921253204, | |
| "logps/chosen": -300.4671630859375, | |
| "logps/rejected": -283.1863708496094, | |
| "loss": 0.3267, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.007321356330066919, | |
| "rewards/margins": 0.014485524967312813, | |
| "rewards/rejected": -0.007164168171584606, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.98936170212766e-06, | |
| "logits/chosen": 0.14375978708267212, | |
| "logits/rejected": 0.21812143921852112, | |
| "logps/chosen": -331.6351318359375, | |
| "logps/rejected": -298.9814453125, | |
| "loss": 0.3286, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": 0.008922559209167957, | |
| "rewards/margins": 0.023942312225699425, | |
| "rewards/rejected": -0.015019752085208893, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.1490752249956131, | |
| "logits/rejected": 0.27425771951675415, | |
| "logps/chosen": -365.9402160644531, | |
| "logps/rejected": -357.152587890625, | |
| "loss": 0.297, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.005502852611243725, | |
| "rewards/margins": 0.04779756814241409, | |
| "rewards/rejected": -0.05330042913556099, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.521276595744681e-06, | |
| "logits/chosen": 0.14248177409172058, | |
| "logits/rejected": 0.24203705787658691, | |
| "logps/chosen": -390.8301696777344, | |
| "logps/rejected": -387.36309814453125, | |
| "loss": 0.2796, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.05510077625513077, | |
| "rewards/margins": 0.06211589649319649, | |
| "rewards/rejected": -0.11721666902303696, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.10283932834863663, | |
| "logits/rejected": 0.1948317587375641, | |
| "logps/chosen": -523.36962890625, | |
| "logps/rejected": -596.3675537109375, | |
| "loss": 0.2893, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.17172859609127045, | |
| "rewards/margins": 0.10832329094409943, | |
| "rewards/rejected": -0.2800518572330475, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.999982660399688e-06, | |
| "logits/chosen": 0.13617555797100067, | |
| "logits/rejected": 0.2513347566127777, | |
| "logps/chosen": -615.2536010742188, | |
| "logps/rejected": -724.0734252929688, | |
| "loss": 0.2269, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.24617168307304382, | |
| "rewards/margins": 0.17905732989311218, | |
| "rewards/rejected": -0.425229012966156, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.99937579964398e-06, | |
| "logits/chosen": 0.11373931169509888, | |
| "logits/rejected": 0.21447698771953583, | |
| "logps/chosen": -669.2640380859375, | |
| "logps/rejected": -688.9365844726562, | |
| "loss": 0.3081, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.27371013164520264, | |
| "rewards/margins": 0.124347984790802, | |
| "rewards/rejected": -0.3980581760406494, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9979021993870645e-06, | |
| "logits/chosen": 0.17223823070526123, | |
| "logits/rejected": 0.25994282960891724, | |
| "logps/chosen": -524.0916137695312, | |
| "logps/rejected": -528.10595703125, | |
| "loss": 0.3009, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.1624583899974823, | |
| "rewards/margins": 0.07922794669866562, | |
| "rewards/rejected": -0.2416863739490509, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.995562370647553e-06, | |
| "logits/chosen": 0.13899263739585876, | |
| "logits/rejected": 0.2618511915206909, | |
| "logps/chosen": -438.03125, | |
| "logps/rejected": -467.33013916015625, | |
| "loss": 0.2569, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.10613300651311874, | |
| "rewards/margins": 0.091960109770298, | |
| "rewards/rejected": -0.19809311628341675, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.992357124836838e-06, | |
| "logits/chosen": 0.13999655842781067, | |
| "logits/rejected": 0.26916781067848206, | |
| "logps/chosen": -510.631103515625, | |
| "logps/rejected": -519.5391845703125, | |
| "loss": 0.2849, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.14092272520065308, | |
| "rewards/margins": 0.08064567297697067, | |
| "rewards/rejected": -0.22156842052936554, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882875734777044e-06, | |
| "logits/chosen": 0.15458917617797852, | |
| "logits/rejected": 0.21037283539772034, | |
| "logps/chosen": -470.0352478027344, | |
| "logps/rejected": -535.9059448242188, | |
| "loss": 0.236, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.14089417457580566, | |
| "rewards/margins": 0.12119980901479721, | |
| "rewards/rejected": -0.2620939612388611, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.983355127818882e-06, | |
| "logits/chosen": 0.17321957647800446, | |
| "logits/rejected": 0.3360803425312042, | |
| "logps/chosen": -578.07177734375, | |
| "logps/rejected": -678.3271484375, | |
| "loss": 0.25, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.23716862499713898, | |
| "rewards/margins": 0.15940351784229279, | |
| "rewards/rejected": -0.39657214283943176, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977561498345639e-06, | |
| "logits/chosen": 0.16144976019859314, | |
| "logits/rejected": 0.2567829489707947, | |
| "logps/chosen": -631.9638061523438, | |
| "logps/rejected": -725.3414916992188, | |
| "loss": 0.2686, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.2788200080394745, | |
| "rewards/margins": 0.14786966145038605, | |
| "rewards/rejected": -0.42668968439102173, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.970908694186624e-06, | |
| "logits/chosen": 0.16097748279571533, | |
| "logits/rejected": 0.22262544929981232, | |
| "logps/chosen": -553.2183837890625, | |
| "logps/rejected": -640.4689331054688, | |
| "loss": 0.2476, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.219456285238266, | |
| "rewards/margins": 0.1420574188232422, | |
| "rewards/rejected": -0.3615136742591858, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9633990224171305e-06, | |
| "logits/chosen": 0.14573153853416443, | |
| "logits/rejected": 0.23508696258068085, | |
| "logps/chosen": -539.7109375, | |
| "logps/rejected": -651.232421875, | |
| "loss": 0.2535, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.2373492270708084, | |
| "rewards/margins": 0.14303575456142426, | |
| "rewards/rejected": -0.38038498163223267, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.955035087259046e-06, | |
| "logits/chosen": 0.1654827892780304, | |
| "logits/rejected": 0.2800630033016205, | |
| "logps/chosen": -603.4498901367188, | |
| "logps/rejected": -686.7757568359375, | |
| "loss": 0.2537, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.2507944703102112, | |
| "rewards/margins": 0.16016040742397308, | |
| "rewards/rejected": -0.41095486283302307, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945819789177756e-06, | |
| "logits/chosen": 0.11868251860141754, | |
| "logits/rejected": 0.29606184363365173, | |
| "logps/chosen": -587.1047973632812, | |
| "logps/rejected": -650.8994750976562, | |
| "loss": 0.2541, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.23289060592651367, | |
| "rewards/margins": 0.13534070551395416, | |
| "rewards/rejected": -0.36823132634162903, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.935756323876306e-06, | |
| "logits/chosen": 0.11233693361282349, | |
| "logits/rejected": 0.2625276446342468, | |
| "logps/chosen": -580.7044677734375, | |
| "logps/rejected": -654.3244018554688, | |
| "loss": 0.2557, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.2328997552394867, | |
| "rewards/margins": 0.13923317193984985, | |
| "rewards/rejected": -0.37213292717933655, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924848181187199e-06, | |
| "logits/chosen": 0.14452582597732544, | |
| "logits/rejected": 0.2437099665403366, | |
| "logps/chosen": -606.4398803710938, | |
| "logps/rejected": -706.3465576171875, | |
| "loss": 0.2552, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.24929067492485046, | |
| "rewards/margins": 0.1445845663547516, | |
| "rewards/rejected": -0.39387524127960205, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.913099143862173e-06, | |
| "logits/chosen": 0.09590379893779755, | |
| "logits/rejected": 0.15467569231987, | |
| "logps/chosen": -597.2806396484375, | |
| "logps/rejected": -679.9869384765625, | |
| "loss": 0.2518, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.23923709988594055, | |
| "rewards/margins": 0.15891839563846588, | |
| "rewards/rejected": -0.3981555104255676, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900513286260416e-06, | |
| "logits/chosen": 0.11927614361047745, | |
| "logits/rejected": 0.1604013293981552, | |
| "logps/chosen": -640.5142822265625, | |
| "logps/rejected": -841.5134887695312, | |
| "loss": 0.2283, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.3347136974334717, | |
| "rewards/margins": 0.2090570628643036, | |
| "rewards/rejected": -0.5437707901000977, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.887094972935645e-06, | |
| "logits/chosen": 0.12204344570636749, | |
| "logits/rejected": 0.2500421106815338, | |
| "logps/chosen": -715.3956298828125, | |
| "logps/rejected": -785.560791015625, | |
| "loss": 0.2628, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.34101438522338867, | |
| "rewards/margins": 0.1506863236427307, | |
| "rewards/rejected": -0.4917007386684418, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.87284885712256e-06, | |
| "logits/chosen": 0.15500317513942719, | |
| "logits/rejected": 0.29695573449134827, | |
| "logps/chosen": -633.2789306640625, | |
| "logps/rejected": -729.1109619140625, | |
| "loss": 0.2473, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.2560732960700989, | |
| "rewards/margins": 0.17503255605697632, | |
| "rewards/rejected": -0.4311058521270752, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.857779879123181e-06, | |
| "logits/chosen": 0.0852101594209671, | |
| "logits/rejected": 0.1848641186952591, | |
| "logps/chosen": -561.0950927734375, | |
| "logps/rejected": -703.2201538085938, | |
| "loss": 0.2509, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.25699612498283386, | |
| "rewards/margins": 0.1455649882555008, | |
| "rewards/rejected": -0.40256109833717346, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.841893264593643e-06, | |
| "logits/chosen": 0.12055132538080215, | |
| "logits/rejected": 0.22479526698589325, | |
| "logps/chosen": -564.7427978515625, | |
| "logps/rejected": -672.8660888671875, | |
| "loss": 0.1755, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2189377099275589, | |
| "rewards/margins": 0.18233411014080048, | |
| "rewards/rejected": -0.4012718200683594, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.825194522732023e-06, | |
| "logits/chosen": 0.07654919475317001, | |
| "logits/rejected": 0.18854263424873352, | |
| "logps/chosen": -660.2512817382812, | |
| "logps/rejected": -773.058837890625, | |
| "loss": 0.2193, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.2836855351924896, | |
| "rewards/margins": 0.19571776688098907, | |
| "rewards/rejected": -0.4794033169746399, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807689444367853e-06, | |
| "logits/chosen": 0.09550791233778, | |
| "logits/rejected": 0.19389700889587402, | |
| "logps/chosen": -669.1429443359375, | |
| "logps/rejected": -838.7755126953125, | |
| "loss": 0.2165, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.31033799052238464, | |
| "rewards/margins": 0.2328120917081833, | |
| "rewards/rejected": -0.5431500673294067, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.78938409995396e-06, | |
| "logits/chosen": 0.001080992049537599, | |
| "logits/rejected": 0.14185592532157898, | |
| "logps/chosen": -594.972900390625, | |
| "logps/rejected": -736.0914306640625, | |
| "loss": 0.2722, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.2792796492576599, | |
| "rewards/margins": 0.17098276317119598, | |
| "rewards/rejected": -0.4502624571323395, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770284837461342e-06, | |
| "logits/chosen": 0.08192798495292664, | |
| "logits/rejected": 0.138724684715271, | |
| "logps/chosen": -556.79345703125, | |
| "logps/rejected": -625.5200805664062, | |
| "loss": 0.2453, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.21958652138710022, | |
| "rewards/margins": 0.12120404094457626, | |
| "rewards/rejected": -0.3407905697822571, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7503982801778015e-06, | |
| "logits/chosen": 0.08842509239912033, | |
| "logits/rejected": 0.12143252789974213, | |
| "logps/chosen": -575.1492919921875, | |
| "logps/rejected": -664.153564453125, | |
| "loss": 0.2486, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.24815413355827332, | |
| "rewards/margins": 0.13290396332740784, | |
| "rewards/rejected": -0.38105809688568115, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.729731324411104e-06, | |
| "logits/chosen": 0.0016645323485136032, | |
| "logits/rejected": 0.1351255476474762, | |
| "logps/chosen": -573.5245971679688, | |
| "logps/rejected": -703.4102783203125, | |
| "loss": 0.2377, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.24630823731422424, | |
| "rewards/margins": 0.16149748861789703, | |
| "rewards/rejected": -0.40780574083328247, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7082911370974645e-06, | |
| "logits/chosen": 0.04974130541086197, | |
| "logits/rejected": 0.10218667984008789, | |
| "logps/chosen": -575.5032958984375, | |
| "logps/rejected": -679.7342529296875, | |
| "loss": 0.2552, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.2367144078016281, | |
| "rewards/margins": 0.15034009516239166, | |
| "rewards/rejected": -0.38705453276634216, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68608515331618e-06, | |
| "logits/chosen": 0.03959808871150017, | |
| "logits/rejected": 0.16889765858650208, | |
| "logps/chosen": -601.1052856445312, | |
| "logps/rejected": -735.5942993164062, | |
| "loss": 0.2556, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.2658396363258362, | |
| "rewards/margins": 0.17807722091674805, | |
| "rewards/rejected": -0.44391685724258423, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.663121073711269e-06, | |
| "logits/chosen": 0.060259442776441574, | |
| "logits/rejected": 0.17572340369224548, | |
| "logps/chosen": -625.7338256835938, | |
| "logps/rejected": -786.9043579101562, | |
| "loss": 0.2364, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.29917049407958984, | |
| "rewards/margins": 0.1899183690547943, | |
| "rewards/rejected": -0.48908883333206177, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.63940686182103e-06, | |
| "logits/chosen": 0.05524193122982979, | |
| "logits/rejected": 0.1667497456073761, | |
| "logps/chosen": -640.2739868164062, | |
| "logps/rejected": -716.3453369140625, | |
| "loss": 0.2418, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.28584688901901245, | |
| "rewards/margins": 0.1426083743572235, | |
| "rewards/rejected": -0.42845526337623596, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.614950741316425e-06, | |
| "logits/chosen": 0.05645556375384331, | |
| "logits/rejected": 0.1303979605436325, | |
| "logps/chosen": -594.4942016601562, | |
| "logps/rejected": -682.8673095703125, | |
| "loss": 0.2491, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.25856301188468933, | |
| "rewards/margins": 0.15196993947029114, | |
| "rewards/rejected": -0.41053295135498047, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.589761193149254e-06, | |
| "logits/chosen": 0.07052268087863922, | |
| "logits/rejected": 0.13902485370635986, | |
| "logps/chosen": -585.3961181640625, | |
| "logps/rejected": -691.311767578125, | |
| "loss": 0.2537, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.2726908326148987, | |
| "rewards/margins": 0.1488717794418335, | |
| "rewards/rejected": -0.4215625822544098, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.563846952611112e-06, | |
| "logits/chosen": 0.05198805406689644, | |
| "logits/rejected": 0.10118089616298676, | |
| "logps/chosen": -598.3322143554688, | |
| "logps/rejected": -739.3489990234375, | |
| "loss": 0.2355, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.25981515645980835, | |
| "rewards/margins": 0.2083590030670166, | |
| "rewards/rejected": -0.46817412972450256, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.537217006304141e-06, | |
| "logits/chosen": 0.09554968029260635, | |
| "logits/rejected": 0.22383460402488708, | |
| "logps/chosen": -621.1278076171875, | |
| "logps/rejected": -703.5054931640625, | |
| "loss": 0.2346, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.2537866532802582, | |
| "rewards/margins": 0.1655985414981842, | |
| "rewards/rejected": -0.4193851947784424, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.50988058902464e-06, | |
| "logits/chosen": 0.07675488293170929, | |
| "logits/rejected": 0.22268608212471008, | |
| "logps/chosen": -649.4010009765625, | |
| "logps/rejected": -696.96337890625, | |
| "loss": 0.2337, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.24650602042675018, | |
| "rewards/margins": 0.1643296480178833, | |
| "rewards/rejected": -0.4108356535434723, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481847180560593e-06, | |
| "logits/chosen": -0.009524239227175713, | |
| "logits/rejected": 0.13664120435714722, | |
| "logps/chosen": -608.6438598632812, | |
| "logps/rejected": -744.18701171875, | |
| "loss": 0.2511, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.2557665705680847, | |
| "rewards/margins": 0.1905471682548523, | |
| "rewards/rejected": -0.4463137090206146, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.453126502404253e-06, | |
| "logits/chosen": 0.04823983460664749, | |
| "logits/rejected": 0.21495167911052704, | |
| "logps/chosen": -668.1390991210938, | |
| "logps/rejected": -764.8587036132812, | |
| "loss": 0.2063, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.2886661887168884, | |
| "rewards/margins": 0.20503409206867218, | |
| "rewards/rejected": -0.4937002658843994, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.423728514380892e-06, | |
| "logits/chosen": 0.14659680426120758, | |
| "logits/rejected": 0.19975139200687408, | |
| "logps/chosen": -691.0910034179688, | |
| "logps/rejected": -818.7584228515625, | |
| "loss": 0.216, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.33451852202415466, | |
| "rewards/margins": 0.21299397945404053, | |
| "rewards/rejected": -0.5475125312805176, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.393663411194918e-06, | |
| "logits/chosen": 0.10968673229217529, | |
| "logits/rejected": 0.2234468162059784, | |
| "logps/chosen": -732.4443359375, | |
| "logps/rejected": -899.6062622070312, | |
| "loss": 0.2183, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.33378008008003235, | |
| "rewards/margins": 0.2530228793621063, | |
| "rewards/rejected": -0.5868029594421387, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.362941618894523e-06, | |
| "logits/chosen": 0.10406269133090973, | |
| "logits/rejected": 0.15065929293632507, | |
| "logps/chosen": -659.9013671875, | |
| "logps/rejected": -766.6561279296875, | |
| "loss": 0.2346, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.29572516679763794, | |
| "rewards/margins": 0.1765248328447342, | |
| "rewards/rejected": -0.47224998474121094, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.331573791256116e-06, | |
| "logits/chosen": 0.0664634108543396, | |
| "logits/rejected": 0.16555871069431305, | |
| "logps/chosen": -600.4996948242188, | |
| "logps/rejected": -713.78466796875, | |
| "loss": 0.2461, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.2576759457588196, | |
| "rewards/margins": 0.1549091637134552, | |
| "rewards/rejected": -0.4125851094722748, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.299570806089786e-06, | |
| "logits/chosen": 0.16024425625801086, | |
| "logits/rejected": 0.22183863818645477, | |
| "logps/chosen": -619.4130859375, | |
| "logps/rejected": -696.956787109375, | |
| "loss": 0.2457, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.24661366641521454, | |
| "rewards/margins": 0.1785852015018463, | |
| "rewards/rejected": -0.42519888281822205, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.266943761467057e-06, | |
| "logits/chosen": 0.07459579408168793, | |
| "logits/rejected": 0.20276884734630585, | |
| "logps/chosen": -657.4212646484375, | |
| "logps/rejected": -731.3056640625, | |
| "loss": 0.231, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.2757970690727234, | |
| "rewards/margins": 0.1686132848262787, | |
| "rewards/rejected": -0.44441038370132446, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.233703971872287e-06, | |
| "logits/chosen": 0.056164585053920746, | |
| "logits/rejected": 0.22885584831237793, | |
| "logps/chosen": -652.3689575195312, | |
| "logps/rejected": -749.9366455078125, | |
| "loss": 0.227, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.29479923844337463, | |
| "rewards/margins": 0.16815118491649628, | |
| "rewards/rejected": -0.4629504680633545, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.1998629642789925e-06, | |
| "logits/chosen": 0.10070328414440155, | |
| "logits/rejected": 0.16475458443164825, | |
| "logps/chosen": -708.42529296875, | |
| "logps/rejected": -811.713134765625, | |
| "loss": 0.2415, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.33783578872680664, | |
| "rewards/margins": 0.18223688006401062, | |
| "rewards/rejected": -0.5200726389884949, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.165432474152505e-06, | |
| "logits/chosen": 0.10311365127563477, | |
| "logits/rejected": 0.15643465518951416, | |
| "logps/chosen": -583.4866333007812, | |
| "logps/rejected": -729.7852783203125, | |
| "loss": 0.2016, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2622414231300354, | |
| "rewards/margins": 0.19530150294303894, | |
| "rewards/rejected": -0.45754295587539673, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.130424441380308e-06, | |
| "logits/chosen": 0.10025729984045029, | |
| "logits/rejected": 0.2146105021238327, | |
| "logps/chosen": -681.4263916015625, | |
| "logps/rejected": -823.4031372070312, | |
| "loss": 0.2163, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3132783770561218, | |
| "rewards/margins": 0.21751663088798523, | |
| "rewards/rejected": -0.5307949781417847, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.09485100613151e-06, | |
| "logits/chosen": 0.10832244157791138, | |
| "logits/rejected": 0.20945067703723907, | |
| "logps/chosen": -660.5101318359375, | |
| "logps/rejected": -768.7022705078125, | |
| "loss": 0.2215, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.30939123034477234, | |
| "rewards/margins": 0.18488983809947968, | |
| "rewards/rejected": -0.49428099393844604, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.058724504646834e-06, | |
| "logits/chosen": 0.08686210215091705, | |
| "logits/rejected": 0.2045850306749344, | |
| "logps/chosen": -723.7673950195312, | |
| "logps/rejected": -883.3922119140625, | |
| "loss": 0.2512, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.38144949078559875, | |
| "rewards/margins": 0.195582315325737, | |
| "rewards/rejected": -0.5770317912101746, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.022057464960632e-06, | |
| "logits/chosen": 0.09274314343929291, | |
| "logits/rejected": 0.1920190155506134, | |
| "logps/chosen": -618.1563720703125, | |
| "logps/rejected": -740.8644409179688, | |
| "loss": 0.2629, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.3001014292240143, | |
| "rewards/margins": 0.1581810563802719, | |
| "rewards/rejected": -0.458282470703125, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.984862602556383e-06, | |
| "logits/chosen": 0.021902620792388916, | |
| "logits/rejected": 0.11925847828388214, | |
| "logps/chosen": -595.6156005859375, | |
| "logps/rejected": -711.0296630859375, | |
| "loss": 0.2626, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.2705914080142975, | |
| "rewards/margins": 0.16306385397911072, | |
| "rewards/rejected": -0.4336552619934082, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.947152815957187e-06, | |
| "logits/chosen": 0.05186104774475098, | |
| "logits/rejected": 0.20855280756950378, | |
| "logps/chosen": -584.5855102539062, | |
| "logps/rejected": -713.7994384765625, | |
| "loss": 0.1958, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.25383633375167847, | |
| "rewards/margins": 0.1767672598361969, | |
| "rewards/rejected": -0.43060359358787537, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.908941182252785e-06, | |
| "logits/chosen": 0.022304801270365715, | |
| "logits/rejected": 0.14363619685173035, | |
| "logps/chosen": -643.5025634765625, | |
| "logps/rejected": -748.3091430664062, | |
| "loss": 0.2237, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.29282766580581665, | |
| "rewards/margins": 0.17381563782691956, | |
| "rewards/rejected": -0.4666433334350586, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.8702409525646535e-06, | |
| "logits/chosen": 0.09528463333845139, | |
| "logits/rejected": 0.18656165897846222, | |
| "logps/chosen": -692.3035888671875, | |
| "logps/rejected": -819.2098388671875, | |
| "loss": 0.2097, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.31982243061065674, | |
| "rewards/margins": 0.20729240775108337, | |
| "rewards/rejected": -0.5271148085594177, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.8310655474507495e-06, | |
| "logits/chosen": 0.04973004013299942, | |
| "logits/rejected": 0.19504304230213165, | |
| "logps/chosen": -688.0816650390625, | |
| "logps/rejected": -852.9588012695312, | |
| "loss": 0.2432, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.36149436235427856, | |
| "rewards/margins": 0.1907968968153, | |
| "rewards/rejected": -0.5522913336753845, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7914285522515002e-06, | |
| "logits/chosen": 0.09976281225681305, | |
| "logits/rejected": 0.19117242097854614, | |
| "logps/chosen": -680.7174072265625, | |
| "logps/rejected": -797.1549072265625, | |
| "loss": 0.2341, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.3415672183036804, | |
| "rewards/margins": 0.1845674216747284, | |
| "rewards/rejected": -0.5261346101760864, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.751343712378639e-06, | |
| "logits/chosen": 0.07142799347639084, | |
| "logits/rejected": 0.22169947624206543, | |
| "logps/chosen": -626.5697021484375, | |
| "logps/rejected": -787.6502075195312, | |
| "loss": 0.2383, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2981518507003784, | |
| "rewards/margins": 0.206342414021492, | |
| "rewards/rejected": -0.504494309425354, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.710824928548546e-06, | |
| "logits/chosen": 0.1079743355512619, | |
| "logits/rejected": 0.13973113894462585, | |
| "logps/chosen": -665.4769287109375, | |
| "logps/rejected": -744.9660034179688, | |
| "loss": 0.2612, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.3199656009674072, | |
| "rewards/margins": 0.16452810168266296, | |
| "rewards/rejected": -0.4844936728477478, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6698862519617225e-06, | |
| "logits/chosen": 0.075187087059021, | |
| "logits/rejected": 0.15009218454360962, | |
| "logps/chosen": -644.90478515625, | |
| "logps/rejected": -790.0805053710938, | |
| "loss": 0.2429, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.30303993821144104, | |
| "rewards/margins": 0.19713596999645233, | |
| "rewards/rejected": -0.500175952911377, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6285418794300793e-06, | |
| "logits/chosen": 0.06037820130586624, | |
| "logits/rejected": 0.12935443222522736, | |
| "logps/chosen": -659.0757446289062, | |
| "logps/rejected": -834.96533203125, | |
| "loss": 0.207, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.31061092019081116, | |
| "rewards/margins": 0.22491443157196045, | |
| "rewards/rejected": -0.535525381565094, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.5868061484537365e-06, | |
| "logits/chosen": 0.062171898782253265, | |
| "logits/rejected": 0.20288972556591034, | |
| "logps/chosen": -650.5526733398438, | |
| "logps/rejected": -812.0040893554688, | |
| "loss": 0.2295, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3222615122795105, | |
| "rewards/margins": 0.2239852249622345, | |
| "rewards/rejected": -0.5462467670440674, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5446935322490285e-06, | |
| "logits/chosen": 0.07936866581439972, | |
| "logits/rejected": 0.23025266826152802, | |
| "logps/chosen": -730.1929321289062, | |
| "logps/rejected": -875.7835693359375, | |
| "loss": 0.1995, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.3494511842727661, | |
| "rewards/margins": 0.2297430783510208, | |
| "rewards/rejected": -0.5791942477226257, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.502218634729447e-06, | |
| "logits/chosen": 0.11655671894550323, | |
| "logits/rejected": 0.1666901409626007, | |
| "logps/chosen": -654.1530151367188, | |
| "logps/rejected": -815.29296875, | |
| "loss": 0.2193, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.30700844526290894, | |
| "rewards/margins": 0.21489982306957245, | |
| "rewards/rejected": -0.5219083428382874, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.459396185441265e-06, | |
| "logits/chosen": 0.11936108767986298, | |
| "logits/rejected": 0.18859058618545532, | |
| "logps/chosen": -630.1988525390625, | |
| "logps/rejected": -762.0612182617188, | |
| "loss": 0.2079, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.25710731744766235, | |
| "rewards/margins": 0.2074425220489502, | |
| "rewards/rejected": -0.46454983949661255, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4162410344555834e-06, | |
| "logits/chosen": 0.05629768967628479, | |
| "logits/rejected": 0.1666196882724762, | |
| "logps/chosen": -649.73046875, | |
| "logps/rejected": -818.6950073242188, | |
| "loss": 0.1792, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.2980211675167084, | |
| "rewards/margins": 0.23819315433502197, | |
| "rewards/rejected": -0.5362143516540527, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3727681472185937e-06, | |
| "logits/chosen": 0.04574333503842354, | |
| "logits/rejected": 0.17197957634925842, | |
| "logps/chosen": -702.0049438476562, | |
| "logps/rejected": -867.20947265625, | |
| "loss": 0.2038, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.35734453797340393, | |
| "rewards/margins": 0.22116133570671082, | |
| "rewards/rejected": -0.5785058736801147, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3289925993618217e-06, | |
| "logits/chosen": 0.055036939680576324, | |
| "logits/rejected": 0.18666990101337433, | |
| "logps/chosen": -772.3389282226562, | |
| "logps/rejected": -934.3049926757812, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.3967970013618469, | |
| "rewards/margins": 0.2379632443189621, | |
| "rewards/rejected": -0.6347602605819702, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2849295714741643e-06, | |
| "logits/chosen": 0.06958106905221939, | |
| "logits/rejected": 0.1442635953426361, | |
| "logps/chosen": -669.5227661132812, | |
| "logps/rejected": -853.2138671875, | |
| "loss": 0.207, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3451857566833496, | |
| "rewards/margins": 0.2383837252855301, | |
| "rewards/rejected": -0.5835694074630737, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2405943438375287e-06, | |
| "logits/chosen": 0.0513598695397377, | |
| "logits/rejected": 0.13928762078285217, | |
| "logps/chosen": -653.1226806640625, | |
| "logps/rejected": -795.4114379882812, | |
| "loss": 0.2438, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.327537477016449, | |
| "rewards/margins": 0.18883965909481049, | |
| "rewards/rejected": -0.5163770914077759, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1960022911279036e-06, | |
| "logits/chosen": 0.07376402616500854, | |
| "logits/rejected": 0.17226745188236237, | |
| "logps/chosen": -704.4319458007812, | |
| "logps/rejected": -898.2445068359375, | |
| "loss": 0.1766, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.33583196997642517, | |
| "rewards/margins": 0.25600650906562805, | |
| "rewards/rejected": -0.5918384790420532, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1511688770836844e-06, | |
| "logits/chosen": 0.04906494915485382, | |
| "logits/rejected": 0.1545862853527069, | |
| "logps/chosen": -670.8369140625, | |
| "logps/rejected": -803.107421875, | |
| "loss": 0.2318, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3352956175804138, | |
| "rewards/margins": 0.2003110647201538, | |
| "rewards/rejected": -0.5356066823005676, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.1061096491431307e-06, | |
| "logits/chosen": 0.12640194594860077, | |
| "logits/rejected": 0.219802588224411, | |
| "logps/chosen": -742.6751098632812, | |
| "logps/rejected": -895.0465698242188, | |
| "loss": 0.1921, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.3457973599433899, | |
| "rewards/margins": 0.24558237195014954, | |
| "rewards/rejected": -0.591379702091217, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0608402330527796e-06, | |
| "logits/chosen": -0.01623944379389286, | |
| "logits/rejected": 0.16946300864219666, | |
| "logps/chosen": -649.9053955078125, | |
| "logps/rejected": -832.1677856445312, | |
| "loss": 0.1958, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.3138830065727234, | |
| "rewards/margins": 0.23707744479179382, | |
| "rewards/rejected": -0.5509604215621948, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0153763274487176e-06, | |
| "logits/chosen": 0.05111056566238403, | |
| "logits/rejected": 0.14114698767662048, | |
| "logps/chosen": -662.3326416015625, | |
| "logps/rejected": -784.4324951171875, | |
| "loss": 0.2388, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.32007449865341187, | |
| "rewards/margins": 0.1715754121541977, | |
| "rewards/rejected": -0.49164995551109314, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9697336984125683e-06, | |
| "logits/chosen": -0.006915999110788107, | |
| "logits/rejected": 0.11657528579235077, | |
| "logps/chosen": -666.0967407226562, | |
| "logps/rejected": -806.9496459960938, | |
| "loss": 0.2169, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.30281534790992737, | |
| "rewards/margins": 0.22429080307483673, | |
| "rewards/rejected": -0.5271061062812805, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.923928174004094e-06, | |
| "logits/chosen": 0.056006837636232376, | |
| "logits/rejected": 0.17199097573757172, | |
| "logps/chosen": -652.8565673828125, | |
| "logps/rejected": -816.8966674804688, | |
| "loss": 0.2207, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.27928978204727173, | |
| "rewards/margins": 0.23044073581695557, | |
| "rewards/rejected": -0.5097305178642273, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8779756387723036e-06, | |
| "logits/chosen": 0.06072944402694702, | |
| "logits/rejected": 0.13912425935268402, | |
| "logps/chosen": -622.7828369140625, | |
| "logps/rejected": -809.6715698242188, | |
| "loss": 0.2059, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.2813548147678375, | |
| "rewards/margins": 0.23472395539283752, | |
| "rewards/rejected": -0.5160787105560303, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.831892028246968e-06, | |
| "logits/chosen": -0.0010059855412691832, | |
| "logits/rejected": 0.11881496012210846, | |
| "logps/chosen": -682.4347534179688, | |
| "logps/rejected": -841.9749755859375, | |
| "loss": 0.2177, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3291272521018982, | |
| "rewards/margins": 0.23086945712566376, | |
| "rewards/rejected": -0.5599966645240784, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7856933234124617e-06, | |
| "logits/chosen": 0.05394923686981201, | |
| "logits/rejected": 0.16719438135623932, | |
| "logps/chosen": -690.3303833007812, | |
| "logps/rejected": -856.0970458984375, | |
| "loss": 0.2469, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.3471989035606384, | |
| "rewards/margins": 0.21049991250038147, | |
| "rewards/rejected": -0.5576988458633423, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7393955451658387e-06, | |
| "logits/chosen": 0.043146662414073944, | |
| "logits/rejected": 0.08542972058057785, | |
| "logps/chosen": -734.964599609375, | |
| "logps/rejected": -885.3410034179688, | |
| "loss": 0.2057, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3733413815498352, | |
| "rewards/margins": 0.21346192061901093, | |
| "rewards/rejected": -0.5868033170700073, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6930147487610667e-06, | |
| "logits/chosen": 0.01986471749842167, | |
| "logits/rejected": 0.1354392021894455, | |
| "logps/chosen": -722.3784790039062, | |
| "logps/rejected": -825.0411376953125, | |
| "loss": 0.2134, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.33837828040122986, | |
| "rewards/margins": 0.21411871910095215, | |
| "rewards/rejected": -0.5524970293045044, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6465670182413487e-06, | |
| "logits/chosen": 0.04280179738998413, | |
| "logits/rejected": 0.16137579083442688, | |
| "logps/chosen": -761.4157104492188, | |
| "logps/rejected": -863.3331909179688, | |
| "loss": 0.2193, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3615965247154236, | |
| "rewards/margins": 0.20891360938549042, | |
| "rewards/rejected": -0.5705101490020752, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.6000684608614594e-06, | |
| "logits/chosen": 0.028211206197738647, | |
| "logits/rejected": 0.09348297864198685, | |
| "logps/chosen": -724.6305541992188, | |
| "logps/rejected": -880.3488159179688, | |
| "loss": 0.2079, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.36276954412460327, | |
| "rewards/margins": 0.22737035155296326, | |
| "rewards/rejected": -0.5901398658752441, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.5535352015020338e-06, | |
| "logits/chosen": 0.05322765186429024, | |
| "logits/rejected": 0.13077041506767273, | |
| "logps/chosen": -760.7427978515625, | |
| "logps/rejected": -912.9437255859375, | |
| "loss": 0.2429, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.41152581572532654, | |
| "rewards/margins": 0.21826231479644775, | |
| "rewards/rejected": -0.6297882199287415, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.506983377077741e-06, | |
| "logits/chosen": -0.0053452602587640285, | |
| "logits/rejected": 0.112821564078331, | |
| "logps/chosen": -674.3397827148438, | |
| "logps/rejected": -873.40087890625, | |
| "loss": 0.1812, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.34911829233169556, | |
| "rewards/margins": 0.24309155344963074, | |
| "rewards/rejected": -0.5922098159790039, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.460429130941289e-06, | |
| "logits/chosen": 0.0023634755052626133, | |
| "logits/rejected": 0.12815037369728088, | |
| "logps/chosen": -758.9469604492188, | |
| "logps/rejected": -931.2481689453125, | |
| "loss": 0.2024, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.37650245428085327, | |
| "rewards/margins": 0.25763195753097534, | |
| "rewards/rejected": -0.6341344118118286, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.413888607285192e-06, | |
| "logits/chosen": 0.035070545971393585, | |
| "logits/rejected": 0.12803152203559875, | |
| "logps/chosen": -736.931396484375, | |
| "logps/rejected": -923.0338745117188, | |
| "loss": 0.184, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.36943623423576355, | |
| "rewards/margins": 0.2507568597793579, | |
| "rewards/rejected": -0.6201931238174438, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.367377945543249e-06, | |
| "logits/chosen": 5.705729199689813e-05, | |
| "logits/rejected": 0.1263124793767929, | |
| "logps/chosen": -706.70068359375, | |
| "logps/rejected": -814.0146484375, | |
| "loss": 0.2331, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.3658052086830139, | |
| "rewards/margins": 0.1808270514011383, | |
| "rewards/rejected": -0.5466322898864746, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.320913274793676e-06, | |
| "logits/chosen": 0.015436547808349133, | |
| "logits/rejected": 0.10843756049871445, | |
| "logps/chosen": -703.6624755859375, | |
| "logps/rejected": -858.7509765625, | |
| "loss": 0.1987, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.31511586904525757, | |
| "rewards/margins": 0.2355271279811859, | |
| "rewards/rejected": -0.5506430268287659, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.27451070816582e-06, | |
| "logits/chosen": 0.012071605771780014, | |
| "logits/rejected": 0.11224579811096191, | |
| "logps/chosen": -666.1785888671875, | |
| "logps/rejected": -801.8255004882812, | |
| "loss": 0.2259, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.34429460763931274, | |
| "rewards/margins": 0.19534102082252502, | |
| "rewards/rejected": -0.5396356582641602, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.228186337252414e-06, | |
| "logits/chosen": -0.010557098314166069, | |
| "logits/rejected": 0.06913888454437256, | |
| "logps/chosen": -734.2570190429688, | |
| "logps/rejected": -892.15625, | |
| "loss": 0.2044, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.36991119384765625, | |
| "rewards/margins": 0.2224087417125702, | |
| "rewards/rejected": -0.5923199653625488, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1819562265292946e-06, | |
| "logits/chosen": -0.006013315636664629, | |
| "logits/rejected": 0.08018209785223007, | |
| "logps/chosen": -695.9881591796875, | |
| "logps/rejected": -833.2312622070312, | |
| "loss": 0.2152, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.36053889989852905, | |
| "rewards/margins": 0.20316271483898163, | |
| "rewards/rejected": -0.5637016296386719, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1358364077845236e-06, | |
| "logits/chosen": -0.008084287866950035, | |
| "logits/rejected": 0.0774918869137764, | |
| "logps/chosen": -716.2960815429688, | |
| "logps/rejected": -869.6041259765625, | |
| "loss": 0.2309, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3821774125099182, | |
| "rewards/margins": 0.20834413170814514, | |
| "rewards/rejected": -0.5905215740203857, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.089842874558849e-06, | |
| "logits/chosen": 0.007766352035105228, | |
| "logits/rejected": 0.030095476657152176, | |
| "logps/chosen": -695.3117065429688, | |
| "logps/rejected": -871.33837890625, | |
| "loss": 0.1987, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.3631244897842407, | |
| "rewards/margins": 0.22518320381641388, | |
| "rewards/rejected": -0.5883077383041382, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0439915765994242e-06, | |
| "logits/chosen": -0.024640878662467003, | |
| "logits/rejected": 0.09226598590612411, | |
| "logps/chosen": -703.4818115234375, | |
| "logps/rejected": -877.7662353515625, | |
| "loss": 0.2056, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.3533559739589691, | |
| "rewards/margins": 0.22752220928668976, | |
| "rewards/rejected": -0.5808781385421753, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9982984143287186e-06, | |
| "logits/chosen": 0.009741904214024544, | |
| "logits/rejected": 0.17097178101539612, | |
| "logps/chosen": -810.0487670898438, | |
| "logps/rejected": -879.8194580078125, | |
| "loss": 0.1982, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.3632999062538147, | |
| "rewards/margins": 0.2137196958065033, | |
| "rewards/rejected": -0.5770195722579956, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.95277923333053e-06, | |
| "logits/chosen": 0.012245150282979012, | |
| "logits/rejected": 0.12731657922267914, | |
| "logps/chosen": -676.3506469726562, | |
| "logps/rejected": -802.7940063476562, | |
| "loss": 0.2275, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.33614057302474976, | |
| "rewards/margins": 0.19641144573688507, | |
| "rewards/rejected": -0.5325520634651184, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9074498188550156e-06, | |
| "logits/chosen": 0.02413441613316536, | |
| "logits/rejected": 0.19431506097316742, | |
| "logps/chosen": -700.3193359375, | |
| "logps/rejected": -871.38232421875, | |
| "loss": 0.1795, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.32141950726509094, | |
| "rewards/margins": 0.23818500339984894, | |
| "rewards/rejected": -0.5596045255661011, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.862325890344643e-06, | |
| "logits/chosen": 0.041377656161785126, | |
| "logits/rejected": 0.15879663825035095, | |
| "logps/chosen": -665.6008911132812, | |
| "logps/rejected": -830.9329223632812, | |
| "loss": 0.2267, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.342972993850708, | |
| "rewards/margins": 0.20079226791858673, | |
| "rewards/rejected": -0.5437651872634888, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.817423095982972e-06, | |
| "logits/chosen": 0.022486714646220207, | |
| "logits/rejected": 0.146213561296463, | |
| "logps/chosen": -672.7999877929688, | |
| "logps/rejected": -847.8994140625, | |
| "loss": 0.1819, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.3093092739582062, | |
| "rewards/margins": 0.2545778453350067, | |
| "rewards/rejected": -0.5638871192932129, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.7727570072681293e-06, | |
| "logits/chosen": 0.04687700420618057, | |
| "logits/rejected": 0.12050887197256088, | |
| "logps/chosen": -672.9434814453125, | |
| "logps/rejected": -819.170654296875, | |
| "loss": 0.2352, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.3423709273338318, | |
| "rewards/margins": 0.20281191170215607, | |
| "rewards/rejected": -0.5451828241348267, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7283431136128961e-06, | |
| "logits/chosen": 0.045673031359910965, | |
| "logits/rejected": 0.11997739970684052, | |
| "logps/chosen": -611.8385009765625, | |
| "logps/rejected": -820.4505615234375, | |
| "loss": 0.2063, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.29818734526634216, | |
| "rewards/margins": 0.22259333729743958, | |
| "rewards/rejected": -0.5207806825637817, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6841968169732478e-06, | |
| "logits/chosen": 0.06439246237277985, | |
| "logits/rejected": 0.10217195749282837, | |
| "logps/chosen": -665.1314697265625, | |
| "logps/rejected": -801.7761840820312, | |
| "loss": 0.2283, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3135600686073303, | |
| "rewards/margins": 0.20070858299732208, | |
| "rewards/rejected": -0.514268696308136, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6403334265072284e-06, | |
| "logits/chosen": 0.03967912867665291, | |
| "logits/rejected": 0.1717870980501175, | |
| "logps/chosen": -683.311767578125, | |
| "logps/rejected": -844.0208129882812, | |
| "loss": 0.2371, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3242947459220886, | |
| "rewards/margins": 0.2034534513950348, | |
| "rewards/rejected": -0.5277482271194458, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5967681532660066e-06, | |
| "logits/chosen": 0.013939693570137024, | |
| "logits/rejected": 0.12744280695915222, | |
| "logps/chosen": -689.4755859375, | |
| "logps/rejected": -812.0987548828125, | |
| "loss": 0.2429, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3416514992713928, | |
| "rewards/margins": 0.1776062697172165, | |
| "rewards/rejected": -0.5192577838897705, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5535161049189463e-06, | |
| "logits/chosen": 0.06095554679632187, | |
| "logits/rejected": 0.14684705436229706, | |
| "logps/chosen": -654.1356811523438, | |
| "logps/rejected": -810.3081665039062, | |
| "loss": 0.2229, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.3114492893218994, | |
| "rewards/margins": 0.1974024623632431, | |
| "rewards/rejected": -0.5088517665863037, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5105922805145356e-06, | |
| "logits/chosen": 0.05724464729428291, | |
| "logits/rejected": 0.14580979943275452, | |
| "logps/chosen": -657.7913818359375, | |
| "logps/rejected": -799.1402587890625, | |
| "loss": 0.2055, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.308960497379303, | |
| "rewards/margins": 0.21290500462055206, | |
| "rewards/rejected": -0.5218654870986938, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4680115652789823e-06, | |
| "logits/chosen": 0.04883120581507683, | |
| "logits/rejected": 0.11742043495178223, | |
| "logps/chosen": -645.9979248046875, | |
| "logps/rejected": -819.2106323242188, | |
| "loss": 0.1943, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.3161103129386902, | |
| "rewards/margins": 0.21235115826129913, | |
| "rewards/rejected": -0.5284615159034729, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.4257887254542767e-06, | |
| "logits/chosen": 0.08411203324794769, | |
| "logits/rejected": 0.14393514394760132, | |
| "logps/chosen": -695.6365966796875, | |
| "logps/rejected": -878.0914306640625, | |
| "loss": 0.2218, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.34891873598098755, | |
| "rewards/margins": 0.24423813819885254, | |
| "rewards/rejected": -0.5931568741798401, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.3839384031775227e-06, | |
| "logits/chosen": 0.045291513204574585, | |
| "logits/rejected": 0.13728824257850647, | |
| "logps/chosen": -707.1044311523438, | |
| "logps/rejected": -854.4798583984375, | |
| "loss": 0.2101, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3448121249675751, | |
| "rewards/margins": 0.23182924091815948, | |
| "rewards/rejected": -0.5766413807868958, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.342475111403298e-06, | |
| "logits/chosen": 0.029178116470575333, | |
| "logits/rejected": 0.17620958387851715, | |
| "logps/chosen": -674.8609008789062, | |
| "logps/rejected": -858.0894775390625, | |
| "loss": 0.1804, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.3337288498878479, | |
| "rewards/margins": 0.24131789803504944, | |
| "rewards/rejected": -0.5750466585159302, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3014132288708209e-06, | |
| "logits/chosen": 0.038364656269550323, | |
| "logits/rejected": 0.15578509867191315, | |
| "logps/chosen": -692.9742431640625, | |
| "logps/rejected": -857.2974853515625, | |
| "loss": 0.2094, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3217311501502991, | |
| "rewards/margins": 0.2241102159023285, | |
| "rewards/rejected": -0.54584139585495, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2607669951176549e-06, | |
| "logits/chosen": 0.01667640171945095, | |
| "logits/rejected": 0.10969080030918121, | |
| "logps/chosen": -627.9285888671875, | |
| "logps/rejected": -815.8270263671875, | |
| "loss": 0.2083, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.3187568783760071, | |
| "rewards/margins": 0.22442683577537537, | |
| "rewards/rejected": -0.5431837439537048, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2205505055416891e-06, | |
| "logits/chosen": 0.04231597110629082, | |
| "logits/rejected": 0.14465740323066711, | |
| "logps/chosen": -721.646484375, | |
| "logps/rejected": -825.8064575195312, | |
| "loss": 0.2264, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3264230787754059, | |
| "rewards/margins": 0.21109537780284882, | |
| "rewards/rejected": -0.5375184416770935, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1807777065131002e-06, | |
| "logits/chosen": 0.0638522133231163, | |
| "logits/rejected": 0.14387431740760803, | |
| "logps/chosen": -669.5394287109375, | |
| "logps/rejected": -838.8292236328125, | |
| "loss": 0.1997, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.33911409974098206, | |
| "rewards/margins": 0.22880196571350098, | |
| "rewards/rejected": -0.5679160356521606, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1414623905380012e-06, | |
| "logits/chosen": 0.014098724350333214, | |
| "logits/rejected": 0.13527414202690125, | |
| "logps/chosen": -669.8994750976562, | |
| "logps/rejected": -806.6534423828125, | |
| "loss": 0.2604, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.34625038504600525, | |
| "rewards/margins": 0.17370744049549103, | |
| "rewards/rejected": -0.5199578404426575, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.1026181914754388e-06, | |
| "logits/chosen": 0.027355173602700233, | |
| "logits/rejected": 0.12094994634389877, | |
| "logps/chosen": -686.2830200195312, | |
| "logps/rejected": -793.8912353515625, | |
| "loss": 0.2006, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3198310434818268, | |
| "rewards/margins": 0.2038339078426361, | |
| "rewards/rejected": -0.5236650109291077, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0642585798094136e-06, | |
| "logits/chosen": 0.011197770945727825, | |
| "logits/rejected": 0.1233413964509964, | |
| "logps/chosen": -663.2740478515625, | |
| "logps/rejected": -827.7921142578125, | |
| "loss": 0.2023, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.31737273931503296, | |
| "rewards/margins": 0.2423689365386963, | |
| "rewards/rejected": -0.5597416162490845, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0263968579775522e-06, | |
| "logits/chosen": 0.012365484610199928, | |
| "logits/rejected": 0.14399036765098572, | |
| "logps/chosen": -723.01611328125, | |
| "logps/rejected": -888.5731201171875, | |
| "loss": 0.2107, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.3377194106578827, | |
| "rewards/margins": 0.25036248564720154, | |
| "rewards/rejected": -0.588081955909729, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.89046155758058e-07, | |
| "logits/chosen": 0.03630852699279785, | |
| "logits/rejected": 0.1697433888912201, | |
| "logps/chosen": -692.0089111328125, | |
| "logps/rejected": -841.7423706054688, | |
| "loss": 0.2225, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.33607909083366394, | |
| "rewards/margins": 0.2240186482667923, | |
| "rewards/rejected": -0.5600977540016174, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.52219425716534e-07, | |
| "logits/chosen": 0.06855352222919464, | |
| "logits/rejected": 0.16431955993175507, | |
| "logps/chosen": -735.5272216796875, | |
| "logps/rejected": -859.2053833007812, | |
| "loss": 0.2056, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3497410714626312, | |
| "rewards/margins": 0.21472422778606415, | |
| "rewards/rejected": -0.564465343952179, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.15929438714262e-07, | |
| "logits/chosen": -0.0006916738930158317, | |
| "logits/rejected": 0.11941851675510406, | |
| "logps/chosen": -708.4884643554688, | |
| "logps/rejected": -896.8206787109375, | |
| "loss": 0.1853, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.35898059606552124, | |
| "rewards/margins": 0.240576833486557, | |
| "rewards/rejected": -0.5995574593544006, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.801887794794911e-07, | |
| "logits/chosen": 0.039211105555295944, | |
| "logits/rejected": 0.11498727649450302, | |
| "logps/chosen": -711.9271850585938, | |
| "logps/rejected": -853.4754638671875, | |
| "loss": 0.2018, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.3567042946815491, | |
| "rewards/margins": 0.21128018200397491, | |
| "rewards/rejected": -0.567984402179718, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.450098422432787e-07, | |
| "logits/chosen": 0.02443738654255867, | |
| "logits/rejected": 0.05825776979327202, | |
| "logps/chosen": -668.7024536132812, | |
| "logps/rejected": -816.4102172851562, | |
| "loss": 0.2303, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.332243412733078, | |
| "rewards/margins": 0.21151785552501678, | |
| "rewards/rejected": -0.5437612533569336, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.104048264413858e-07, | |
| "logits/chosen": 0.04351818934082985, | |
| "logits/rejected": 0.10868921130895615, | |
| "logps/chosen": -673.7401123046875, | |
| "logps/rejected": -815.3800659179688, | |
| "loss": 0.2565, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.35063451528549194, | |
| "rewards/margins": 0.17919650673866272, | |
| "rewards/rejected": -0.529831051826477, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.763857324837321e-07, | |
| "logits/chosen": 0.035469502210617065, | |
| "logits/rejected": 0.10574381053447723, | |
| "logps/chosen": -697.9148559570312, | |
| "logps/rejected": -829.6324462890625, | |
| "loss": 0.191, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.3119833469390869, | |
| "rewards/margins": 0.24292249977588654, | |
| "rewards/rejected": -0.5549058318138123, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.429643575928605e-07, | |
| "logits/chosen": -0.019414827227592468, | |
| "logits/rejected": 0.18325455486774445, | |
| "logps/chosen": -718.9652099609375, | |
| "logps/rejected": -853.3225708007812, | |
| "loss": 0.2266, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.35121822357177734, | |
| "rewards/margins": 0.2009017914533615, | |
| "rewards/rejected": -0.5521200299263, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.101522917128709e-07, | |
| "logits/chosen": 0.023696620017290115, | |
| "logits/rejected": 0.10542090982198715, | |
| "logps/chosen": -635.32373046875, | |
| "logps/rejected": -819.9827880859375, | |
| "loss": 0.177, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.30631598830223083, | |
| "rewards/margins": 0.2354573756456375, | |
| "rewards/rejected": -0.5417734384536743, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.779609134902312e-07, | |
| "logits/chosen": 0.0738513395190239, | |
| "logits/rejected": 0.12634918093681335, | |
| "logps/chosen": -659.0910034179688, | |
| "logps/rejected": -865.4376831054688, | |
| "loss": 0.2054, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.3390515446662903, | |
| "rewards/margins": 0.23071372509002686, | |
| "rewards/rejected": -0.5697652697563171, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.464013863278629e-07, | |
| "logits/chosen": 0.0322534441947937, | |
| "logits/rejected": 0.1811327189207077, | |
| "logps/chosen": -716.4608764648438, | |
| "logps/rejected": -811.6767578125, | |
| "loss": 0.231, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3598640263080597, | |
| "rewards/margins": 0.19128559529781342, | |
| "rewards/rejected": -0.5511496663093567, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.154846545138696e-07, | |
| "logits/chosen": -0.00073561369208619, | |
| "logits/rejected": 0.0718742161989212, | |
| "logps/chosen": -638.3681640625, | |
| "logps/rejected": -820.8011474609375, | |
| "loss": 0.2099, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.3302134573459625, | |
| "rewards/margins": 0.2071600705385208, | |
| "rewards/rejected": -0.5373734831809998, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.852214394262515e-07, | |
| "logits/chosen": 0.00898817926645279, | |
| "logits/rejected": 0.10342226177453995, | |
| "logps/chosen": -722.0171508789062, | |
| "logps/rejected": -861.2794189453125, | |
| "loss": 0.2237, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3676268458366394, | |
| "rewards/margins": 0.21328619122505188, | |
| "rewards/rejected": -0.5809130072593689, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.556222358149191e-07, | |
| "logits/chosen": 0.01620279997587204, | |
| "logits/rejected": 0.12123896926641464, | |
| "logps/chosen": -617.8934326171875, | |
| "logps/rejected": -802.05859375, | |
| "loss": 0.2168, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.299948513507843, | |
| "rewards/margins": 0.219092458486557, | |
| "rewards/rejected": -0.5190409421920776, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.266973081622992e-07, | |
| "logits/chosen": 0.0352124460041523, | |
| "logits/rejected": 0.1431238204240799, | |
| "logps/chosen": -697.825927734375, | |
| "logps/rejected": -875.0059814453125, | |
| "loss": 0.2137, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3430352807044983, | |
| "rewards/margins": 0.21984973549842834, | |
| "rewards/rejected": -0.562885046005249, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.984566871237942e-07, | |
| "logits/chosen": 0.04245247319340706, | |
| "logits/rejected": 0.13526387512683868, | |
| "logps/chosen": -697.025390625, | |
| "logps/rejected": -813.5718994140625, | |
| "loss": 0.2386, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.33000361919403076, | |
| "rewards/margins": 0.21133120357990265, | |
| "rewards/rejected": -0.5413348078727722, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.709101660493251e-07, | |
| "logits/chosen": 0.02729245461523533, | |
| "logits/rejected": 0.14574703574180603, | |
| "logps/chosen": -676.8394775390625, | |
| "logps/rejected": -825.9822998046875, | |
| "loss": 0.2333, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3324502110481262, | |
| "rewards/margins": 0.20352134108543396, | |
| "rewards/rejected": -0.5359715223312378, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.440672975871743e-07, | |
| "logits/chosen": 0.04532911255955696, | |
| "logits/rejected": 0.12752528488636017, | |
| "logps/chosen": -633.0133056640625, | |
| "logps/rejected": -786.7761840820312, | |
| "loss": 0.2063, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.29472070932388306, | |
| "rewards/margins": 0.21680741012096405, | |
| "rewards/rejected": -0.5115281343460083, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1793739037129134e-07, | |
| "logits/chosen": 0.07257268577814102, | |
| "logits/rejected": 0.11432480812072754, | |
| "logps/chosen": -690.2648315429688, | |
| "logps/rejected": -863.9197998046875, | |
| "loss": 0.2096, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.3343331217765808, | |
| "rewards/margins": 0.23166866600513458, | |
| "rewards/rejected": -0.5660017728805542, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.9252950579322405e-07, | |
| "logits/chosen": 0.06963463127613068, | |
| "logits/rejected": 0.12501554191112518, | |
| "logps/chosen": -634.1485595703125, | |
| "logps/rejected": -776.090576171875, | |
| "loss": 0.2422, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.32659441232681274, | |
| "rewards/margins": 0.19114482402801514, | |
| "rewards/rejected": -0.5177391767501831, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.6785245485978864e-07, | |
| "logits/chosen": 0.0035268261563032866, | |
| "logits/rejected": 0.03672543540596962, | |
| "logps/chosen": -623.8313598632812, | |
| "logps/rejected": -851.9779052734375, | |
| "loss": 0.2306, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3241678476333618, | |
| "rewards/margins": 0.22523315250873566, | |
| "rewards/rejected": -0.5494009852409363, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.43914795137566e-07, | |
| "logits/chosen": 0.03905852884054184, | |
| "logits/rejected": 0.0981665775179863, | |
| "logps/chosen": -628.8436889648438, | |
| "logps/rejected": -775.3961791992188, | |
| "loss": 0.2274, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.3110875189304352, | |
| "rewards/margins": 0.2002953737974167, | |
| "rewards/rejected": -0.5113828778266907, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.207248277852901e-07, | |
| "logits/chosen": 0.06323795020580292, | |
| "logits/rejected": 0.1348869502544403, | |
| "logps/chosen": -645.1129150390625, | |
| "logps/rejected": -793.5126953125, | |
| "loss": 0.1864, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3031459450721741, | |
| "rewards/margins": 0.22209510207176208, | |
| "rewards/rejected": -0.5252410173416138, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.9829059467515074e-07, | |
| "logits/chosen": 0.07657162100076675, | |
| "logits/rejected": 0.11810870468616486, | |
| "logps/chosen": -679.9815673828125, | |
| "logps/rejected": -822.6639404296875, | |
| "loss": 0.2164, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.342891126871109, | |
| "rewards/margins": 0.2010626345872879, | |
| "rewards/rejected": -0.5439537167549133, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.766198756040153e-07, | |
| "logits/chosen": 0.065867580473423, | |
| "logits/rejected": 0.15143200755119324, | |
| "logps/chosen": -630.640869140625, | |
| "logps/rejected": -778.2337036132812, | |
| "loss": 0.1918, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3079833388328552, | |
| "rewards/margins": 0.2207847535610199, | |
| "rewards/rejected": -0.5287681221961975, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.5572018559553155e-07, | |
| "logits/chosen": 0.07310919463634491, | |
| "logits/rejected": 0.12045961618423462, | |
| "logps/chosen": -711.3140869140625, | |
| "logps/rejected": -828.66015625, | |
| "loss": 0.2413, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.34624183177948, | |
| "rewards/margins": 0.1804010570049286, | |
| "rewards/rejected": -0.526642918586731, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3559877229404864e-07, | |
| "logits/chosen": 0.038972992449998856, | |
| "logits/rejected": 0.1098276823759079, | |
| "logps/chosen": -679.78271484375, | |
| "logps/rejected": -838.59765625, | |
| "loss": 0.2205, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3377990126609802, | |
| "rewards/margins": 0.21535663306713104, | |
| "rewards/rejected": -0.5531556606292725, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.1626261345126576e-07, | |
| "logits/chosen": 0.05230052396655083, | |
| "logits/rejected": 0.14600692689418793, | |
| "logps/chosen": -679.2337036132812, | |
| "logps/rejected": -837.9591674804688, | |
| "loss": 0.2071, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.3297780156135559, | |
| "rewards/margins": 0.22520343959331512, | |
| "rewards/rejected": -0.5549814701080322, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.9771841450646505e-07, | |
| "logits/chosen": 0.09260080009698868, | |
| "logits/rejected": 0.1418459266424179, | |
| "logps/chosen": -681.7306518554688, | |
| "logps/rejected": -802.3380126953125, | |
| "loss": 0.2068, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.31467029452323914, | |
| "rewards/margins": 0.21341577172279358, | |
| "rewards/rejected": -0.5280860662460327, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.7997260626118758e-07, | |
| "logits/chosen": 0.01892709732055664, | |
| "logits/rejected": 0.10615235567092896, | |
| "logps/chosen": -638.789794921875, | |
| "logps/rejected": -835.4816284179688, | |
| "loss": 0.2051, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.3251439332962036, | |
| "rewards/margins": 0.21179835498332977, | |
| "rewards/rejected": -0.5369422435760498, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.6303134264914365e-07, | |
| "logits/chosen": 0.07935929298400879, | |
| "logits/rejected": 0.2099631130695343, | |
| "logps/chosen": -729.1849365234375, | |
| "logps/rejected": -823.9190673828125, | |
| "loss": 0.2305, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.32880938053131104, | |
| "rewards/margins": 0.21041882038116455, | |
| "rewards/rejected": -0.5392282009124756, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.469004986021355e-07, | |
| "logits/chosen": 0.056362926959991455, | |
| "logits/rejected": 0.13600429892539978, | |
| "logps/chosen": -694.4349975585938, | |
| "logps/rejected": -821.8929443359375, | |
| "loss": 0.2202, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.3222964107990265, | |
| "rewards/margins": 0.22275909781455994, | |
| "rewards/rejected": -0.5450555086135864, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.315856680127367e-07, | |
| "logits/chosen": 0.07223354279994965, | |
| "logits/rejected": 0.13364526629447937, | |
| "logps/chosen": -760.1466674804688, | |
| "logps/rejected": -892.9208984375, | |
| "loss": 0.1978, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.37695029377937317, | |
| "rewards/margins": 0.20259904861450195, | |
| "rewards/rejected": -0.5795493721961975, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1709216179442817e-07, | |
| "logits/chosen": 0.004451720509678125, | |
| "logits/rejected": 0.1476518213748932, | |
| "logps/chosen": -679.5355224609375, | |
| "logps/rejected": -841.3785400390625, | |
| "loss": 0.2127, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.3298358917236328, | |
| "rewards/margins": 0.2245192527770996, | |
| "rewards/rejected": -0.5543551445007324, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0342500603986421e-07, | |
| "logits/chosen": 0.008707459084689617, | |
| "logits/rejected": 0.09261924773454666, | |
| "logps/chosen": -672.3406982421875, | |
| "logps/rejected": -848.3216552734375, | |
| "loss": 0.1905, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3172846734523773, | |
| "rewards/margins": 0.24588418006896973, | |
| "rewards/rejected": -0.5631688833236694, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.058894027791643e-08, | |
| "logits/chosen": 0.0769379660487175, | |
| "logits/rejected": 0.10007413476705551, | |
| "logps/chosen": -699.5477294921875, | |
| "logps/rejected": -884.2034301757812, | |
| "loss": 0.1939, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.3175678253173828, | |
| "rewards/margins": 0.25022247433662415, | |
| "rewards/rejected": -0.5677902698516846, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.858841583008592e-08, | |
| "logits/chosen": 0.055988211184740067, | |
| "logits/rejected": 0.1475806087255478, | |
| "logps/chosen": -652.2313842773438, | |
| "logps/rejected": -836.2672119140625, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3141983151435852, | |
| "rewards/margins": 0.24290838837623596, | |
| "rewards/rejected": -0.5571067333221436, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.742759426686313e-08, | |
| "logits/chosen": 0.04174142703413963, | |
| "logits/rejected": 0.14616337418556213, | |
| "logps/chosen": -674.041015625, | |
| "logps/rejected": -825.32470703125, | |
| "loss": 0.1955, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.31099969148635864, | |
| "rewards/margins": 0.22915931046009064, | |
| "rewards/rejected": -0.5401589870452881, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.7110345964571104e-08, | |
| "logits/chosen": 0.057897310703992844, | |
| "logits/rejected": 0.157648503780365, | |
| "logps/chosen": -703.7222290039062, | |
| "logps/rejected": -847.68115234375, | |
| "loss": 0.2134, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3606089949607849, | |
| "rewards/margins": 0.21621307730674744, | |
| "rewards/rejected": -0.5768221616744995, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.764024876318357e-08, | |
| "logits/chosen": 0.06264643371105194, | |
| "logits/rejected": 0.14495016634464264, | |
| "logps/chosen": -730.0210571289062, | |
| "logps/rejected": -846.86474609375, | |
| "loss": 0.2213, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.35915127396583557, | |
| "rewards/margins": 0.19883516430854797, | |
| "rewards/rejected": -0.5579863786697388, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.902058672559633e-08, | |
| "logits/chosen": 0.07288487255573273, | |
| "logits/rejected": 0.1731829047203064, | |
| "logps/chosen": -698.0350341796875, | |
| "logps/rejected": -848.1388549804688, | |
| "loss": 0.1737, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.31695953011512756, | |
| "rewards/margins": 0.2423754632472992, | |
| "rewards/rejected": -0.5593350529670715, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.125434899876933e-08, | |
| "logits/chosen": 0.06433729827404022, | |
| "logits/rejected": 0.13777321577072144, | |
| "logps/chosen": -716.5806884765625, | |
| "logps/rejected": -871.9490356445312, | |
| "loss": 0.2003, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3185502290725708, | |
| "rewards/margins": 0.23725280165672302, | |
| "rewards/rejected": -0.5558030009269714, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.4344228777145873e-08, | |
| "logits/chosen": 0.038981080055236816, | |
| "logits/rejected": 0.1305188238620758, | |
| "logps/chosen": -728.6278076171875, | |
| "logps/rejected": -865.8338012695312, | |
| "loss": 0.2074, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.34264928102493286, | |
| "rewards/margins": 0.23139305412769318, | |
| "rewards/rejected": -0.5740423202514648, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.829262236869772e-08, | |
| "logits/chosen": 0.06916898488998413, | |
| "logits/rejected": 0.18797878921031952, | |
| "logps/chosen": -702.5946044921875, | |
| "logps/rejected": -872.0211791992188, | |
| "loss": 0.2173, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.34207233786582947, | |
| "rewards/margins": 0.2262537032365799, | |
| "rewards/rejected": -0.5683261156082153, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3101628363929586e-08, | |
| "logits/chosen": 0.05317959189414978, | |
| "logits/rejected": 0.11542042344808578, | |
| "logps/chosen": -715.3739624023438, | |
| "logps/rejected": -858.2398681640625, | |
| "loss": 0.2227, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.34564128518104553, | |
| "rewards/margins": 0.21440550684928894, | |
| "rewards/rejected": -0.5600467920303345, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.773046908123195e-09, | |
| "logits/chosen": 0.07714705169200897, | |
| "logits/rejected": 0.1786491423845291, | |
| "logps/chosen": -677.5328979492188, | |
| "logps/rejected": -854.1397705078125, | |
| "loss": 0.2085, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.32333892583847046, | |
| "rewards/margins": 0.22118084132671356, | |
| "rewards/rejected": -0.5445197820663452, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.308379077080817e-09, | |
| "logits/chosen": 0.09814608097076416, | |
| "logits/rejected": 0.09552597999572754, | |
| "logps/chosen": -678.3449096679688, | |
| "logps/rejected": -837.9964599609375, | |
| "loss": 0.2152, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3395656943321228, | |
| "rewards/margins": 0.21091222763061523, | |
| "rewards/rejected": -0.5504779815673828, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.7088263565760996e-09, | |
| "logits/chosen": 0.08845305442810059, | |
| "logits/rejected": 0.13582393527030945, | |
| "logps/chosen": -649.8729248046875, | |
| "logps/rejected": -840.5818481445312, | |
| "loss": 0.2022, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.32875552773475647, | |
| "rewards/margins": 0.22628939151763916, | |
| "rewards/rejected": -0.555044949054718, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.752902257023633e-10, | |
| "logits/chosen": 0.041667476296424866, | |
| "logits/rejected": 0.1802738904953003, | |
| "logps/chosen": -713.4054565429688, | |
| "logps/rejected": -848.8229370117188, | |
| "loss": 0.2413, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3549792766571045, | |
| "rewards/margins": 0.2078593522310257, | |
| "rewards/rejected": -0.5628384947776794, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.083718442532189e-10, | |
| "logits/chosen": 0.07584629207849503, | |
| "logits/rejected": 0.15728022158145905, | |
| "logps/chosen": -741.0794677734375, | |
| "logps/rejected": -853.7579956054688, | |
| "loss": 0.2253, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3356108069419861, | |
| "rewards/margins": 0.20086529850959778, | |
| "rewards/rejected": -0.5364760756492615, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1875, | |
| "total_flos": 0.0, | |
| "train_loss": 0.2332912155151367, | |
| "train_runtime": 17312.5985, | |
| "train_samples_per_second": 1.733, | |
| "train_steps_per_second": 0.108 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1875, | |
| "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 | |
| } | |