| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9980364656381484, | |
| "eval_steps": 100, | |
| "global_step": 2004, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04488078541374474, | |
| "grad_norm": 4.790558815002441, | |
| "learning_rate": 9.850299401197606e-05, | |
| "logits/chosen": -3.3742988109588623, | |
| "logits/rejected": -3.0817112922668457, | |
| "logps/chosen": -273.48614501953125, | |
| "logps/rejected": -234.3329315185547, | |
| "loss": 0.6831, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": 0.0712718814611435, | |
| "rewards/margins": 0.024287192150950432, | |
| "rewards/rejected": 0.04698468744754791, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08976157082748948, | |
| "grad_norm": 5.551278591156006, | |
| "learning_rate": 9.700598802395209e-05, | |
| "logits/chosen": -3.378220558166504, | |
| "logits/rejected": -3.129826307296753, | |
| "logps/chosen": -267.0759582519531, | |
| "logps/rejected": -238.60873413085938, | |
| "loss": 0.6691, | |
| "rewards/accuracies": 0.590624988079071, | |
| "rewards/chosen": 0.21243497729301453, | |
| "rewards/margins": 0.07040555775165558, | |
| "rewards/rejected": 0.14202943444252014, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13464235624123422, | |
| "grad_norm": 7.342463493347168, | |
| "learning_rate": 9.550898203592816e-05, | |
| "logits/chosen": -3.3940162658691406, | |
| "logits/rejected": -3.142778158187866, | |
| "logps/chosen": -267.77581787109375, | |
| "logps/rejected": -233.1001434326172, | |
| "loss": 0.6586, | |
| "rewards/accuracies": 0.6145833134651184, | |
| "rewards/chosen": 0.2707298994064331, | |
| "rewards/margins": 0.12286876887083054, | |
| "rewards/rejected": 0.14786113798618317, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1496026180458158, | |
| "eval_logits/chosen": -3.403446674346924, | |
| "eval_logits/rejected": -3.1215860843658447, | |
| "eval_logps/chosen": -268.9533386230469, | |
| "eval_logps/rejected": -229.84756469726562, | |
| "eval_loss": 0.6490052342414856, | |
| "eval_rewards/accuracies": 0.6090127229690552, | |
| "eval_rewards/chosen": 0.2928723990917206, | |
| "eval_rewards/margins": 0.17060735821723938, | |
| "eval_rewards/rejected": 0.12226507067680359, | |
| "eval_runtime": 1689.4226, | |
| "eval_samples_per_second": 3.166, | |
| "eval_steps_per_second": 3.166, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.17952314165497896, | |
| "grad_norm": 7.352132797241211, | |
| "learning_rate": 9.40119760479042e-05, | |
| "logits/chosen": -3.416881561279297, | |
| "logits/rejected": -3.130246877670288, | |
| "logps/chosen": -271.2243347167969, | |
| "logps/rejected": -233.06614685058594, | |
| "loss": 0.6385, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": 0.295549601316452, | |
| "rewards/margins": 0.20651350915431976, | |
| "rewards/rejected": 0.08903612196445465, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2244039270687237, | |
| "grad_norm": 7.124303340911865, | |
| "learning_rate": 9.251497005988024e-05, | |
| "logits/chosen": -3.3911712169647217, | |
| "logits/rejected": -3.1303460597991943, | |
| "logps/chosen": -265.11651611328125, | |
| "logps/rejected": -234.63478088378906, | |
| "loss": 0.6436, | |
| "rewards/accuracies": 0.6322916746139526, | |
| "rewards/chosen": 0.23116879165172577, | |
| "rewards/margins": 0.23007448017597198, | |
| "rewards/rejected": 0.0010943154338747263, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.26928471248246844, | |
| "grad_norm": 6.228757381439209, | |
| "learning_rate": 9.101796407185628e-05, | |
| "logits/chosen": -3.4022183418273926, | |
| "logits/rejected": -3.151683807373047, | |
| "logps/chosen": -267.1882019042969, | |
| "logps/rejected": -232.48471069335938, | |
| "loss": 0.6485, | |
| "rewards/accuracies": 0.6322916746139526, | |
| "rewards/chosen": 0.3013507127761841, | |
| "rewards/margins": 0.20289267599582672, | |
| "rewards/rejected": 0.09845803678035736, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2992052360916316, | |
| "eval_logits/chosen": -3.4198923110961914, | |
| "eval_logits/rejected": -3.143324613571167, | |
| "eval_logps/chosen": -268.99163818359375, | |
| "eval_logps/rejected": -230.40538024902344, | |
| "eval_loss": 0.6396481394767761, | |
| "eval_rewards/accuracies": 0.6142483353614807, | |
| "eval_rewards/chosen": 0.2890413999557495, | |
| "eval_rewards/margins": 0.22255805134773254, | |
| "eval_rewards/rejected": 0.06648338586091995, | |
| "eval_runtime": 1688.7699, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3141654978962132, | |
| "grad_norm": 6.4461493492126465, | |
| "learning_rate": 8.952095808383235e-05, | |
| "logits/chosen": -3.434596300125122, | |
| "logits/rejected": -3.132395029067993, | |
| "logps/chosen": -267.54437255859375, | |
| "logps/rejected": -224.43540954589844, | |
| "loss": 0.638, | |
| "rewards/accuracies": 0.6197916865348816, | |
| "rewards/chosen": 0.2697771489620209, | |
| "rewards/margins": 0.2309388816356659, | |
| "rewards/rejected": 0.03883826732635498, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3590462833099579, | |
| "grad_norm": 4.610179424285889, | |
| "learning_rate": 8.80239520958084e-05, | |
| "logits/chosen": -3.428438901901245, | |
| "logits/rejected": -3.169167995452881, | |
| "logps/chosen": -265.898193359375, | |
| "logps/rejected": -230.3724822998047, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.6166666746139526, | |
| "rewards/chosen": 0.28266531229019165, | |
| "rewards/margins": 0.23549547791481018, | |
| "rewards/rejected": 0.04716984182596207, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.40392706872370265, | |
| "grad_norm": 5.838982582092285, | |
| "learning_rate": 8.652694610778443e-05, | |
| "logits/chosen": -3.4263041019439697, | |
| "logits/rejected": -3.1684255599975586, | |
| "logps/chosen": -267.129150390625, | |
| "logps/rejected": -233.6484832763672, | |
| "loss": 0.6251, | |
| "rewards/accuracies": 0.6302083134651184, | |
| "rewards/chosen": 0.1708066761493683, | |
| "rewards/margins": 0.28462281823158264, | |
| "rewards/rejected": -0.11381613463163376, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.4488078541374474, | |
| "grad_norm": 5.045168876647949, | |
| "learning_rate": 8.502994011976048e-05, | |
| "logits/chosen": -3.439959764480591, | |
| "logits/rejected": -3.173079490661621, | |
| "logps/chosen": -273.4431457519531, | |
| "logps/rejected": -236.24371337890625, | |
| "loss": 0.6327, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": 0.17748431861400604, | |
| "rewards/margins": 0.269964337348938, | |
| "rewards/rejected": -0.09247999638319016, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4488078541374474, | |
| "eval_logits/chosen": -3.424139976501465, | |
| "eval_logits/rejected": -3.150641918182373, | |
| "eval_logps/chosen": -269.5113525390625, | |
| "eval_logps/rejected": -231.47146606445312, | |
| "eval_loss": 0.6353974938392639, | |
| "eval_rewards/accuracies": 0.6181750297546387, | |
| "eval_rewards/chosen": 0.23706810176372528, | |
| "eval_rewards/margins": 0.2771916091442108, | |
| "eval_rewards/rejected": -0.04012349247932434, | |
| "eval_runtime": 1688.647, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.49368863955119213, | |
| "grad_norm": 5.068808555603027, | |
| "learning_rate": 8.353293413173653e-05, | |
| "logits/chosen": -3.4144017696380615, | |
| "logits/rejected": -3.1683106422424316, | |
| "logps/chosen": -272.5228271484375, | |
| "logps/rejected": -239.20681762695312, | |
| "loss": 0.646, | |
| "rewards/accuracies": 0.6208333373069763, | |
| "rewards/chosen": 0.22192205488681793, | |
| "rewards/margins": 0.24012483656406403, | |
| "rewards/rejected": -0.018202781677246094, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5385694249649369, | |
| "grad_norm": 6.0258941650390625, | |
| "learning_rate": 8.203592814371259e-05, | |
| "logits/chosen": -3.4079012870788574, | |
| "logits/rejected": -3.1440396308898926, | |
| "logps/chosen": -276.3011474609375, | |
| "logps/rejected": -235.62054443359375, | |
| "loss": 0.6228, | |
| "rewards/accuracies": 0.6270833611488342, | |
| "rewards/chosen": 0.09719991683959961, | |
| "rewards/margins": 0.2948659658432007, | |
| "rewards/rejected": -0.19766603410243988, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.5834502103786816, | |
| "grad_norm": 5.713747024536133, | |
| "learning_rate": 8.053892215568862e-05, | |
| "logits/chosen": -3.3723533153533936, | |
| "logits/rejected": -3.1148476600646973, | |
| "logps/chosen": -274.2776794433594, | |
| "logps/rejected": -234.34136962890625, | |
| "loss": 0.6342, | |
| "rewards/accuracies": 0.6270833611488342, | |
| "rewards/chosen": 0.17114956676959991, | |
| "rewards/margins": 0.29201894998550415, | |
| "rewards/rejected": -0.12086938321590424, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5984104721832632, | |
| "eval_logits/chosen": -3.390080451965332, | |
| "eval_logits/rejected": -3.111392021179199, | |
| "eval_logps/chosen": -268.8110656738281, | |
| "eval_logps/rejected": -230.63925170898438, | |
| "eval_loss": 0.6309967041015625, | |
| "eval_rewards/accuracies": 0.630142092704773, | |
| "eval_rewards/chosen": 0.30709749460220337, | |
| "eval_rewards/margins": 0.2640005946159363, | |
| "eval_rewards/rejected": 0.04309689626097679, | |
| "eval_runtime": 1688.4508, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.6283309957924264, | |
| "grad_norm": 5.401741027832031, | |
| "learning_rate": 7.904191616766467e-05, | |
| "logits/chosen": -3.389784574508667, | |
| "logits/rejected": -3.113330602645874, | |
| "logps/chosen": -270.7179260253906, | |
| "logps/rejected": -234.7329864501953, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.6260416507720947, | |
| "rewards/chosen": 0.2668881416320801, | |
| "rewards/margins": 0.25750893354415894, | |
| "rewards/rejected": 0.009379198774695396, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.6732117812061711, | |
| "grad_norm": 6.569407939910889, | |
| "learning_rate": 7.754491017964072e-05, | |
| "logits/chosen": -3.420933246612549, | |
| "logits/rejected": -3.107722520828247, | |
| "logps/chosen": -279.6606750488281, | |
| "logps/rejected": -232.16326904296875, | |
| "loss": 0.6152, | |
| "rewards/accuracies": 0.6395833492279053, | |
| "rewards/chosen": 0.23483921587467194, | |
| "rewards/margins": 0.30834510922431946, | |
| "rewards/rejected": -0.07350588589906693, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7180925666199158, | |
| "grad_norm": 4.889843463897705, | |
| "learning_rate": 7.604790419161677e-05, | |
| "logits/chosen": -3.4380805492401123, | |
| "logits/rejected": -3.1253116130828857, | |
| "logps/chosen": -279.8207092285156, | |
| "logps/rejected": -233.9474639892578, | |
| "loss": 0.612, | |
| "rewards/accuracies": 0.6697916388511658, | |
| "rewards/chosen": 0.06458248198032379, | |
| "rewards/margins": 0.378538578748703, | |
| "rewards/rejected": -0.3139561414718628, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.748013090229079, | |
| "eval_logits/chosen": -3.419874429702759, | |
| "eval_logits/rejected": -3.1424779891967773, | |
| "eval_logps/chosen": -270.2199401855469, | |
| "eval_logps/rejected": -232.4933319091797, | |
| "eval_loss": 0.6269693374633789, | |
| "eval_rewards/accuracies": 0.627711296081543, | |
| "eval_rewards/chosen": 0.16621026396751404, | |
| "eval_rewards/margins": 0.3085208237171173, | |
| "eval_rewards/rejected": -0.14231054484844208, | |
| "eval_runtime": 1688.3228, | |
| "eval_samples_per_second": 3.168, | |
| "eval_steps_per_second": 3.168, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7629733520336606, | |
| "grad_norm": 4.36655330657959, | |
| "learning_rate": 7.455089820359282e-05, | |
| "logits/chosen": -3.4343178272247314, | |
| "logits/rejected": -3.1612308025360107, | |
| "logps/chosen": -272.99578857421875, | |
| "logps/rejected": -234.4145965576172, | |
| "loss": 0.629, | |
| "rewards/accuracies": 0.6427083611488342, | |
| "rewards/chosen": 0.13686993718147278, | |
| "rewards/margins": 0.3045249283313751, | |
| "rewards/rejected": -0.16765499114990234, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.8078541374474053, | |
| "grad_norm": 4.971264839172363, | |
| "learning_rate": 7.305389221556886e-05, | |
| "logits/chosen": -3.4246087074279785, | |
| "logits/rejected": -3.172884464263916, | |
| "logps/chosen": -267.14556884765625, | |
| "logps/rejected": -233.85691833496094, | |
| "loss": 0.6269, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 0.22364649176597595, | |
| "rewards/margins": 0.28497254848480225, | |
| "rewards/rejected": -0.06132606416940689, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.85273492286115, | |
| "grad_norm": 5.077197074890137, | |
| "learning_rate": 7.155688622754491e-05, | |
| "logits/chosen": -3.4349772930145264, | |
| "logits/rejected": -3.1722497940063477, | |
| "logps/chosen": -268.02630615234375, | |
| "logps/rejected": -231.99020385742188, | |
| "loss": 0.63, | |
| "rewards/accuracies": 0.621874988079071, | |
| "rewards/chosen": 0.2208840399980545, | |
| "rewards/margins": 0.2860158383846283, | |
| "rewards/rejected": -0.06513180583715439, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.8976157082748948, | |
| "grad_norm": 4.760651111602783, | |
| "learning_rate": 7.005988023952096e-05, | |
| "logits/chosen": -3.4018094539642334, | |
| "logits/rejected": -3.1606853008270264, | |
| "logps/chosen": -268.86090087890625, | |
| "logps/rejected": -233.84007263183594, | |
| "loss": 0.6432, | |
| "rewards/accuracies": 0.6208333373069763, | |
| "rewards/chosen": 0.25363439321517944, | |
| "rewards/margins": 0.2553554177284241, | |
| "rewards/rejected": -0.0017210314981639385, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.8976157082748948, | |
| "eval_logits/chosen": -3.4228434562683105, | |
| "eval_logits/rejected": -3.145069122314453, | |
| "eval_logps/chosen": -269.40740966796875, | |
| "eval_logps/rejected": -231.58685302734375, | |
| "eval_loss": 0.6246524453163147, | |
| "eval_rewards/accuracies": 0.6312640309333801, | |
| "eval_rewards/chosen": 0.24746553599834442, | |
| "eval_rewards/margins": 0.2991257905960083, | |
| "eval_rewards/rejected": -0.051660239696502686, | |
| "eval_runtime": 1688.5162, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.9424964936886395, | |
| "grad_norm": 5.144285678863525, | |
| "learning_rate": 6.856287425149701e-05, | |
| "logits/chosen": -3.4329488277435303, | |
| "logits/rejected": -3.1452276706695557, | |
| "logps/chosen": -269.5411376953125, | |
| "logps/rejected": -228.07046508789062, | |
| "loss": 0.6185, | |
| "rewards/accuracies": 0.6260416507720947, | |
| "rewards/chosen": 0.26634886860847473, | |
| "rewards/margins": 0.3162167966365814, | |
| "rewards/rejected": -0.049867913126945496, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.9873772791023843, | |
| "grad_norm": 4.551113128662109, | |
| "learning_rate": 6.706586826347305e-05, | |
| "logits/chosen": -3.435673713684082, | |
| "logits/rejected": -3.1743414402008057, | |
| "logps/chosen": -273.6510314941406, | |
| "logps/rejected": -241.5527801513672, | |
| "loss": 0.6236, | |
| "rewards/accuracies": 0.6364583373069763, | |
| "rewards/chosen": 0.11331641674041748, | |
| "rewards/margins": 0.32483571767807007, | |
| "rewards/rejected": -0.2115192860364914, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.032258064516129, | |
| "grad_norm": 4.672567844390869, | |
| "learning_rate": 6.55688622754491e-05, | |
| "logits/chosen": -3.4276251792907715, | |
| "logits/rejected": -3.1490509510040283, | |
| "logps/chosen": -269.5851135253906, | |
| "logps/rejected": -237.02989196777344, | |
| "loss": 0.5554, | |
| "rewards/accuracies": 0.7302083373069763, | |
| "rewards/chosen": 0.08210794627666473, | |
| "rewards/margins": 0.49889788031578064, | |
| "rewards/rejected": -0.4167899191379547, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.0472183263207107, | |
| "eval_logits/chosen": -3.4185428619384766, | |
| "eval_logits/rejected": -3.1414475440979004, | |
| "eval_logps/chosen": -270.3139343261719, | |
| "eval_logps/rejected": -232.80120849609375, | |
| "eval_loss": 0.6221644282341003, | |
| "eval_rewards/accuracies": 0.6338818073272705, | |
| "eval_rewards/chosen": 0.15681201219558716, | |
| "eval_rewards/margins": 0.3299100995063782, | |
| "eval_rewards/rejected": -0.17309808731079102, | |
| "eval_runtime": 1688.7954, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.0771388499298737, | |
| "grad_norm": 4.2797369956970215, | |
| "learning_rate": 6.407185628742515e-05, | |
| "logits/chosen": -3.445012092590332, | |
| "logits/rejected": -3.1330997943878174, | |
| "logps/chosen": -266.95782470703125, | |
| "logps/rejected": -227.27879333496094, | |
| "loss": 0.5249, | |
| "rewards/accuracies": 0.7635416388511658, | |
| "rewards/chosen": 0.25232160091400146, | |
| "rewards/margins": 0.5462218523025513, | |
| "rewards/rejected": -0.2939002215862274, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.1220196353436185, | |
| "grad_norm": 5.101881980895996, | |
| "learning_rate": 6.25748502994012e-05, | |
| "logits/chosen": -3.425431728363037, | |
| "logits/rejected": -3.1551194190979004, | |
| "logps/chosen": -271.9197082519531, | |
| "logps/rejected": -233.2086639404297, | |
| "loss": 0.5308, | |
| "rewards/accuracies": 0.7385416626930237, | |
| "rewards/chosen": 0.2957630157470703, | |
| "rewards/margins": 0.5949270129203796, | |
| "rewards/rejected": -0.29916396737098694, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.1669004207573632, | |
| "grad_norm": 5.360141754150391, | |
| "learning_rate": 6.107784431137725e-05, | |
| "logits/chosen": -3.4009079933166504, | |
| "logits/rejected": -3.1200320720672607, | |
| "logps/chosen": -272.1022644042969, | |
| "logps/rejected": -236.18499755859375, | |
| "loss": 0.5226, | |
| "rewards/accuracies": 0.7520833611488342, | |
| "rewards/chosen": 0.30986490845680237, | |
| "rewards/margins": 0.59710294008255, | |
| "rewards/rejected": -0.2872380018234253, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.1968209443665265, | |
| "eval_logits/chosen": -3.405834674835205, | |
| "eval_logits/rejected": -3.1333200931549072, | |
| "eval_logps/chosen": -272.1116638183594, | |
| "eval_logps/rejected": -235.3762664794922, | |
| "eval_loss": 0.6281805038452148, | |
| "eval_rewards/accuracies": 0.6335078477859497, | |
| "eval_rewards/chosen": -0.022958112880587578, | |
| "eval_rewards/margins": 0.40764307975769043, | |
| "eval_rewards/rejected": -0.43060120940208435, | |
| "eval_runtime": 1688.3487, | |
| "eval_samples_per_second": 3.168, | |
| "eval_steps_per_second": 3.168, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.211781206171108, | |
| "grad_norm": 5.8285088539123535, | |
| "learning_rate": 5.95808383233533e-05, | |
| "logits/chosen": -3.3979651927948, | |
| "logits/rejected": -3.1520321369171143, | |
| "logps/chosen": -274.0641174316406, | |
| "logps/rejected": -240.42205810546875, | |
| "loss": 0.5402, | |
| "rewards/accuracies": 0.7354166507720947, | |
| "rewards/chosen": 0.1310122311115265, | |
| "rewards/margins": 0.5752567052841187, | |
| "rewards/rejected": -0.44424447417259216, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.2566619915848527, | |
| "grad_norm": 5.5216851234436035, | |
| "learning_rate": 5.808383233532935e-05, | |
| "logits/chosen": -3.4025676250457764, | |
| "logits/rejected": -3.1448330879211426, | |
| "logps/chosen": -274.1934509277344, | |
| "logps/rejected": -243.01490783691406, | |
| "loss": 0.5201, | |
| "rewards/accuracies": 0.7552083134651184, | |
| "rewards/chosen": 0.14752289652824402, | |
| "rewards/margins": 0.626766562461853, | |
| "rewards/rejected": -0.479243665933609, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.3015427769985974, | |
| "grad_norm": 5.673742294311523, | |
| "learning_rate": 5.6586826347305385e-05, | |
| "logits/chosen": -3.3895277976989746, | |
| "logits/rejected": -3.1401288509368896, | |
| "logps/chosen": -273.1130676269531, | |
| "logps/rejected": -241.987060546875, | |
| "loss": 0.5497, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.02567141316831112, | |
| "rewards/margins": 0.5613437294960022, | |
| "rewards/rejected": -0.5356722474098206, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.3464235624123422, | |
| "grad_norm": 6.6557440757751465, | |
| "learning_rate": 5.508982035928144e-05, | |
| "logits/chosen": -3.3836898803710938, | |
| "logits/rejected": -3.1433603763580322, | |
| "logps/chosen": -266.1312561035156, | |
| "logps/rejected": -238.70309448242188, | |
| "loss": 0.5474, | |
| "rewards/accuracies": 0.7322916388511658, | |
| "rewards/chosen": 0.016369260847568512, | |
| "rewards/margins": 0.563011109828949, | |
| "rewards/rejected": -0.5466418862342834, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.3464235624123422, | |
| "eval_logits/chosen": -3.3754773139953613, | |
| "eval_logits/rejected": -3.106959819793701, | |
| "eval_logps/chosen": -271.97869873046875, | |
| "eval_logps/rejected": -234.78985595703125, | |
| "eval_loss": 0.629611074924469, | |
| "eval_rewards/accuracies": 0.6299551129341125, | |
| "eval_rewards/chosen": -0.009662697091698647, | |
| "eval_rewards/margins": 0.3622985780239105, | |
| "eval_rewards/rejected": -0.3719612658023834, | |
| "eval_runtime": 1688.3293, | |
| "eval_samples_per_second": 3.168, | |
| "eval_steps_per_second": 3.168, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.391304347826087, | |
| "grad_norm": 5.2359724044799805, | |
| "learning_rate": 5.359281437125748e-05, | |
| "logits/chosen": -3.3651976585388184, | |
| "logits/rejected": -3.123444080352783, | |
| "logps/chosen": -271.6989440917969, | |
| "logps/rejected": -236.84417724609375, | |
| "loss": 0.54, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 0.1108192428946495, | |
| "rewards/margins": 0.5575817823410034, | |
| "rewards/rejected": -0.4467625319957733, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.4361851332398317, | |
| "grad_norm": 5.669713497161865, | |
| "learning_rate": 5.209580838323354e-05, | |
| "logits/chosen": -3.3611793518066406, | |
| "logits/rejected": -3.099807024002075, | |
| "logps/chosen": -274.7862243652344, | |
| "logps/rejected": -237.52139282226562, | |
| "loss": 0.5405, | |
| "rewards/accuracies": 0.7260416746139526, | |
| "rewards/chosen": 0.027847904711961746, | |
| "rewards/margins": 0.5692722797393799, | |
| "rewards/rejected": -0.5414243936538696, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.4810659186535764, | |
| "grad_norm": 6.500946044921875, | |
| "learning_rate": 5.059880239520959e-05, | |
| "logits/chosen": -3.3827614784240723, | |
| "logits/rejected": -3.09436297416687, | |
| "logps/chosen": -276.1948547363281, | |
| "logps/rejected": -238.23243713378906, | |
| "loss": 0.5235, | |
| "rewards/accuracies": 0.7593749761581421, | |
| "rewards/chosen": -0.051430441439151764, | |
| "rewards/margins": 0.6151652932167053, | |
| "rewards/rejected": -0.6665957570075989, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.496026180458158, | |
| "eval_logits/chosen": -3.366751194000244, | |
| "eval_logits/rejected": -3.1013710498809814, | |
| "eval_logps/chosen": -272.0386047363281, | |
| "eval_logps/rejected": -234.93768310546875, | |
| "eval_loss": 0.628265380859375, | |
| "eval_rewards/accuracies": 0.6325729489326477, | |
| "eval_rewards/chosen": -0.01565566658973694, | |
| "eval_rewards/margins": 0.37109050154685974, | |
| "eval_rewards/rejected": -0.38674619793891907, | |
| "eval_runtime": 1688.5815, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.5259467040673211, | |
| "grad_norm": 5.458423614501953, | |
| "learning_rate": 4.910179640718563e-05, | |
| "logits/chosen": -3.3471567630767822, | |
| "logits/rejected": -3.1277146339416504, | |
| "logps/chosen": -269.085205078125, | |
| "logps/rejected": -243.0161895751953, | |
| "loss": 0.5338, | |
| "rewards/accuracies": 0.7489583492279053, | |
| "rewards/chosen": 0.08274559676647186, | |
| "rewards/margins": 0.5906849503517151, | |
| "rewards/rejected": -0.5079393982887268, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.5708274894810659, | |
| "grad_norm": 7.57076358795166, | |
| "learning_rate": 4.7604790419161675e-05, | |
| "logits/chosen": -3.3720383644104004, | |
| "logits/rejected": -3.0789096355438232, | |
| "logps/chosen": -266.96270751953125, | |
| "logps/rejected": -228.08006286621094, | |
| "loss": 0.5331, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": 0.014148912392556667, | |
| "rewards/margins": 0.5797646045684814, | |
| "rewards/rejected": -0.5656156539916992, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.6157082748948106, | |
| "grad_norm": 5.330569744110107, | |
| "learning_rate": 4.610778443113773e-05, | |
| "logits/chosen": -3.359792947769165, | |
| "logits/rejected": -3.09041166305542, | |
| "logps/chosen": -276.38226318359375, | |
| "logps/rejected": -243.1844024658203, | |
| "loss": 0.5232, | |
| "rewards/accuracies": 0.7395833134651184, | |
| "rewards/chosen": 0.06125294789671898, | |
| "rewards/margins": 0.6237131953239441, | |
| "rewards/rejected": -0.56246018409729, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.645628798503974, | |
| "eval_logits/chosen": -3.3514223098754883, | |
| "eval_logits/rejected": -3.0870354175567627, | |
| "eval_logps/chosen": -271.9643859863281, | |
| "eval_logps/rejected": -234.57383728027344, | |
| "eval_loss": 0.6333222389221191, | |
| "eval_rewards/accuracies": 0.620792806148529, | |
| "eval_rewards/chosen": -0.008234047330915928, | |
| "eval_rewards/margins": 0.34212782979011536, | |
| "eval_rewards/rejected": -0.3503618538379669, | |
| "eval_runtime": 1688.4687, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.6605890603085554, | |
| "grad_norm": 6.506576061248779, | |
| "learning_rate": 4.4610778443113777e-05, | |
| "logits/chosen": -3.3484690189361572, | |
| "logits/rejected": -3.130286455154419, | |
| "logps/chosen": -273.8847961425781, | |
| "logps/rejected": -247.9475860595703, | |
| "loss": 0.5384, | |
| "rewards/accuracies": 0.7322916388511658, | |
| "rewards/chosen": 0.10436714440584183, | |
| "rewards/margins": 0.5445392727851868, | |
| "rewards/rejected": -0.44017213582992554, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.7054698457223, | |
| "grad_norm": 6.028050422668457, | |
| "learning_rate": 4.311377245508982e-05, | |
| "logits/chosen": -3.3555943965911865, | |
| "logits/rejected": -3.125157356262207, | |
| "logps/chosen": -267.73516845703125, | |
| "logps/rejected": -236.43356323242188, | |
| "loss": 0.5549, | |
| "rewards/accuracies": 0.7354166507720947, | |
| "rewards/chosen": -0.052730146795511246, | |
| "rewards/margins": 0.5250240564346313, | |
| "rewards/rejected": -0.5777541995048523, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.7503506311360448, | |
| "grad_norm": 6.8205647468566895, | |
| "learning_rate": 4.161676646706587e-05, | |
| "logits/chosen": -3.383364677429199, | |
| "logits/rejected": -3.1156816482543945, | |
| "logps/chosen": -273.9105529785156, | |
| "logps/rejected": -237.84432983398438, | |
| "loss": 0.523, | |
| "rewards/accuracies": 0.746874988079071, | |
| "rewards/chosen": -0.06969426572322845, | |
| "rewards/margins": 0.6354466676712036, | |
| "rewards/rejected": -0.705141007900238, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.7952314165497896, | |
| "grad_norm": 6.074138641357422, | |
| "learning_rate": 4.0119760479041915e-05, | |
| "logits/chosen": -3.391815185546875, | |
| "logits/rejected": -3.1276473999023438, | |
| "logps/chosen": -279.9575500488281, | |
| "logps/rejected": -244.89010620117188, | |
| "loss": 0.5156, | |
| "rewards/accuracies": 0.7479166388511658, | |
| "rewards/chosen": -0.07707042992115021, | |
| "rewards/margins": 0.6421669125556946, | |
| "rewards/rejected": -0.7192373871803284, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.7952314165497896, | |
| "eval_logits/chosen": -3.379970073699951, | |
| "eval_logits/rejected": -3.1176722049713135, | |
| "eval_logps/chosen": -274.0703430175781, | |
| "eval_logps/rejected": -237.28604125976562, | |
| "eval_loss": 0.6306100487709045, | |
| "eval_rewards/accuracies": 0.6350037455558777, | |
| "eval_rewards/chosen": -0.21883098781108856, | |
| "eval_rewards/margins": 0.4027484953403473, | |
| "eval_rewards/rejected": -0.6215794086456299, | |
| "eval_runtime": 1688.6398, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.8401122019635343, | |
| "grad_norm": 5.888485431671143, | |
| "learning_rate": 3.8622754491017966e-05, | |
| "logits/chosen": -3.3851656913757324, | |
| "logits/rejected": -3.1222336292266846, | |
| "logps/chosen": -272.52117919921875, | |
| "logps/rejected": -237.61158752441406, | |
| "loss": 0.5372, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -0.14584079384803772, | |
| "rewards/margins": 0.6132307052612305, | |
| "rewards/rejected": -0.7590714693069458, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.884992987377279, | |
| "grad_norm": 6.371288299560547, | |
| "learning_rate": 3.712574850299401e-05, | |
| "logits/chosen": -3.379087209701538, | |
| "logits/rejected": -3.1119582653045654, | |
| "logps/chosen": -273.9693603515625, | |
| "logps/rejected": -238.9442901611328, | |
| "loss": 0.5142, | |
| "rewards/accuracies": 0.7614583373069763, | |
| "rewards/chosen": -0.13823945820331573, | |
| "rewards/margins": 0.6301066279411316, | |
| "rewards/rejected": -0.7683460116386414, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.9298737727910238, | |
| "grad_norm": 6.35048246383667, | |
| "learning_rate": 3.562874251497006e-05, | |
| "logits/chosen": -3.3981616497039795, | |
| "logits/rejected": -3.162132740020752, | |
| "logps/chosen": -268.2223205566406, | |
| "logps/rejected": -237.77635192871094, | |
| "loss": 0.5352, | |
| "rewards/accuracies": 0.7364583611488342, | |
| "rewards/chosen": -0.19809262454509735, | |
| "rewards/margins": 0.6094833016395569, | |
| "rewards/rejected": -0.8075758814811707, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.9448340345956054, | |
| "eval_logits/chosen": -3.3784019947052, | |
| "eval_logits/rejected": -3.116711378097534, | |
| "eval_logps/chosen": -274.5598449707031, | |
| "eval_logps/rejected": -237.6013946533203, | |
| "eval_loss": 0.6299869418144226, | |
| "eval_rewards/accuracies": 0.6327599287033081, | |
| "eval_rewards/chosen": -0.2677817940711975, | |
| "eval_rewards/margins": 0.38533419370651245, | |
| "eval_rewards/rejected": -0.6531160473823547, | |
| "eval_runtime": 1688.4156, | |
| "eval_samples_per_second": 3.167, | |
| "eval_steps_per_second": 3.167, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.9747545582047685, | |
| "grad_norm": 6.9590959548950195, | |
| "learning_rate": 3.413173652694611e-05, | |
| "logits/chosen": -3.376148223876953, | |
| "logits/rejected": -3.1122324466705322, | |
| "logps/chosen": -282.5127258300781, | |
| "logps/rejected": -247.69212341308594, | |
| "loss": 0.5232, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -0.10552702099084854, | |
| "rewards/margins": 0.6293079257011414, | |
| "rewards/rejected": -0.7348350286483765, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.0196353436185133, | |
| "grad_norm": 5.992002010345459, | |
| "learning_rate": 3.263473053892216e-05, | |
| "logits/chosen": -3.397113800048828, | |
| "logits/rejected": -3.136545419692993, | |
| "logps/chosen": -278.75390625, | |
| "logps/rejected": -246.4496307373047, | |
| "loss": 0.5015, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.11011376976966858, | |
| "rewards/margins": 0.6908566355705261, | |
| "rewards/rejected": -0.8009704351425171, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.064516129032258, | |
| "grad_norm": 7.003468990325928, | |
| "learning_rate": 3.1137724550898205e-05, | |
| "logits/chosen": -3.370246410369873, | |
| "logits/rejected": -3.093019723892212, | |
| "logps/chosen": -279.3021240234375, | |
| "logps/rejected": -242.58258056640625, | |
| "loss": 0.446, | |
| "rewards/accuracies": 0.8114583492279053, | |
| "rewards/chosen": -0.05044478550553322, | |
| "rewards/margins": 0.8847902417182922, | |
| "rewards/rejected": -0.9352350234985352, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.0944366526414213, | |
| "eval_logits/chosen": -3.3703560829162598, | |
| "eval_logits/rejected": -3.111078977584839, | |
| "eval_logps/chosen": -274.6524353027344, | |
| "eval_logps/rejected": -237.6984405517578, | |
| "eval_loss": 0.6312919855117798, | |
| "eval_rewards/accuracies": 0.6325729489326477, | |
| "eval_rewards/chosen": -0.27703869342803955, | |
| "eval_rewards/margins": 0.38578376173973083, | |
| "eval_rewards/rejected": -0.6628224849700928, | |
| "eval_runtime": 1685.7321, | |
| "eval_samples_per_second": 3.173, | |
| "eval_steps_per_second": 3.173, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.1093969144460027, | |
| "grad_norm": 5.607975959777832, | |
| "learning_rate": 2.9640718562874252e-05, | |
| "logits/chosen": -3.365170955657959, | |
| "logits/rejected": -3.1272387504577637, | |
| "logps/chosen": -271.7861022949219, | |
| "logps/rejected": -242.59573364257812, | |
| "loss": 0.4698, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.10354464501142502, | |
| "rewards/margins": 0.7553961277008057, | |
| "rewards/rejected": -0.8589407801628113, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.1542776998597475, | |
| "grad_norm": 5.758065223693848, | |
| "learning_rate": 2.81437125748503e-05, | |
| "logits/chosen": -3.362076997756958, | |
| "logits/rejected": -3.1057257652282715, | |
| "logps/chosen": -267.41705322265625, | |
| "logps/rejected": -235.0531463623047, | |
| "loss": 0.4586, | |
| "rewards/accuracies": 0.8166666626930237, | |
| "rewards/chosen": -0.11131696403026581, | |
| "rewards/margins": 0.785390317440033, | |
| "rewards/rejected": -0.89670729637146, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.1991584852734922, | |
| "grad_norm": 6.8294501304626465, | |
| "learning_rate": 2.6646706586826347e-05, | |
| "logits/chosen": -3.368708610534668, | |
| "logits/rejected": -3.0964319705963135, | |
| "logps/chosen": -269.26409912109375, | |
| "logps/rejected": -237.1984405517578, | |
| "loss": 0.4496, | |
| "rewards/accuracies": 0.8145833611488342, | |
| "rewards/chosen": -0.10699882358312607, | |
| "rewards/margins": 0.8314520120620728, | |
| "rewards/rejected": -0.938450813293457, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.244039270687237, | |
| "grad_norm": 6.356990337371826, | |
| "learning_rate": 2.5149700598802394e-05, | |
| "logits/chosen": -3.374453067779541, | |
| "logits/rejected": -3.129500389099121, | |
| "logps/chosen": -271.7542724609375, | |
| "logps/rejected": -241.45423889160156, | |
| "loss": 0.4552, | |
| "rewards/accuracies": 0.8135416507720947, | |
| "rewards/chosen": -0.20287248492240906, | |
| "rewards/margins": 0.8104608058929443, | |
| "rewards/rejected": -1.0133334398269653, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.244039270687237, | |
| "eval_logits/chosen": -3.360283613204956, | |
| "eval_logits/rejected": -3.1080915927886963, | |
| "eval_logps/chosen": -276.3040466308594, | |
| "eval_logps/rejected": -239.7833251953125, | |
| "eval_loss": 0.6368128657341003, | |
| "eval_rewards/accuracies": 0.6351907253265381, | |
| "eval_rewards/chosen": -0.44220101833343506, | |
| "eval_rewards/margins": 0.4291093647480011, | |
| "eval_rewards/rejected": -0.871310293674469, | |
| "eval_runtime": 1686.3289, | |
| "eval_samples_per_second": 3.171, | |
| "eval_steps_per_second": 3.171, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.2889200561009817, | |
| "grad_norm": 6.016663551330566, | |
| "learning_rate": 2.3652694610778445e-05, | |
| "logits/chosen": -3.3569111824035645, | |
| "logits/rejected": -3.123525857925415, | |
| "logps/chosen": -274.6582946777344, | |
| "logps/rejected": -241.02252197265625, | |
| "loss": 0.4577, | |
| "rewards/accuracies": 0.7947916388511658, | |
| "rewards/chosen": -0.20317865908145905, | |
| "rewards/margins": 0.8162151575088501, | |
| "rewards/rejected": -1.0193939208984375, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.3338008415147264, | |
| "grad_norm": 5.684780120849609, | |
| "learning_rate": 2.2155688622754492e-05, | |
| "logits/chosen": -3.3533644676208496, | |
| "logits/rejected": -3.146190881729126, | |
| "logps/chosen": -271.4990234375, | |
| "logps/rejected": -242.20486450195312, | |
| "loss": 0.4674, | |
| "rewards/accuracies": 0.7989583611488342, | |
| "rewards/chosen": -0.12297000735998154, | |
| "rewards/margins": 0.8095114827156067, | |
| "rewards/rejected": -0.9324816465377808, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.378681626928471, | |
| "grad_norm": 7.419367790222168, | |
| "learning_rate": 2.065868263473054e-05, | |
| "logits/chosen": -3.364116907119751, | |
| "logits/rejected": -3.092254638671875, | |
| "logps/chosen": -270.5090026855469, | |
| "logps/rejected": -237.64964294433594, | |
| "loss": 0.4443, | |
| "rewards/accuracies": 0.8177083134651184, | |
| "rewards/chosen": -0.22570447623729706, | |
| "rewards/margins": 0.84433513879776, | |
| "rewards/rejected": -1.0700395107269287, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.393641888733053, | |
| "eval_logits/chosen": -3.354207992553711, | |
| "eval_logits/rejected": -3.103837013244629, | |
| "eval_logps/chosen": -276.166015625, | |
| "eval_logps/rejected": -239.59542846679688, | |
| "eval_loss": 0.6390828490257263, | |
| "eval_rewards/accuracies": 0.6344428062438965, | |
| "eval_rewards/chosen": -0.4283973276615143, | |
| "eval_rewards/margins": 0.42412257194519043, | |
| "eval_rewards/rejected": -0.8525198101997375, | |
| "eval_runtime": 1685.5337, | |
| "eval_samples_per_second": 3.173, | |
| "eval_steps_per_second": 3.173, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.423562412342216, | |
| "grad_norm": 7.774267196655273, | |
| "learning_rate": 1.916167664670659e-05, | |
| "logits/chosen": -3.355332851409912, | |
| "logits/rejected": -3.1059510707855225, | |
| "logps/chosen": -277.3658752441406, | |
| "logps/rejected": -247.025146484375, | |
| "loss": 0.4466, | |
| "rewards/accuracies": 0.8052083253860474, | |
| "rewards/chosen": -0.20841935276985168, | |
| "rewards/margins": 0.860171377658844, | |
| "rewards/rejected": -1.0685906410217285, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.4684431977559607, | |
| "grad_norm": 7.505038738250732, | |
| "learning_rate": 1.7664670658682637e-05, | |
| "logits/chosen": -3.345045804977417, | |
| "logits/rejected": -3.1149702072143555, | |
| "logps/chosen": -278.6654968261719, | |
| "logps/rejected": -250.63827514648438, | |
| "loss": 0.4452, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -0.24601925909519196, | |
| "rewards/margins": 0.8955973982810974, | |
| "rewards/rejected": -1.141616702079773, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.5133239831697054, | |
| "grad_norm": 7.055671215057373, | |
| "learning_rate": 1.6167664670658684e-05, | |
| "logits/chosen": -3.3556442260742188, | |
| "logits/rejected": -3.086568593978882, | |
| "logps/chosen": -275.1831359863281, | |
| "logps/rejected": -240.82598876953125, | |
| "loss": 0.4564, | |
| "rewards/accuracies": 0.8052083253860474, | |
| "rewards/chosen": -0.29922303557395935, | |
| "rewards/margins": 0.8765833377838135, | |
| "rewards/rejected": -1.1758064031600952, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.5432445067788687, | |
| "eval_logits/chosen": -3.349193811416626, | |
| "eval_logits/rejected": -3.098674774169922, | |
| "eval_logps/chosen": -277.01483154296875, | |
| "eval_logps/rejected": -240.51805114746094, | |
| "eval_loss": 0.642005980014801, | |
| "eval_rewards/accuracies": 0.6297681331634521, | |
| "eval_rewards/chosen": -0.5132736563682556, | |
| "eval_rewards/margins": 0.4315095543861389, | |
| "eval_rewards/rejected": -0.9447831511497498, | |
| "eval_runtime": 1686.0549, | |
| "eval_samples_per_second": 3.172, | |
| "eval_steps_per_second": 3.172, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.55820476858345, | |
| "grad_norm": 7.247310638427734, | |
| "learning_rate": 1.467065868263473e-05, | |
| "logits/chosen": -3.3303916454315186, | |
| "logits/rejected": -3.118966817855835, | |
| "logps/chosen": -276.04510498046875, | |
| "logps/rejected": -250.57984924316406, | |
| "loss": 0.4615, | |
| "rewards/accuracies": 0.8072916865348816, | |
| "rewards/chosen": -0.27220281958580017, | |
| "rewards/margins": 0.8313066363334656, | |
| "rewards/rejected": -1.103509545326233, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.603085553997195, | |
| "grad_norm": 6.719433784484863, | |
| "learning_rate": 1.317365269461078e-05, | |
| "logits/chosen": -3.3551132678985596, | |
| "logits/rejected": -3.1186678409576416, | |
| "logps/chosen": -277.4861755371094, | |
| "logps/rejected": -251.39437866210938, | |
| "loss": 0.455, | |
| "rewards/accuracies": 0.8031250238418579, | |
| "rewards/chosen": -0.23053057491779327, | |
| "rewards/margins": 0.8569380640983582, | |
| "rewards/rejected": -1.0874686241149902, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.6479663394109396, | |
| "grad_norm": 6.049899101257324, | |
| "learning_rate": 1.1676646706586828e-05, | |
| "logits/chosen": -3.3462002277374268, | |
| "logits/rejected": -3.0950281620025635, | |
| "logps/chosen": -279.08447265625, | |
| "logps/rejected": -243.8069305419922, | |
| "loss": 0.4414, | |
| "rewards/accuracies": 0.8072916865348816, | |
| "rewards/chosen": -0.24068358540534973, | |
| "rewards/margins": 0.8998420238494873, | |
| "rewards/rejected": -1.1405255794525146, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.6928471248246844, | |
| "grad_norm": 7.545809268951416, | |
| "learning_rate": 1.0179640718562875e-05, | |
| "logits/chosen": -3.346256732940674, | |
| "logits/rejected": -3.112372875213623, | |
| "logps/chosen": -270.18499755859375, | |
| "logps/rejected": -240.69723510742188, | |
| "loss": 0.4603, | |
| "rewards/accuracies": 0.8083333373069763, | |
| "rewards/chosen": -0.25966814160346985, | |
| "rewards/margins": 0.8120385408401489, | |
| "rewards/rejected": -1.071706771850586, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.6928471248246844, | |
| "eval_logits/chosen": -3.3438971042633057, | |
| "eval_logits/rejected": -3.0931475162506104, | |
| "eval_logps/chosen": -276.7391662597656, | |
| "eval_logps/rejected": -240.1473388671875, | |
| "eval_loss": 0.6427502036094666, | |
| "eval_rewards/accuracies": 0.6297681331634521, | |
| "eval_rewards/chosen": -0.48571139574050903, | |
| "eval_rewards/margins": 0.4220017194747925, | |
| "eval_rewards/rejected": -0.9077131152153015, | |
| "eval_runtime": 1686.025, | |
| "eval_samples_per_second": 3.172, | |
| "eval_steps_per_second": 3.172, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.737727910238429, | |
| "grad_norm": 5.611355304718018, | |
| "learning_rate": 8.682634730538922e-06, | |
| "logits/chosen": -3.347557306289673, | |
| "logits/rejected": -3.109966993331909, | |
| "logps/chosen": -275.6930236816406, | |
| "logps/rejected": -247.47406005859375, | |
| "loss": 0.4457, | |
| "rewards/accuracies": 0.8291666507720947, | |
| "rewards/chosen": -0.3010416030883789, | |
| "rewards/margins": 0.8664290308952332, | |
| "rewards/rejected": -1.1674706935882568, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.782608695652174, | |
| "grad_norm": 8.53209114074707, | |
| "learning_rate": 7.18562874251497e-06, | |
| "logits/chosen": -3.3400204181671143, | |
| "logits/rejected": -3.103865623474121, | |
| "logps/chosen": -285.2029724121094, | |
| "logps/rejected": -255.09078979492188, | |
| "loss": 0.4524, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.22924675047397614, | |
| "rewards/margins": 0.8349610567092896, | |
| "rewards/rejected": -1.0642077922821045, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.8274894810659186, | |
| "grad_norm": 7.011772155761719, | |
| "learning_rate": 5.688622754491018e-06, | |
| "logits/chosen": -3.3375208377838135, | |
| "logits/rejected": -3.0882365703582764, | |
| "logps/chosen": -269.7694091796875, | |
| "logps/rejected": -238.83042907714844, | |
| "loss": 0.4511, | |
| "rewards/accuracies": 0.8031250238418579, | |
| "rewards/chosen": -0.3288494944572449, | |
| "rewards/margins": 0.8713601231575012, | |
| "rewards/rejected": -1.2002094984054565, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.8424497428705005, | |
| "eval_logits/chosen": -3.3421826362609863, | |
| "eval_logits/rejected": -3.0923619270324707, | |
| "eval_logps/chosen": -277.226806640625, | |
| "eval_logps/rejected": -240.6798858642578, | |
| "eval_loss": 0.6432516574859619, | |
| "eval_rewards/accuracies": 0.6295811533927917, | |
| "eval_rewards/chosen": -0.5344744324684143, | |
| "eval_rewards/margins": 0.42649218440055847, | |
| "eval_rewards/rejected": -0.9609667062759399, | |
| "eval_runtime": 1686.3389, | |
| "eval_samples_per_second": 3.171, | |
| "eval_steps_per_second": 3.171, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.8723702664796633, | |
| "grad_norm": 7.099593162536621, | |
| "learning_rate": 4.191616766467066e-06, | |
| "logits/chosen": -3.359609365463257, | |
| "logits/rejected": -3.0945444107055664, | |
| "logps/chosen": -280.75030517578125, | |
| "logps/rejected": -245.13504028320312, | |
| "loss": 0.4418, | |
| "rewards/accuracies": 0.8197916746139526, | |
| "rewards/chosen": -0.30151474475860596, | |
| "rewards/margins": 0.8953721523284912, | |
| "rewards/rejected": -1.1968867778778076, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.917251051893408, | |
| "grad_norm": 7.788060188293457, | |
| "learning_rate": 2.6946107784431138e-06, | |
| "logits/chosen": -3.3403496742248535, | |
| "logits/rejected": -3.091184139251709, | |
| "logps/chosen": -280.9390869140625, | |
| "logps/rejected": -247.351806640625, | |
| "loss": 0.444, | |
| "rewards/accuracies": 0.8302083611488342, | |
| "rewards/chosen": -0.2591714859008789, | |
| "rewards/margins": 0.885311484336853, | |
| "rewards/rejected": -1.1444830894470215, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.962131837307153, | |
| "grad_norm": 7.973437786102295, | |
| "learning_rate": 1.1976047904191619e-06, | |
| "logits/chosen": -3.328507900238037, | |
| "logits/rejected": -3.088214635848999, | |
| "logps/chosen": -271.0534362792969, | |
| "logps/rejected": -242.70079040527344, | |
| "loss": 0.4531, | |
| "rewards/accuracies": 0.815625011920929, | |
| "rewards/chosen": -0.357342392206192, | |
| "rewards/margins": 0.865265429019928, | |
| "rewards/rejected": -1.2226077318191528, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.992052360916316, | |
| "eval_logits/chosen": -3.342743158340454, | |
| "eval_logits/rejected": -3.0929155349731445, | |
| "eval_logps/chosen": -277.3058166503906, | |
| "eval_logps/rejected": -240.79270935058594, | |
| "eval_loss": 0.6429719924926758, | |
| "eval_rewards/accuracies": 0.6299551129341125, | |
| "eval_rewards/chosen": -0.5423800349235535, | |
| "eval_rewards/margins": 0.42986956238746643, | |
| "eval_rewards/rejected": -0.9722495079040527, | |
| "eval_runtime": 1684.4785, | |
| "eval_samples_per_second": 3.175, | |
| "eval_steps_per_second": 3.175, | |
| "step": 2000 | |
| } | |
| ], | |
| "logging_steps": 30, | |
| "max_steps": 2004, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |