|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.9986120749479528, |
|
"eval_steps": 1000, |
|
"global_step": 1440, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0013879250520471894, |
|
"grad_norm": 7.310771886732375, |
|
"learning_rate": 3.4722222222222217e-09, |
|
"logits/chosen": -2.658149242401123, |
|
"logits/rejected": -2.6729652881622314, |
|
"logps/chosen": -310.6693115234375, |
|
"logps/rejected": -336.3360595703125, |
|
"loss": 0.6931, |
|
"rewards/accuracies": 0.0, |
|
"rewards/chosen": 0.0, |
|
"rewards/margins": 0.0, |
|
"rewards/rejected": 0.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.013879250520471894, |
|
"grad_norm": 6.989456718606319, |
|
"learning_rate": 3.472222222222222e-08, |
|
"logits/chosen": -2.7207493782043457, |
|
"logits/rejected": -2.678452968597412, |
|
"logps/chosen": -329.1288146972656, |
|
"logps/rejected": -334.9566650390625, |
|
"loss": 0.6932, |
|
"rewards/accuracies": 0.4236111044883728, |
|
"rewards/chosen": -0.00011655675189103931, |
|
"rewards/margins": -0.00021387077867984772, |
|
"rewards/rejected": 9.731399040902033e-05, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.027758501040943788, |
|
"grad_norm": 7.28858517103727, |
|
"learning_rate": 6.944444444444444e-08, |
|
"logits/chosen": -2.690325975418091, |
|
"logits/rejected": -2.6607680320739746, |
|
"logps/chosen": -317.3731384277344, |
|
"logps/rejected": -331.0643005371094, |
|
"loss": 0.6928, |
|
"rewards/accuracies": 0.5375000238418579, |
|
"rewards/chosen": 0.00018671144789550453, |
|
"rewards/margins": 0.00033597589936107397, |
|
"rewards/rejected": -0.00014926446601748466, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.041637751561415685, |
|
"grad_norm": 7.1293734693123785, |
|
"learning_rate": 1.0416666666666667e-07, |
|
"logits/chosen": -2.6790878772735596, |
|
"logits/rejected": -2.645470142364502, |
|
"logps/chosen": -351.1817321777344, |
|
"logps/rejected": -351.5977478027344, |
|
"loss": 0.6914, |
|
"rewards/accuracies": 0.690625011920929, |
|
"rewards/chosen": 0.0023157999385148287, |
|
"rewards/margins": 0.004029616713523865, |
|
"rewards/rejected": -0.0017138172406703234, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.055517002081887576, |
|
"grad_norm": 7.102945109434136, |
|
"learning_rate": 1.3888888888888888e-07, |
|
"logits/chosen": -2.698319435119629, |
|
"logits/rejected": -2.6377086639404297, |
|
"logps/chosen": -352.13153076171875, |
|
"logps/rejected": -337.7492370605469, |
|
"loss": 0.6867, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": 0.010404938831925392, |
|
"rewards/margins": 0.012554061599075794, |
|
"rewards/rejected": -0.0021491218358278275, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.06939625260235947, |
|
"grad_norm": 6.472817798536912, |
|
"learning_rate": 1.736111111111111e-07, |
|
"logits/chosen": -2.6700968742370605, |
|
"logits/rejected": -2.656541347503662, |
|
"logps/chosen": -325.9363708496094, |
|
"logps/rejected": -359.7778625488281, |
|
"loss": 0.6778, |
|
"rewards/accuracies": 0.7906249761581421, |
|
"rewards/chosen": 0.026792842894792557, |
|
"rewards/margins": 0.031667567789554596, |
|
"rewards/rejected": -0.004874727688729763, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.08327550312283137, |
|
"grad_norm": 7.812961434750235, |
|
"learning_rate": 2.0833333333333333e-07, |
|
"logits/chosen": -2.7250404357910156, |
|
"logits/rejected": -2.680410861968994, |
|
"logps/chosen": -325.12628173828125, |
|
"logps/rejected": -345.4599914550781, |
|
"loss": 0.6575, |
|
"rewards/accuracies": 0.815625011920929, |
|
"rewards/chosen": 0.05238135904073715, |
|
"rewards/margins": 0.06940947473049164, |
|
"rewards/rejected": -0.017028113827109337, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.09715475364330327, |
|
"grad_norm": 8.252964521813333, |
|
"learning_rate": 2.4305555555555555e-07, |
|
"logits/chosen": -2.6680197715759277, |
|
"logits/rejected": -2.6488897800445557, |
|
"logps/chosen": -336.755615234375, |
|
"logps/rejected": -372.3600158691406, |
|
"loss": 0.6231, |
|
"rewards/accuracies": 0.809374988079071, |
|
"rewards/chosen": 0.04352443665266037, |
|
"rewards/margins": 0.14054368436336517, |
|
"rewards/rejected": -0.0970192551612854, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.11103400416377515, |
|
"grad_norm": 10.907322040862939, |
|
"learning_rate": 2.7777777777777776e-07, |
|
"logits/chosen": -2.6945643424987793, |
|
"logits/rejected": -2.635817050933838, |
|
"logps/chosen": -345.4059143066406, |
|
"logps/rejected": -395.4034423828125, |
|
"loss": 0.5649, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": -0.05337335914373398, |
|
"rewards/margins": 0.348626971244812, |
|
"rewards/rejected": -0.402000367641449, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.12491325468424705, |
|
"grad_norm": 14.954457613666033, |
|
"learning_rate": 3.1249999999999997e-07, |
|
"logits/chosen": -2.7117972373962402, |
|
"logits/rejected": -2.659916877746582, |
|
"logps/chosen": -357.34344482421875, |
|
"logps/rejected": -424.046875, |
|
"loss": 0.4833, |
|
"rewards/accuracies": 0.815625011920929, |
|
"rewards/chosen": -0.24065232276916504, |
|
"rewards/margins": 0.6489619016647339, |
|
"rewards/rejected": -0.8896142840385437, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.13879250520471895, |
|
"grad_norm": 18.8809663980661, |
|
"learning_rate": 3.472222222222222e-07, |
|
"logits/chosen": -2.691723108291626, |
|
"logits/rejected": -2.676143169403076, |
|
"logps/chosen": -429.1051330566406, |
|
"logps/rejected": -508.2496032714844, |
|
"loss": 0.4648, |
|
"rewards/accuracies": 0.8062499761581421, |
|
"rewards/chosen": -0.836562991142273, |
|
"rewards/margins": 0.7794925570487976, |
|
"rewards/rejected": -1.6160557270050049, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.15267175572519084, |
|
"grad_norm": 17.153622207375392, |
|
"learning_rate": 3.819444444444444e-07, |
|
"logits/chosen": -2.7029433250427246, |
|
"logits/rejected": -2.6785738468170166, |
|
"logps/chosen": -463.09588623046875, |
|
"logps/rejected": -551.4292602539062, |
|
"loss": 0.4393, |
|
"rewards/accuracies": 0.78125, |
|
"rewards/chosen": -1.1502994298934937, |
|
"rewards/margins": 0.9367601275444031, |
|
"rewards/rejected": -2.087059736251831, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.16655100624566274, |
|
"grad_norm": 17.322428984395128, |
|
"learning_rate": 4.1666666666666667e-07, |
|
"logits/chosen": -2.740248203277588, |
|
"logits/rejected": -2.715928554534912, |
|
"logps/chosen": -466.61614990234375, |
|
"logps/rejected": -588.0739135742188, |
|
"loss": 0.416, |
|
"rewards/accuracies": 0.746874988079071, |
|
"rewards/chosen": -1.314789891242981, |
|
"rewards/margins": 1.0913175344467163, |
|
"rewards/rejected": -2.4061074256896973, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.18043025676613464, |
|
"grad_norm": 17.727320951964273, |
|
"learning_rate": 4.513888888888889e-07, |
|
"logits/chosen": -2.732884168624878, |
|
"logits/rejected": -2.693039894104004, |
|
"logps/chosen": -470.36712646484375, |
|
"logps/rejected": -637.1500854492188, |
|
"loss": 0.3922, |
|
"rewards/accuracies": 0.831250011920929, |
|
"rewards/chosen": -1.2835638523101807, |
|
"rewards/margins": 1.5643529891967773, |
|
"rewards/rejected": -2.847916841506958, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.19430950728660654, |
|
"grad_norm": 23.308808504878304, |
|
"learning_rate": 4.861111111111111e-07, |
|
"logits/chosen": -2.410512924194336, |
|
"logits/rejected": -2.2582736015319824, |
|
"logps/chosen": -466.9541931152344, |
|
"logps/rejected": -714.3003540039062, |
|
"loss": 0.355, |
|
"rewards/accuracies": 0.846875011920929, |
|
"rewards/chosen": -1.4372889995574951, |
|
"rewards/margins": 2.2647476196289062, |
|
"rewards/rejected": -3.7020363807678223, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.2081887578070784, |
|
"grad_norm": 15.792878048667099, |
|
"learning_rate": 4.999735579817769e-07, |
|
"logits/chosen": -1.8029206991195679, |
|
"logits/rejected": -1.4160174131393433, |
|
"logps/chosen": -489.02960205078125, |
|
"logps/rejected": -732.9075927734375, |
|
"loss": 0.3505, |
|
"rewards/accuracies": 0.856249988079071, |
|
"rewards/chosen": -1.6447805166244507, |
|
"rewards/margins": 2.2803826332092285, |
|
"rewards/rejected": -3.9251632690429688, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.2220680083275503, |
|
"grad_norm": 27.767234596146235, |
|
"learning_rate": 4.998119881260575e-07, |
|
"logits/chosen": -1.729731559753418, |
|
"logits/rejected": -1.0307753086090088, |
|
"logps/chosen": -489.806396484375, |
|
"logps/rejected": -750.17333984375, |
|
"loss": 0.3419, |
|
"rewards/accuracies": 0.809374988079071, |
|
"rewards/chosen": -1.7482877969741821, |
|
"rewards/margins": 2.432042121887207, |
|
"rewards/rejected": -4.1803297996521, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.2359472588480222, |
|
"grad_norm": 16.170525819937215, |
|
"learning_rate": 4.995036332451857e-07, |
|
"logits/chosen": -1.9091663360595703, |
|
"logits/rejected": -0.9273399114608765, |
|
"logps/chosen": -495.46661376953125, |
|
"logps/rejected": -758.0303955078125, |
|
"loss": 0.3289, |
|
"rewards/accuracies": 0.831250011920929, |
|
"rewards/chosen": -1.4501034021377563, |
|
"rewards/margins": 2.5703396797180176, |
|
"rewards/rejected": -4.020443439483643, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.2498265093684941, |
|
"grad_norm": 19.933464291596422, |
|
"learning_rate": 4.990486745229364e-07, |
|
"logits/chosen": -1.4646329879760742, |
|
"logits/rejected": -0.523992121219635, |
|
"logps/chosen": -475.5484924316406, |
|
"logps/rejected": -712.8101806640625, |
|
"loss": 0.3238, |
|
"rewards/accuracies": 0.8218749761581421, |
|
"rewards/chosen": -1.4877039194107056, |
|
"rewards/margins": 2.2666311264038086, |
|
"rewards/rejected": -3.7543349266052246, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.263705759888966, |
|
"grad_norm": 13.788976405207086, |
|
"learning_rate": 4.984473792848607e-07, |
|
"logits/chosen": -1.0008184909820557, |
|
"logits/rejected": 0.3863092064857483, |
|
"logps/chosen": -459.58367919921875, |
|
"logps/rejected": -734.6851196289062, |
|
"loss": 0.3131, |
|
"rewards/accuracies": 0.84375, |
|
"rewards/chosen": -1.3406341075897217, |
|
"rewards/margins": 2.7795493602752686, |
|
"rewards/rejected": -4.120182991027832, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.2775850104094379, |
|
"grad_norm": 20.384550977715232, |
|
"learning_rate": 4.977001008412112e-07, |
|
"logits/chosen": -0.7814763188362122, |
|
"logits/rejected": 0.5796900987625122, |
|
"logps/chosen": -522.8018798828125, |
|
"logps/rejected": -799.6663208007812, |
|
"loss": 0.333, |
|
"rewards/accuracies": 0.828125, |
|
"rewards/chosen": -1.9407739639282227, |
|
"rewards/margins": 2.682741165161133, |
|
"rewards/rejected": -4.6235151290893555, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.2914642609299098, |
|
"grad_norm": 18.232716128762664, |
|
"learning_rate": 4.968072782793435e-07, |
|
"logits/chosen": -0.9184268116950989, |
|
"logits/rejected": 0.4480651319026947, |
|
"logps/chosen": -499.62127685546875, |
|
"logps/rejected": -834.462890625, |
|
"loss": 0.3089, |
|
"rewards/accuracies": 0.846875011920929, |
|
"rewards/chosen": -1.7784448862075806, |
|
"rewards/margins": 3.2359156608581543, |
|
"rewards/rejected": -5.0143609046936035, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.3053435114503817, |
|
"grad_norm": 19.076527739229473, |
|
"learning_rate": 4.957694362057149e-07, |
|
"logits/chosen": -0.9672171473503113, |
|
"logits/rejected": 0.16848711669445038, |
|
"logps/chosen": -560.64453125, |
|
"logps/rejected": -855.81591796875, |
|
"loss": 0.3149, |
|
"rewards/accuracies": 0.846875011920929, |
|
"rewards/chosen": -1.9354803562164307, |
|
"rewards/margins": 3.0875911712646484, |
|
"rewards/rejected": -5.0230712890625, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.3192227619708536, |
|
"grad_norm": 16.258599026908875, |
|
"learning_rate": 4.945871844376368e-07, |
|
"logits/chosen": -1.0649207830429077, |
|
"logits/rejected": 0.31253132224082947, |
|
"logps/chosen": -538.4542846679688, |
|
"logps/rejected": -874.3626098632812, |
|
"loss": 0.3188, |
|
"rewards/accuracies": 0.8968750238418579, |
|
"rewards/chosen": -1.9632246494293213, |
|
"rewards/margins": 3.402304172515869, |
|
"rewards/rejected": -5.365528583526611, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.3331020124913255, |
|
"grad_norm": 18.705612187457067, |
|
"learning_rate": 4.932612176449559e-07, |
|
"logits/chosen": -0.9302694201469421, |
|
"logits/rejected": 0.36828285455703735, |
|
"logps/chosen": -505.27606201171875, |
|
"logps/rejected": -816.0523681640625, |
|
"loss": 0.3025, |
|
"rewards/accuracies": 0.859375, |
|
"rewards/chosen": -1.567478895187378, |
|
"rewards/margins": 2.9575300216674805, |
|
"rewards/rejected": -4.525008678436279, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.3469812630117974, |
|
"grad_norm": 24.629743659903227, |
|
"learning_rate": 4.917923149418791e-07, |
|
"logits/chosen": 0.6533899307250977, |
|
"logits/rejected": 1.9370393753051758, |
|
"logps/chosen": -599.1996459960938, |
|
"logps/rejected": -975.5636596679688, |
|
"loss": 0.2998, |
|
"rewards/accuracies": 0.84375, |
|
"rewards/chosen": -2.498326301574707, |
|
"rewards/margins": 3.716355562210083, |
|
"rewards/rejected": -6.214681625366211, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.3608605135322693, |
|
"grad_norm": 17.05308334694298, |
|
"learning_rate": 4.901813394291801e-07, |
|
"logits/chosen": -0.2267475426197052, |
|
"logits/rejected": 1.061628818511963, |
|
"logps/chosen": -510.58990478515625, |
|
"logps/rejected": -825.64892578125, |
|
"loss": 0.3068, |
|
"rewards/accuracies": 0.8343750238418579, |
|
"rewards/chosen": -1.8009761571884155, |
|
"rewards/margins": 3.0452382564544678, |
|
"rewards/rejected": -4.846214771270752, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.3747397640527412, |
|
"grad_norm": 16.458651892780324, |
|
"learning_rate": 4.884292376870567e-07, |
|
"logits/chosen": -0.5495095252990723, |
|
"logits/rejected": 0.9298044443130493, |
|
"logps/chosen": -518.9110107421875, |
|
"logps/rejected": -838.7022705078125, |
|
"loss": 0.3019, |
|
"rewards/accuracies": 0.8218749761581421, |
|
"rewards/chosen": -1.7417328357696533, |
|
"rewards/margins": 3.202885866165161, |
|
"rewards/rejected": -4.944618225097656, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.3886190145732131, |
|
"grad_norm": 17.412372040942056, |
|
"learning_rate": 4.865370392189376e-07, |
|
"logits/chosen": -1.044090986251831, |
|
"logits/rejected": 0.6990992426872253, |
|
"logps/chosen": -521.0147094726562, |
|
"logps/rejected": -908.5431518554688, |
|
"loss": 0.3086, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": -1.7731910943984985, |
|
"rewards/margins": 3.855863571166992, |
|
"rewards/rejected": -5.629055023193359, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.4024982650936849, |
|
"grad_norm": 18.962541699001296, |
|
"learning_rate": 4.845058558465645e-07, |
|
"logits/chosen": 0.2687569260597229, |
|
"logits/rejected": 1.7490098476409912, |
|
"logps/chosen": -562.8821411132812, |
|
"logps/rejected": -911.21142578125, |
|
"loss": 0.2941, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": -2.1526246070861816, |
|
"rewards/margins": 3.554647445678711, |
|
"rewards/rejected": -5.707272052764893, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.4163775156141568, |
|
"grad_norm": 18.283817174579646, |
|
"learning_rate": 4.823368810567056e-07, |
|
"logits/chosen": 0.27320951223373413, |
|
"logits/rejected": 1.6727936267852783, |
|
"logps/chosen": -508.94744873046875, |
|
"logps/rejected": -807.9446411132812, |
|
"loss": 0.3081, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": -1.861353874206543, |
|
"rewards/margins": 3.0898642539978027, |
|
"rewards/rejected": -4.951218605041504, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.4302567661346287, |
|
"grad_norm": 15.422967764865048, |
|
"learning_rate": 4.800313892998847e-07, |
|
"logits/chosen": -0.3293236196041107, |
|
"logits/rejected": 1.3189566135406494, |
|
"logps/chosen": -509.6499938964844, |
|
"logps/rejected": -856.3865966796875, |
|
"loss": 0.2919, |
|
"rewards/accuracies": 0.856249988079071, |
|
"rewards/chosen": -1.683760404586792, |
|
"rewards/margins": 3.3120670318603516, |
|
"rewards/rejected": -4.995827674865723, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.4441360166551006, |
|
"grad_norm": 18.804961569512887, |
|
"learning_rate": 4.775907352415367e-07, |
|
"logits/chosen": 0.22868101298809052, |
|
"logits/rejected": 1.9417445659637451, |
|
"logps/chosen": -556.9364013671875, |
|
"logps/rejected": -950.5734252929688, |
|
"loss": 0.2808, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -1.9100916385650635, |
|
"rewards/margins": 4.022894859313965, |
|
"rewards/rejected": -5.932986259460449, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.4580152671755725, |
|
"grad_norm": 17.099019764148846, |
|
"learning_rate": 4.7501635296603025e-07, |
|
"logits/chosen": 1.049659013748169, |
|
"logits/rejected": 2.4073307514190674, |
|
"logps/chosen": -531.3202514648438, |
|
"logps/rejected": -922.6915893554688, |
|
"loss": 0.292, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -1.8814361095428467, |
|
"rewards/margins": 3.822434902191162, |
|
"rewards/rejected": -5.7038702964782715, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.4718945176960444, |
|
"grad_norm": 18.898740164063316, |
|
"learning_rate": 4.723097551340265e-07, |
|
"logits/chosen": 1.2799204587936401, |
|
"logits/rejected": 2.3121683597564697, |
|
"logps/chosen": -494.0704040527344, |
|
"logps/rejected": -854.8629760742188, |
|
"loss": 0.292, |
|
"rewards/accuracies": 0.8656250238418579, |
|
"rewards/chosen": -1.7092489004135132, |
|
"rewards/margins": 3.4398162364959717, |
|
"rewards/rejected": -5.149065971374512, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.4857737682165163, |
|
"grad_norm": 20.5834993963944, |
|
"learning_rate": 4.6947253209366613e-07, |
|
"logits/chosen": 1.0228136777877808, |
|
"logits/rejected": 2.397416114807129, |
|
"logps/chosen": -545.7423706054688, |
|
"logps/rejected": -912.7175903320312, |
|
"loss": 0.2827, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -2.025628089904785, |
|
"rewards/margins": 3.6154162883758545, |
|
"rewards/rejected": -5.6410441398620605, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.4996530187369882, |
|
"grad_norm": 14.253483616771307, |
|
"learning_rate": 4.6650635094610966e-07, |
|
"logits/chosen": 1.0244512557983398, |
|
"logits/rejected": 2.2805395126342773, |
|
"logps/chosen": -517.4966430664062, |
|
"logps/rejected": -885.3656005859375, |
|
"loss": 0.2767, |
|
"rewards/accuracies": 0.893750011920929, |
|
"rewards/chosen": -1.9133589267730713, |
|
"rewards/margins": 3.5970351696014404, |
|
"rewards/rejected": -5.510394096374512, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.5135322692574601, |
|
"grad_norm": 23.570388915292295, |
|
"learning_rate": 4.6341295456597906e-07, |
|
"logits/chosen": 0.6538206338882446, |
|
"logits/rejected": 1.800736665725708, |
|
"logps/chosen": -486.4225158691406, |
|
"logps/rejected": -818.6160278320312, |
|
"loss": 0.2972, |
|
"rewards/accuracies": 0.840624988079071, |
|
"rewards/chosen": -1.574295997619629, |
|
"rewards/margins": 3.247091770172119, |
|
"rewards/rejected": -4.821388244628906, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.527411519777932, |
|
"grad_norm": 22.92056805222348, |
|
"learning_rate": 4.6019416057727577e-07, |
|
"logits/chosen": 0.5617297291755676, |
|
"logits/rejected": 1.9416172504425049, |
|
"logps/chosen": -554.2542724609375, |
|
"logps/rejected": -1017.8567504882812, |
|
"loss": 0.2817, |
|
"rewards/accuracies": 0.856249988079071, |
|
"rewards/chosen": -1.9359180927276611, |
|
"rewards/margins": 4.510493278503418, |
|
"rewards/rejected": -6.4464111328125, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.5412907702984039, |
|
"grad_norm": 21.96046117437538, |
|
"learning_rate": 4.5685186028537756e-07, |
|
"logits/chosen": 0.08615957945585251, |
|
"logits/rejected": 1.9250423908233643, |
|
"logps/chosen": -521.3770751953125, |
|
"logps/rejected": -1026.9183349609375, |
|
"loss": 0.2752, |
|
"rewards/accuracies": 0.903124988079071, |
|
"rewards/chosen": -1.688780426979065, |
|
"rewards/margins": 4.929998397827148, |
|
"rewards/rejected": -6.618779182434082, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.5551700208188758, |
|
"grad_norm": 16.35709631938306, |
|
"learning_rate": 4.5338801756574185e-07, |
|
"logits/chosen": 0.8631726503372192, |
|
"logits/rejected": 2.426805019378662, |
|
"logps/chosen": -550.8397216796875, |
|
"logps/rejected": -960.1754760742188, |
|
"loss": 0.2975, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -1.955361008644104, |
|
"rewards/margins": 4.027670860290527, |
|
"rewards/rejected": -5.983031749725342, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.5690492713393477, |
|
"grad_norm": 14.715456208460227, |
|
"learning_rate": 4.498046677099674e-07, |
|
"logits/chosen": 0.137635737657547, |
|
"logits/rejected": 1.9270877838134766, |
|
"logps/chosen": -470.42767333984375, |
|
"logps/rejected": -795.9421997070312, |
|
"loss": 0.2826, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": -1.4737566709518433, |
|
"rewards/margins": 3.3337607383728027, |
|
"rewards/rejected": -4.807517051696777, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.5829285218598196, |
|
"grad_norm": 15.380893236552625, |
|
"learning_rate": 4.461039162298939e-07, |
|
"logits/chosen": 1.0668681859970093, |
|
"logits/rejected": 2.6232194900512695, |
|
"logps/chosen": -546.1395874023438, |
|
"logps/rejected": -970.1839599609375, |
|
"loss": 0.271, |
|
"rewards/accuracies": 0.8687499761581421, |
|
"rewards/chosen": -1.9840848445892334, |
|
"rewards/margins": 4.236409664154053, |
|
"rewards/rejected": -6.220494270324707, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.5968077723802915, |
|
"grad_norm": 20.12909286909629, |
|
"learning_rate": 4.4228793762044126e-07, |
|
"logits/chosen": 1.8645107746124268, |
|
"logits/rejected": 3.282402515411377, |
|
"logps/chosen": -556.1341552734375, |
|
"logps/rejected": -999.4012451171875, |
|
"loss": 0.286, |
|
"rewards/accuracies": 0.859375, |
|
"rewards/chosen": -2.260993719100952, |
|
"rewards/margins": 4.435044288635254, |
|
"rewards/rejected": -6.696038246154785, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.6106870229007634, |
|
"grad_norm": 16.373631758502654, |
|
"learning_rate": 4.3835897408191513e-07, |
|
"logits/chosen": 1.3662347793579102, |
|
"logits/rejected": 2.8811306953430176, |
|
"logps/chosen": -512.532958984375, |
|
"logps/rejected": -906.1589965820312, |
|
"loss": 0.2862, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": -1.7359873056411743, |
|
"rewards/margins": 3.9433860778808594, |
|
"rewards/rejected": -5.679372787475586, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.6245662734212353, |
|
"grad_norm": 20.352147452552153, |
|
"learning_rate": 4.34319334202531e-07, |
|
"logits/chosen": 1.8553016185760498, |
|
"logits/rejected": 3.245082139968872, |
|
"logps/chosen": -553.4808349609375, |
|
"logps/rejected": -1043.968505859375, |
|
"loss": 0.2671, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": -2.185727834701538, |
|
"rewards/margins": 4.84250545501709, |
|
"rewards/rejected": -7.028233528137207, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.6384455239417072, |
|
"grad_norm": 14.530740198443842, |
|
"learning_rate": 4.301713916019286e-07, |
|
"logits/chosen": 1.8998327255249023, |
|
"logits/rejected": 3.121591091156006, |
|
"logps/chosen": -545.904296875, |
|
"logps/rejected": -980.6796875, |
|
"loss": 0.277, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -1.9527307748794556, |
|
"rewards/margins": 4.239319324493408, |
|
"rewards/rejected": -6.192049980163574, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.6523247744621791, |
|
"grad_norm": 14.543324773787445, |
|
"learning_rate": 4.2591758353647643e-07, |
|
"logits/chosen": 1.0112075805664062, |
|
"logits/rejected": 2.648533821105957, |
|
"logps/chosen": -551.0354614257812, |
|
"logps/rejected": -958.7569580078125, |
|
"loss": 0.2778, |
|
"rewards/accuracies": 0.8656250238418579, |
|
"rewards/chosen": -1.892492651939392, |
|
"rewards/margins": 4.073439598083496, |
|
"rewards/rejected": -5.965932846069336, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.666204024982651, |
|
"grad_norm": 20.665751040711644, |
|
"learning_rate": 4.2156040946718343e-07, |
|
"logits/chosen": 1.5330281257629395, |
|
"logits/rejected": 3.1015706062316895, |
|
"logps/chosen": -588.2686157226562, |
|
"logps/rejected": -1030.510498046875, |
|
"loss": 0.2843, |
|
"rewards/accuracies": 0.878125011920929, |
|
"rewards/chosen": -2.3880248069763184, |
|
"rewards/margins": 4.504977226257324, |
|
"rewards/rejected": -6.893001556396484, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.6800832755031229, |
|
"grad_norm": 15.05601087138497, |
|
"learning_rate": 4.1710242959106056e-07, |
|
"logits/chosen": 0.5176582336425781, |
|
"logits/rejected": 2.4557127952575684, |
|
"logps/chosen": -507.008544921875, |
|
"logps/rejected": -886.51416015625, |
|
"loss": 0.294, |
|
"rewards/accuracies": 0.846875011920929, |
|
"rewards/chosen": -1.5845837593078613, |
|
"rewards/margins": 3.810807704925537, |
|
"rewards/rejected": -5.395391464233398, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.6939625260235948, |
|
"grad_norm": 16.718450087742482, |
|
"learning_rate": 4.125462633367959e-07, |
|
"logits/chosen": 1.4314930438995361, |
|
"logits/rejected": 2.9996209144592285, |
|
"logps/chosen": -532.1755981445312, |
|
"logps/rejected": -1021.0213623046875, |
|
"loss": 0.266, |
|
"rewards/accuracies": 0.878125011920929, |
|
"rewards/chosen": -2.0497827529907227, |
|
"rewards/margins": 4.718562126159668, |
|
"rewards/rejected": -6.768344879150391, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.7078417765440667, |
|
"grad_norm": 19.85458205442182, |
|
"learning_rate": 4.0789458782562435e-07, |
|
"logits/chosen": 1.293492317199707, |
|
"logits/rejected": 2.812124013900757, |
|
"logps/chosen": -551.9437866210938, |
|
"logps/rejected": -1100.38720703125, |
|
"loss": 0.2609, |
|
"rewards/accuracies": 0.887499988079071, |
|
"rewards/chosen": -2.3401436805725098, |
|
"rewards/margins": 5.148820400238037, |
|
"rewards/rejected": -7.4889631271362305, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.7217210270645386, |
|
"grad_norm": 14.477360180818952, |
|
"learning_rate": 4.031501362983007e-07, |
|
"logits/chosen": 0.7156480550765991, |
|
"logits/rejected": 2.6717166900634766, |
|
"logps/chosen": -510.4054260253906, |
|
"logps/rejected": -1019.923828125, |
|
"loss": 0.2805, |
|
"rewards/accuracies": 0.887499988079071, |
|
"rewards/chosen": -1.7749239206314087, |
|
"rewards/margins": 4.909750938415527, |
|
"rewards/rejected": -6.6846747398376465, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.7356002775850105, |
|
"grad_norm": 17.058981872372126, |
|
"learning_rate": 3.9831569650909553e-07, |
|
"logits/chosen": 0.01914766989648342, |
|
"logits/rejected": 2.122251510620117, |
|
"logps/chosen": -551.1067504882812, |
|
"logps/rejected": -933.5851440429688, |
|
"loss": 0.2736, |
|
"rewards/accuracies": 0.846875011920929, |
|
"rewards/chosen": -2.0334556102752686, |
|
"rewards/margins": 3.904972791671753, |
|
"rewards/rejected": -5.9384284019470215, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.7494795281054824, |
|
"grad_norm": 19.755034017893955, |
|
"learning_rate": 3.933941090877615e-07, |
|
"logits/chosen": -0.34956273436546326, |
|
"logits/rejected": 2.1510236263275146, |
|
"logps/chosen": -553.6760864257812, |
|
"logps/rejected": -1035.0694580078125, |
|
"loss": 0.2598, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -2.195675849914551, |
|
"rewards/margins": 4.830143928527832, |
|
"rewards/rejected": -7.025820732116699, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.7633587786259542, |
|
"grad_norm": 15.194278158742483, |
|
"learning_rate": 3.883882658704306e-07, |
|
"logits/chosen": -0.1008995771408081, |
|
"logits/rejected": 2.1417980194091797, |
|
"logps/chosen": -562.4782104492188, |
|
"logps/rejected": -1068.9832763671875, |
|
"loss": 0.2769, |
|
"rewards/accuracies": 0.8656250238418579, |
|
"rewards/chosen": -2.3539323806762695, |
|
"rewards/margins": 5.010300636291504, |
|
"rewards/rejected": -7.364233493804932, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.7772380291464261, |
|
"grad_norm": 17.553015958549693, |
|
"learning_rate": 3.833011082004228e-07, |
|
"logits/chosen": -0.8681015968322754, |
|
"logits/rejected": 1.4696462154388428, |
|
"logps/chosen": -573.9002685546875, |
|
"logps/rejected": -1092.654296875, |
|
"loss": 0.2772, |
|
"rewards/accuracies": 0.828125, |
|
"rewards/chosen": -2.370056390762329, |
|
"rewards/margins": 5.011393070220947, |
|
"rewards/rejected": -7.381448268890381, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.7911172796668979, |
|
"grad_norm": 15.00893179617878, |
|
"learning_rate": 3.781356251999663e-07, |
|
"logits/chosen": -1.1506744623184204, |
|
"logits/rejected": 0.8982523679733276, |
|
"logps/chosen": -535.85595703125, |
|
"logps/rejected": -979.8224487304688, |
|
"loss": 0.2904, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": -2.1066572666168213, |
|
"rewards/margins": 4.214187145233154, |
|
"rewards/rejected": -6.320844650268555, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.8049965301873698, |
|
"grad_norm": 17.833191557593317, |
|
"learning_rate": 3.728948520138426e-07, |
|
"logits/chosen": -0.45408257842063904, |
|
"logits/rejected": 1.915353775024414, |
|
"logps/chosen": -540.576171875, |
|
"logps/rejected": -994.28564453125, |
|
"loss": 0.2725, |
|
"rewards/accuracies": 0.859375, |
|
"rewards/chosen": -2.1601402759552, |
|
"rewards/margins": 4.492788314819336, |
|
"rewards/rejected": -6.652928829193115, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.8188757807078417, |
|
"grad_norm": 16.37664073810551, |
|
"learning_rate": 3.6758186802599064e-07, |
|
"logits/chosen": 0.19857454299926758, |
|
"logits/rejected": 2.2812840938568115, |
|
"logps/chosen": -550.7555541992188, |
|
"logps/rejected": -1052.941162109375, |
|
"loss": 0.2681, |
|
"rewards/accuracies": 0.878125011920929, |
|
"rewards/chosen": -2.186647415161133, |
|
"rewards/margins": 4.857481956481934, |
|
"rewards/rejected": -7.044129848480225, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.8327550312283136, |
|
"grad_norm": 15.489252159692187, |
|
"learning_rate": 3.6219979505011555e-07, |
|
"logits/chosen": 0.13169285655021667, |
|
"logits/rejected": 2.4862585067749023, |
|
"logps/chosen": -545.8336181640625, |
|
"logps/rejected": -1015.2984619140625, |
|
"loss": 0.2599, |
|
"rewards/accuracies": 0.8843749761581421, |
|
"rewards/chosen": -2.0449271202087402, |
|
"rewards/margins": 4.806704521179199, |
|
"rewards/rejected": -6.851631164550781, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.8466342817487855, |
|
"grad_norm": 16.034098237394367, |
|
"learning_rate": 3.5675179549536786e-07, |
|
"logits/chosen": 1.026650309562683, |
|
"logits/rejected": 2.891331672668457, |
|
"logps/chosen": -562.756103515625, |
|
"logps/rejected": -1112.7884521484375, |
|
"loss": 0.2655, |
|
"rewards/accuracies": 0.871874988079071, |
|
"rewards/chosen": -2.344050884246826, |
|
"rewards/margins": 5.390646934509277, |
|
"rewards/rejected": -7.7346978187561035, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.8605135322692574, |
|
"grad_norm": 15.964353907588587, |
|
"learning_rate": 3.512410705081684e-07, |
|
"logits/chosen": 0.5978251695632935, |
|
"logits/rejected": 2.6756691932678223, |
|
"logps/chosen": -596.0946655273438, |
|
"logps/rejected": -1164.9000244140625, |
|
"loss": 0.2654, |
|
"rewards/accuracies": 0.890625, |
|
"rewards/chosen": -2.3005499839782715, |
|
"rewards/margins": 5.751313209533691, |
|
"rewards/rejected": -8.051862716674805, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.8743927827897293, |
|
"grad_norm": 15.534990200131912, |
|
"learning_rate": 3.4567085809127245e-07, |
|
"logits/chosen": 0.7654634714126587, |
|
"logits/rejected": 2.563817262649536, |
|
"logps/chosen": -553.0380859375, |
|
"logps/rejected": -1015.8709106445312, |
|
"loss": 0.2837, |
|
"rewards/accuracies": 0.8343750238418579, |
|
"rewards/chosen": -2.105071544647217, |
|
"rewards/margins": 4.519524097442627, |
|
"rewards/rejected": -6.624594688415527, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.8882720333102012, |
|
"grad_norm": 15.473999576589652, |
|
"learning_rate": 3.400444312011776e-07, |
|
"logits/chosen": 1.6810909509658813, |
|
"logits/rejected": 3.29026460647583, |
|
"logps/chosen": -547.7821044921875, |
|
"logps/rejected": -979.19580078125, |
|
"loss": 0.2808, |
|
"rewards/accuracies": 0.871874988079071, |
|
"rewards/chosen": -2.2526745796203613, |
|
"rewards/margins": 4.376776695251465, |
|
"rewards/rejected": -6.629450798034668, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.9021512838306731, |
|
"grad_norm": 18.65677766855292, |
|
"learning_rate": 3.343650958249935e-07, |
|
"logits/chosen": 1.9508390426635742, |
|
"logits/rejected": 3.384152889251709, |
|
"logps/chosen": -611.1751708984375, |
|
"logps/rejected": -1097.154541015625, |
|
"loss": 0.2717, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": -2.623706340789795, |
|
"rewards/margins": 4.866621494293213, |
|
"rewards/rejected": -7.49032735824585, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.916030534351145, |
|
"grad_norm": 15.172341090722101, |
|
"learning_rate": 3.286361890379034e-07, |
|
"logits/chosen": 1.418536901473999, |
|
"logits/rejected": 2.9168481826782227, |
|
"logps/chosen": -542.9044799804688, |
|
"logps/rejected": -998.8190307617188, |
|
"loss": 0.2649, |
|
"rewards/accuracies": 0.878125011920929, |
|
"rewards/chosen": -2.128884792327881, |
|
"rewards/margins": 4.480809688568115, |
|
"rewards/rejected": -6.6096954345703125, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.9299097848716169, |
|
"grad_norm": 21.03897492647871, |
|
"learning_rate": 3.2286107704235875e-07, |
|
"logits/chosen": 0.6511715650558472, |
|
"logits/rejected": 2.4751362800598145, |
|
"logps/chosen": -516.9407348632812, |
|
"logps/rejected": -994.5736083984375, |
|
"loss": 0.2687, |
|
"rewards/accuracies": 0.856249988079071, |
|
"rewards/chosen": -1.8717918395996094, |
|
"rewards/margins": 4.618001937866211, |
|
"rewards/rejected": -6.489793300628662, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.9437890353920888, |
|
"grad_norm": 23.147160211659624, |
|
"learning_rate": 3.1704315319015936e-07, |
|
"logits/chosen": 0.42685967683792114, |
|
"logits/rejected": 2.4200854301452637, |
|
"logps/chosen": -569.0222778320312, |
|
"logps/rejected": -1023.0499877929688, |
|
"loss": 0.2637, |
|
"rewards/accuracies": 0.856249988079071, |
|
"rewards/chosen": -2.241736650466919, |
|
"rewards/margins": 4.569226264953613, |
|
"rewards/rejected": -6.8109636306762695, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.9576682859125607, |
|
"grad_norm": 19.074650599712797, |
|
"learning_rate": 3.1118583598858094e-07, |
|
"logits/chosen": 0.3251928687095642, |
|
"logits/rejected": 2.361527442932129, |
|
"logps/chosen": -536.3818969726562, |
|
"logps/rejected": -1084.29150390625, |
|
"loss": 0.269, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": -2.159126043319702, |
|
"rewards/margins": 5.291468620300293, |
|
"rewards/rejected": -7.450594425201416, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.9715475364330326, |
|
"grad_norm": 16.46319504999634, |
|
"learning_rate": 3.052925670917219e-07, |
|
"logits/chosen": 0.3188991844654083, |
|
"logits/rejected": 2.1961517333984375, |
|
"logps/chosen": -590.7283935546875, |
|
"logps/rejected": -1070.7264404296875, |
|
"loss": 0.2754, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -2.466348171234131, |
|
"rewards/margins": 4.865278720855713, |
|
"rewards/rejected": -7.33162784576416, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.9854267869535045, |
|
"grad_norm": 22.275614803003936, |
|
"learning_rate": 2.9936680927824935e-07, |
|
"logits/chosen": 0.7669705152511597, |
|
"logits/rejected": 2.4150214195251465, |
|
"logps/chosen": -574.384033203125, |
|
"logps/rejected": -1084.837646484375, |
|
"loss": 0.2543, |
|
"rewards/accuracies": 0.856249988079071, |
|
"rewards/chosen": -2.393491268157959, |
|
"rewards/margins": 5.025046348571777, |
|
"rewards/rejected": -7.4185380935668945, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.9993060374739764, |
|
"grad_norm": 22.856869601641986, |
|
"learning_rate": 2.934120444167326e-07, |
|
"logits/chosen": 0.018332133069634438, |
|
"logits/rejected": 1.9890985488891602, |
|
"logps/chosen": -554.123291015625, |
|
"logps/rejected": -1072.498046875, |
|
"loss": 0.2651, |
|
"rewards/accuracies": 0.8656250238418579, |
|
"rewards/chosen": -2.201577663421631, |
|
"rewards/margins": 5.162604331970215, |
|
"rewards/rejected": -7.3641815185546875, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 1.0131852879944483, |
|
"grad_norm": 16.323336520132766, |
|
"learning_rate": 2.8743177141975993e-07, |
|
"logits/chosen": 0.2875978350639343, |
|
"logits/rejected": 2.0843756198883057, |
|
"logps/chosen": -598.0647583007812, |
|
"logps/rejected": -1214.6568603515625, |
|
"loss": 0.1848, |
|
"rewards/accuracies": 0.921875, |
|
"rewards/chosen": -2.478050708770752, |
|
"rewards/margins": 6.1301589012146, |
|
"rewards/rejected": -8.608209609985352, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 1.0270645385149202, |
|
"grad_norm": 17.717617068699322, |
|
"learning_rate": 2.814295041880407e-07, |
|
"logits/chosen": 1.5749400854110718, |
|
"logits/rejected": 3.2800605297088623, |
|
"logps/chosen": -720.4405517578125, |
|
"logps/rejected": -1425.5921630859375, |
|
"loss": 0.1738, |
|
"rewards/accuracies": 0.9156249761581421, |
|
"rewards/chosen": -3.6376540660858154, |
|
"rewards/margins": 6.913857460021973, |
|
"rewards/rejected": -10.55151081085205, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 1.040943789035392, |
|
"grad_norm": 16.138966847361562, |
|
"learning_rate": 2.754087695457005e-07, |
|
"logits/chosen": 1.3446929454803467, |
|
"logits/rejected": 3.647613525390625, |
|
"logps/chosen": -660.2525024414062, |
|
"logps/rejected": -1274.7076416015625, |
|
"loss": 0.1695, |
|
"rewards/accuracies": 0.9125000238418579, |
|
"rewards/chosen": -3.0320277214050293, |
|
"rewards/margins": 6.136539936065674, |
|
"rewards/rejected": -9.168566703796387, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 1.054823039555864, |
|
"grad_norm": 15.364782345427345, |
|
"learning_rate": 2.6937310516798275e-07, |
|
"logits/chosen": 1.2250503301620483, |
|
"logits/rejected": 3.3649864196777344, |
|
"logps/chosen": -629.1475830078125, |
|
"logps/rejected": -1302.7933349609375, |
|
"loss": 0.1799, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -2.8058159351348877, |
|
"rewards/margins": 6.627597808837891, |
|
"rewards/rejected": -9.4334135055542, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 1.0687022900763359, |
|
"grad_norm": 18.530915715706502, |
|
"learning_rate": 2.6332605750257456e-07, |
|
"logits/chosen": 0.9841400384902954, |
|
"logits/rejected": 3.341764450073242, |
|
"logps/chosen": -656.33251953125, |
|
"logps/rejected": -1349.439697265625, |
|
"loss": 0.1607, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": -3.064764976501465, |
|
"rewards/margins": 6.942727565765381, |
|
"rewards/rejected": -10.007492065429688, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 1.0825815405968078, |
|
"grad_norm": 18.346652449927596, |
|
"learning_rate": 2.5727117968577785e-07, |
|
"logits/chosen": 1.088732361793518, |
|
"logits/rejected": 3.094304323196411, |
|
"logps/chosen": -656.88134765625, |
|
"logps/rejected": -1397.916259765625, |
|
"loss": 0.1514, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": -3.20575213432312, |
|
"rewards/margins": 7.160604000091553, |
|
"rewards/rejected": -10.366357803344727, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 1.0964607911172797, |
|
"grad_norm": 17.314542797248876, |
|
"learning_rate": 2.5121202945475043e-07, |
|
"logits/chosen": 1.5640583038330078, |
|
"logits/rejected": 3.803560256958008, |
|
"logps/chosen": -630.0179443359375, |
|
"logps/rejected": -1349.983642578125, |
|
"loss": 0.178, |
|
"rewards/accuracies": 0.909375011920929, |
|
"rewards/chosen": -2.848733425140381, |
|
"rewards/margins": 7.226857662200928, |
|
"rewards/rejected": -10.075590133666992, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 1.1103400416377516, |
|
"grad_norm": 14.955502239768675, |
|
"learning_rate": 2.4515216705704393e-07, |
|
"logits/chosen": 1.8664920330047607, |
|
"logits/rejected": 3.7396292686462402, |
|
"logps/chosen": -608.3690795898438, |
|
"logps/rejected": -1366.815673828125, |
|
"loss": 0.1637, |
|
"rewards/accuracies": 0.953125, |
|
"rewards/chosen": -2.7840514183044434, |
|
"rewards/margins": 7.429004669189453, |
|
"rewards/rejected": -10.213056564331055, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.1242192921582235, |
|
"grad_norm": 18.71030295389215, |
|
"learning_rate": 2.39095153158666e-07, |
|
"logits/chosen": 1.694549560546875, |
|
"logits/rejected": 3.747912645339966, |
|
"logps/chosen": -657.016845703125, |
|
"logps/rejected": -1396.2716064453125, |
|
"loss": 0.1581, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -3.1420130729675293, |
|
"rewards/margins": 7.294529914855957, |
|
"rewards/rejected": -10.436542510986328, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 1.1380985426786954, |
|
"grad_norm": 14.829984041140372, |
|
"learning_rate": 2.330445467518977e-07, |
|
"logits/chosen": 0.7256888747215271, |
|
"logits/rejected": 2.9225716590881348, |
|
"logps/chosen": -660.5631103515625, |
|
"logps/rejected": -1347.63134765625, |
|
"loss": 0.1619, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": -3.208239793777466, |
|
"rewards/margins": 6.785590171813965, |
|
"rewards/rejected": -9.993829727172852, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 1.1519777931991673, |
|
"grad_norm": 22.137973289587368, |
|
"learning_rate": 2.270039030640931e-07, |
|
"logits/chosen": 1.578162431716919, |
|
"logits/rejected": 3.47196888923645, |
|
"logps/chosen": -628.019287109375, |
|
"logps/rejected": -1320.9600830078125, |
|
"loss": 0.1749, |
|
"rewards/accuracies": 0.893750011920929, |
|
"rewards/chosen": -2.9641802310943604, |
|
"rewards/margins": 6.771336555480957, |
|
"rewards/rejected": -9.735516548156738, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 1.1658570437196392, |
|
"grad_norm": 19.216307710411925, |
|
"learning_rate": 2.209767714686924e-07, |
|
"logits/chosen": 1.5978351831436157, |
|
"logits/rejected": 3.7053027153015137, |
|
"logps/chosen": -653.354736328125, |
|
"logps/rejected": -1406.637451171875, |
|
"loss": 0.1606, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": -3.2107303142547607, |
|
"rewards/margins": 7.3466925621032715, |
|
"rewards/rejected": -10.557422637939453, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 1.179736294240111, |
|
"grad_norm": 21.415640762520873, |
|
"learning_rate": 2.1496669339967344e-07, |
|
"logits/chosen": 1.8785479068756104, |
|
"logits/rejected": 3.8594677448272705, |
|
"logps/chosen": -668.572265625, |
|
"logps/rejected": -1460.521240234375, |
|
"loss": 0.1611, |
|
"rewards/accuracies": 0.9156249761581421, |
|
"rewards/chosen": -3.3157646656036377, |
|
"rewards/margins": 7.8289475440979, |
|
"rewards/rejected": -11.144712448120117, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 1.193615544760583, |
|
"grad_norm": 19.68985482074639, |
|
"learning_rate": 2.0897720027066897e-07, |
|
"logits/chosen": 1.5298289060592651, |
|
"logits/rejected": 3.88330078125, |
|
"logps/chosen": -660.7781982421875, |
|
"logps/rejected": -1415.609130859375, |
|
"loss": 0.1637, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": -3.227867603302002, |
|
"rewards/margins": 7.6040802001953125, |
|
"rewards/rejected": -10.831949234008789, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 1.2074947952810549, |
|
"grad_norm": 19.76316476703645, |
|
"learning_rate": 2.0301181139997202e-07, |
|
"logits/chosen": 1.5017510652542114, |
|
"logits/rejected": 3.601763963699341, |
|
"logps/chosen": -664.7364501953125, |
|
"logps/rejected": -1365.153564453125, |
|
"loss": 0.1716, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -3.1189022064208984, |
|
"rewards/margins": 6.980868339538574, |
|
"rewards/rejected": -10.099771499633789, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 1.2213740458015268, |
|
"grad_norm": 15.231393479477973, |
|
"learning_rate": 1.970740319426474e-07, |
|
"logits/chosen": 2.0277457237243652, |
|
"logits/rejected": 3.8836147785186768, |
|
"logps/chosen": -615.454833984375, |
|
"logps/rejected": -1219.456298828125, |
|
"loss": 0.1803, |
|
"rewards/accuracies": 0.9156249761581421, |
|
"rewards/chosen": -2.8338332176208496, |
|
"rewards/margins": 5.970818519592285, |
|
"rewards/rejected": -8.804651260375977, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 1.2352532963219987, |
|
"grad_norm": 23.68508206139152, |
|
"learning_rate": 1.911673508309656e-07, |
|
"logits/chosen": 1.9813998937606812, |
|
"logits/rejected": 3.9024734497070312, |
|
"logps/chosen": -628.2916259765625, |
|
"logps/rejected": -1264.0699462890625, |
|
"loss": 0.1643, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": -2.8454182147979736, |
|
"rewards/margins": 6.396851539611816, |
|
"rewards/rejected": -9.242269515991211, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 1.2491325468424705, |
|
"grad_norm": 19.952403887659834, |
|
"learning_rate": 1.8529523872436977e-07, |
|
"logits/chosen": 1.032080054283142, |
|
"logits/rejected": 3.210120677947998, |
|
"logps/chosen": -671.1250610351562, |
|
"logps/rejected": -1549.0615234375, |
|
"loss": 0.16, |
|
"rewards/accuracies": 0.956250011920929, |
|
"rewards/chosen": -3.1327102184295654, |
|
"rewards/margins": 8.625667572021484, |
|
"rewards/rejected": -11.758378982543945, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 1.2630117973629424, |
|
"grad_norm": 22.651112148813382, |
|
"learning_rate": 1.7946114597017808e-07, |
|
"logits/chosen": 1.6984176635742188, |
|
"logits/rejected": 3.7526144981384277, |
|
"logps/chosen": -660.7144165039062, |
|
"logps/rejected": -1447.436767578125, |
|
"loss": 0.1619, |
|
"rewards/accuracies": 0.940625011920929, |
|
"rewards/chosen": -3.2229180335998535, |
|
"rewards/margins": 7.7218732833862305, |
|
"rewards/rejected": -10.944790840148926, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 1.2768910478834143, |
|
"grad_norm": 22.126302318317872, |
|
"learning_rate": 1.7366850057622172e-07, |
|
"logits/chosen": 1.6554454565048218, |
|
"logits/rejected": 3.758930206298828, |
|
"logps/chosen": -654.619140625, |
|
"logps/rejected": -1429.7637939453125, |
|
"loss": 0.1653, |
|
"rewards/accuracies": 0.9468749761581421, |
|
"rewards/chosen": -3.282761812210083, |
|
"rewards/margins": 7.560622215270996, |
|
"rewards/rejected": -10.8433837890625, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 1.2907702984038862, |
|
"grad_norm": 17.175640354603786, |
|
"learning_rate": 1.6792070619660974e-07, |
|
"logits/chosen": 1.6677768230438232, |
|
"logits/rejected": 3.8167777061462402, |
|
"logps/chosen": -654.5197143554688, |
|
"logps/rejected": -1449.1417236328125, |
|
"loss": 0.1654, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": -3.2044639587402344, |
|
"rewards/margins": 7.7717132568359375, |
|
"rewards/rejected": -10.976176261901855, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 1.3046495489243581, |
|
"grad_norm": 19.342272367773113, |
|
"learning_rate": 1.622211401318028e-07, |
|
"logits/chosen": 1.567354440689087, |
|
"logits/rejected": 4.129142761230469, |
|
"logps/chosen": -657.1278686523438, |
|
"logps/rejected": -1453.904541015625, |
|
"loss": 0.1595, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": -3.066843271255493, |
|
"rewards/margins": 7.886987209320068, |
|
"rewards/rejected": -10.953829765319824, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 1.31852879944483, |
|
"grad_norm": 23.040092913699233, |
|
"learning_rate": 1.5657315134417244e-07, |
|
"logits/chosen": 1.8149755001068115, |
|
"logits/rejected": 4.2323408126831055, |
|
"logps/chosen": -688.6448364257812, |
|
"logps/rejected": -1585.87841796875, |
|
"loss": 0.1665, |
|
"rewards/accuracies": 0.9125000238418579, |
|
"rewards/chosen": -3.4780020713806152, |
|
"rewards/margins": 8.797907829284668, |
|
"rewards/rejected": -12.275908470153809, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 1.332408049965302, |
|
"grad_norm": 17.336766543398397, |
|
"learning_rate": 1.5098005849021078e-07, |
|
"logits/chosen": 1.7025401592254639, |
|
"logits/rejected": 4.534233570098877, |
|
"logps/chosen": -703.167724609375, |
|
"logps/rejected": -1630.665283203125, |
|
"loss": 0.1672, |
|
"rewards/accuracies": 0.9281250238418579, |
|
"rewards/chosen": -3.5511913299560547, |
|
"rewards/margins": 9.27750301361084, |
|
"rewards/rejected": -12.828694343566895, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 1.3462873004857738, |
|
"grad_norm": 20.603249005514577, |
|
"learning_rate": 1.454451479705484e-07, |
|
"logits/chosen": 2.531970500946045, |
|
"logits/rejected": 4.76376485824585, |
|
"logps/chosen": -658.04931640625, |
|
"logps/rejected": -1445.8599853515625, |
|
"loss": 0.1611, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -3.303406238555908, |
|
"rewards/margins": 7.701096534729004, |
|
"rewards/rejected": -11.00450325012207, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 1.3601665510062457, |
|
"grad_norm": 26.637520706621387, |
|
"learning_rate": 1.3997167199892385e-07, |
|
"logits/chosen": 2.412917375564575, |
|
"logits/rejected": 4.9495530128479, |
|
"logps/chosen": -706.7518310546875, |
|
"logps/rejected": -1550.7310791015625, |
|
"loss": 0.16, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": -3.5566086769104004, |
|
"rewards/margins": 8.430598258972168, |
|
"rewards/rejected": -11.987205505371094, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 1.3740458015267176, |
|
"grad_norm": 22.552075685705777, |
|
"learning_rate": 1.3456284669124157e-07, |
|
"logits/chosen": 2.5810890197753906, |
|
"logits/rejected": 4.767507076263428, |
|
"logps/chosen": -692.7928466796875, |
|
"logps/rejected": -1483.0191650390625, |
|
"loss": 0.1661, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": -3.5196609497070312, |
|
"rewards/margins": 7.722577095031738, |
|
"rewards/rejected": -11.24223804473877, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 1.3879250520471895, |
|
"grad_norm": 18.63062778053758, |
|
"learning_rate": 1.2922185017584036e-07, |
|
"logits/chosen": 2.9480197429656982, |
|
"logits/rejected": 5.796414375305176, |
|
"logps/chosen": -692.9706420898438, |
|
"logps/rejected": -1486.6146240234375, |
|
"loss": 0.1587, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": -3.4646847248077393, |
|
"rewards/margins": 7.979840278625488, |
|
"rewards/rejected": -11.444524765014648, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.3879250520471895, |
|
"eval_logits/chosen": 2.4397730827331543, |
|
"eval_logits/rejected": 4.560609817504883, |
|
"eval_logps/chosen": -779.3055419921875, |
|
"eval_logps/rejected": -1613.5015869140625, |
|
"eval_loss": 0.24712695181369781, |
|
"eval_rewards/accuracies": 0.8909774422645569, |
|
"eval_rewards/chosen": -3.947213649749756, |
|
"eval_rewards/margins": 8.086578369140625, |
|
"eval_rewards/rejected": -12.033791542053223, |
|
"eval_runtime": 385.5049, |
|
"eval_samples_per_second": 22.026, |
|
"eval_steps_per_second": 0.345, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.4018043025676614, |
|
"grad_norm": 17.720503294892175, |
|
"learning_rate": 1.2395182072608245e-07, |
|
"logits/chosen": 2.783371925354004, |
|
"logits/rejected": 5.293555736541748, |
|
"logps/chosen": -693.8554077148438, |
|
"logps/rejected": -1455.6026611328125, |
|
"loss": 0.1578, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": -3.711820125579834, |
|
"rewards/margins": 7.577448844909668, |
|
"rewards/rejected": -11.289270401000977, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 1.4156835530881333, |
|
"grad_norm": 39.790575003507485, |
|
"learning_rate": 1.1875585491635998e-07, |
|
"logits/chosen": 3.120913028717041, |
|
"logits/rejected": 5.793875217437744, |
|
"logps/chosen": -757.794921875, |
|
"logps/rejected": -1674.984375, |
|
"loss": 0.1529, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": -4.312003135681152, |
|
"rewards/margins": 8.977596282958984, |
|
"rewards/rejected": -13.289599418640137, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 1.4295628036086052, |
|
"grad_norm": 69.28439112059148, |
|
"learning_rate": 1.1363700580260438e-07, |
|
"logits/chosen": 3.0012755393981934, |
|
"logits/rejected": 5.11987829208374, |
|
"logps/chosen": -737.9568481445312, |
|
"logps/rejected": -1678.4951171875, |
|
"loss": 0.161, |
|
"rewards/accuracies": 0.8968750238418579, |
|
"rewards/chosen": -4.191954612731934, |
|
"rewards/margins": 9.063277244567871, |
|
"rewards/rejected": -13.255231857299805, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 1.4434420541290771, |
|
"grad_norm": 18.312529268747, |
|
"learning_rate": 1.0859828112836539e-07, |
|
"logits/chosen": 2.285982847213745, |
|
"logits/rejected": 4.8581390380859375, |
|
"logps/chosen": -739.9261474609375, |
|
"logps/rejected": -1586.6966552734375, |
|
"loss": 0.1639, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -3.9794273376464844, |
|
"rewards/margins": 8.331348419189453, |
|
"rewards/rejected": -12.310776710510254, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 1.457321304649549, |
|
"grad_norm": 25.436043150499486, |
|
"learning_rate": 1.0364264155751487e-07, |
|
"logits/chosen": 2.474147081375122, |
|
"logits/rejected": 4.805792331695557, |
|
"logps/chosen": -723.0252685546875, |
|
"logps/rejected": -1613.719970703125, |
|
"loss": 0.1599, |
|
"rewards/accuracies": 0.9281250238418579, |
|
"rewards/chosen": -4.043066501617432, |
|
"rewards/margins": 8.693578720092773, |
|
"rewards/rejected": -12.736645698547363, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 1.4712005551700207, |
|
"grad_norm": 18.578653218988183, |
|
"learning_rate": 9.877299893461455e-08, |
|
"logits/chosen": 2.1861484050750732, |
|
"logits/rejected": 5.042906761169434, |
|
"logps/chosen": -748.6861572265625, |
|
"logps/rejected": -1565.2005615234375, |
|
"loss": 0.1539, |
|
"rewards/accuracies": 0.9468749761581421, |
|
"rewards/chosen": -4.071117401123047, |
|
"rewards/margins": 8.12061595916748, |
|
"rewards/rejected": -12.191734313964844, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 1.4850798056904928, |
|
"grad_norm": 21.104448029863153, |
|
"learning_rate": 9.39922145739683e-08, |
|
"logits/chosen": 2.260671854019165, |
|
"logits/rejected": 4.919951438903809, |
|
"logps/chosen": -785.3671875, |
|
"logps/rejected": -1578.3946533203125, |
|
"loss": 0.1559, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": -4.341281414031982, |
|
"rewards/margins": 7.95664119720459, |
|
"rewards/rejected": -12.29792308807373, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 1.4989590562109645, |
|
"grad_norm": 30.412482027711867, |
|
"learning_rate": 8.930309757836516e-08, |
|
"logits/chosen": 2.812704563140869, |
|
"logits/rejected": 5.714043140411377, |
|
"logps/chosen": -803.9837036132812, |
|
"logps/rejected": -1662.1617431640625, |
|
"loss": 0.1624, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -4.532739162445068, |
|
"rewards/margins": 8.574421882629395, |
|
"rewards/rejected": -13.107160568237305, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 1.5128383067314366, |
|
"grad_norm": 27.560035648494992, |
|
"learning_rate": 8.470840318850168e-08, |
|
"logits/chosen": 2.2375552654266357, |
|
"logits/rejected": 5.399850368499756, |
|
"logps/chosen": -772.5816650390625, |
|
"logps/rejected": -1577.226806640625, |
|
"loss": 0.1576, |
|
"rewards/accuracies": 0.953125, |
|
"rewards/chosen": -4.188956260681152, |
|
"rewards/margins": 8.190386772155762, |
|
"rewards/rejected": -12.379343032836914, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 1.5267175572519083, |
|
"grad_norm": 21.688406462998472, |
|
"learning_rate": 8.021083116405173e-08, |
|
"logits/chosen": 2.3147189617156982, |
|
"logits/rejected": 5.235360145568848, |
|
"logps/chosen": -789.7681884765625, |
|
"logps/rejected": -1536.5394287109375, |
|
"loss": 0.1576, |
|
"rewards/accuracies": 0.9281250238418579, |
|
"rewards/chosen": -4.277130126953125, |
|
"rewards/margins": 7.6214165687561035, |
|
"rewards/rejected": -11.898547172546387, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 1.5405968077723804, |
|
"grad_norm": 17.348108637258065, |
|
"learning_rate": 7.581302419733632e-08, |
|
"logits/chosen": 2.7001595497131348, |
|
"logits/rejected": 5.277278900146484, |
|
"logps/chosen": -735.4856567382812, |
|
"logps/rejected": -1611.0947265625, |
|
"loss": 0.1519, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": -4.2074875831604, |
|
"rewards/margins": 8.55141830444336, |
|
"rewards/rejected": -12.758907318115234, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 1.554476058292852, |
|
"grad_norm": 26.00795087450143, |
|
"learning_rate": 7.151756636052527e-08, |
|
"logits/chosen": 2.411917209625244, |
|
"logits/rejected": 5.364395618438721, |
|
"logps/chosen": -790.9634399414062, |
|
"logps/rejected": -1729.780517578125, |
|
"loss": 0.1535, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": -4.578307151794434, |
|
"rewards/margins": 9.278319358825684, |
|
"rewards/rejected": -13.856626510620117, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 1.5683553088133242, |
|
"grad_norm": 19.597840266227173, |
|
"learning_rate": 6.732698158728315e-08, |
|
"logits/chosen": 2.3103692531585693, |
|
"logits/rejected": 5.478982448577881, |
|
"logps/chosen": -766.3836059570312, |
|
"logps/rejected": -1625.908203125, |
|
"loss": 0.1485, |
|
"rewards/accuracies": 0.934374988079071, |
|
"rewards/chosen": -4.306478977203369, |
|
"rewards/margins": 8.519509315490723, |
|
"rewards/rejected": -12.82598876953125, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 1.5822345593337959, |
|
"grad_norm": 37.32265853585954, |
|
"learning_rate": 6.324373218975104e-08, |
|
"logits/chosen": 2.6391475200653076, |
|
"logits/rejected": 5.588366508483887, |
|
"logps/chosen": -738.453125, |
|
"logps/rejected": -1503.0093994140625, |
|
"loss": 0.1743, |
|
"rewards/accuracies": 0.9125000238418579, |
|
"rewards/chosen": -4.1589436531066895, |
|
"rewards/margins": 7.595011234283447, |
|
"rewards/rejected": -11.75395393371582, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 1.596113809854268, |
|
"grad_norm": 34.02286786789855, |
|
"learning_rate": 5.927021741173624e-08, |
|
"logits/chosen": 2.2584009170532227, |
|
"logits/rejected": 5.177404403686523, |
|
"logps/chosen": -716.9951171875, |
|
"logps/rejected": -1559.7662353515625, |
|
"loss": 0.1604, |
|
"rewards/accuracies": 0.9468749761581421, |
|
"rewards/chosen": -4.111263275146484, |
|
"rewards/margins": 8.235260963439941, |
|
"rewards/rejected": -12.346525192260742, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 1.6099930603747397, |
|
"grad_norm": 20.932494952987984, |
|
"learning_rate": 5.5408772018959996e-08, |
|
"logits/chosen": 2.0977630615234375, |
|
"logits/rejected": 5.0792131423950195, |
|
"logps/chosen": -744.5228271484375, |
|
"logps/rejected": -1543.290283203125, |
|
"loss": 0.1549, |
|
"rewards/accuracies": 0.9156249761581421, |
|
"rewards/chosen": -3.9662253856658936, |
|
"rewards/margins": 7.838561058044434, |
|
"rewards/rejected": -11.80478572845459, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 1.6238723108952118, |
|
"grad_norm": 18.156628188332558, |
|
"learning_rate": 5.166166492719124e-08, |
|
"logits/chosen": 2.717636823654175, |
|
"logits/rejected": 5.455983638763428, |
|
"logps/chosen": -789.8887939453125, |
|
"logps/rejected": -1659.1243896484375, |
|
"loss": 0.1431, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -4.5077619552612305, |
|
"rewards/margins": 8.738015174865723, |
|
"rewards/rejected": -13.245776176452637, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 1.6377515614156835, |
|
"grad_norm": 31.62818641679718, |
|
"learning_rate": 4.8031097869072225e-08, |
|
"logits/chosen": 2.479032516479492, |
|
"logits/rejected": 5.483838081359863, |
|
"logps/chosen": -843.8903198242188, |
|
"logps/rejected": -1682.486328125, |
|
"loss": 0.1808, |
|
"rewards/accuracies": 0.9312499761581421, |
|
"rewards/chosen": -4.988056182861328, |
|
"rewards/margins": 8.428102493286133, |
|
"rewards/rejected": -13.416158676147461, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 1.6516308119361556, |
|
"grad_norm": 23.367654380160257, |
|
"learning_rate": 4.451920410042048e-08, |
|
"logits/chosen": 2.4087483882904053, |
|
"logits/rejected": 5.190215110778809, |
|
"logps/chosen": -769.288818359375, |
|
"logps/rejected": -1569.325927734375, |
|
"loss": 0.1421, |
|
"rewards/accuracies": 0.956250011920929, |
|
"rewards/chosen": -4.326132297515869, |
|
"rewards/margins": 7.9340925216674805, |
|
"rewards/rejected": -12.260224342346191, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 1.6655100624566272, |
|
"grad_norm": 19.705530443812812, |
|
"learning_rate": 4.112804714676593e-08, |
|
"logits/chosen": 2.707061529159546, |
|
"logits/rejected": 5.535910606384277, |
|
"logps/chosen": -801.08935546875, |
|
"logps/rejected": -1702.814453125, |
|
"loss": 0.1442, |
|
"rewards/accuracies": 0.953125, |
|
"rewards/chosen": -4.545718193054199, |
|
"rewards/margins": 8.991861343383789, |
|
"rewards/rejected": -13.537579536437988, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 1.6793893129770994, |
|
"grad_norm": 18.481993519274262, |
|
"learning_rate": 3.785961959086026e-08, |
|
"logits/chosen": 2.372300624847412, |
|
"logits/rejected": 4.822690010070801, |
|
"logps/chosen": -803.9580688476562, |
|
"logps/rejected": -1608.513671875, |
|
"loss": 0.1649, |
|
"rewards/accuracies": 0.9281250238418579, |
|
"rewards/chosen": -4.596296787261963, |
|
"rewards/margins": 7.976349830627441, |
|
"rewards/rejected": -12.572647094726562, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 1.693268563497571, |
|
"grad_norm": 17.963706056533137, |
|
"learning_rate": 3.4715841901871545e-08, |
|
"logits/chosen": 2.230515956878662, |
|
"logits/rejected": 4.730217456817627, |
|
"logps/chosen": -780.9340209960938, |
|
"logps/rejected": -1548.3511962890625, |
|
"loss": 0.1478, |
|
"rewards/accuracies": 0.956250011920929, |
|
"rewards/chosen": -4.49829626083374, |
|
"rewards/margins": 7.652662754058838, |
|
"rewards/rejected": -12.150960922241211, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 1.7071478140180432, |
|
"grad_norm": 19.452721114097585, |
|
"learning_rate": 3.169856130695106e-08, |
|
"logits/chosen": 2.7935097217559814, |
|
"logits/rejected": 5.348397254943848, |
|
"logps/chosen": -837.3346557617188, |
|
"logps/rejected": -1694.122314453125, |
|
"loss": 0.1637, |
|
"rewards/accuracies": 0.90625, |
|
"rewards/chosen": -4.844791412353516, |
|
"rewards/margins": 8.404356002807617, |
|
"rewards/rejected": -13.249147415161133, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 1.7210270645385148, |
|
"grad_norm": 24.207735428419646, |
|
"learning_rate": 2.8809550705835546e-08, |
|
"logits/chosen": 2.0396385192871094, |
|
"logits/rejected": 5.347299098968506, |
|
"logps/chosen": -809.9459838867188, |
|
"logps/rejected": -1699.661865234375, |
|
"loss": 0.1595, |
|
"rewards/accuracies": 0.9468749761581421, |
|
"rewards/chosen": -4.592886447906494, |
|
"rewards/margins": 8.931785583496094, |
|
"rewards/rejected": -13.52467155456543, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 1.734906315058987, |
|
"grad_norm": 24.611504297665018, |
|
"learning_rate": 2.6050507629123724e-08, |
|
"logits/chosen": 2.3445913791656494, |
|
"logits/rejected": 5.292250633239746, |
|
"logps/chosen": -763.6495361328125, |
|
"logps/rejected": -1591.069091796875, |
|
"loss": 0.1746, |
|
"rewards/accuracies": 0.921875, |
|
"rewards/chosen": -4.202335834503174, |
|
"rewards/margins": 8.237665176391602, |
|
"rewards/rejected": -12.440000534057617, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 1.7487855655794586, |
|
"grad_norm": 17.351441576745916, |
|
"learning_rate": 2.3423053240837514e-08, |
|
"logits/chosen": 2.1456151008605957, |
|
"logits/rejected": 5.197157859802246, |
|
"logps/chosen": -775.988525390625, |
|
"logps/rejected": -1635.9970703125, |
|
"loss": 0.1602, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": -4.428194999694824, |
|
"rewards/margins": 8.511189460754395, |
|
"rewards/rejected": -12.939386367797852, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 1.7626648160999308, |
|
"grad_norm": 23.373280732240797, |
|
"learning_rate": 2.0928731385855548e-08, |
|
"logits/chosen": 2.3203883171081543, |
|
"logits/rejected": 5.307148456573486, |
|
"logps/chosen": -750.8080444335938, |
|
"logps/rejected": -1596.9986572265625, |
|
"loss": 0.1503, |
|
"rewards/accuracies": 0.921875, |
|
"rewards/chosen": -4.260045051574707, |
|
"rewards/margins": 8.443410873413086, |
|
"rewards/rejected": -12.703454971313477, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 1.7765440666204024, |
|
"grad_norm": 21.91095019107269, |
|
"learning_rate": 1.8569007682777415e-08, |
|
"logits/chosen": 2.3937430381774902, |
|
"logits/rejected": 5.018430233001709, |
|
"logps/chosen": -775.3260498046875, |
|
"logps/rejected": -1653.3775634765625, |
|
"loss": 0.1517, |
|
"rewards/accuracies": 0.9468749761581421, |
|
"rewards/chosen": -4.4354681968688965, |
|
"rewards/margins": 8.597478866577148, |
|
"rewards/rejected": -13.032946586608887, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 1.7904233171408745, |
|
"grad_norm": 18.785456105009327, |
|
"learning_rate": 1.6345268662752904e-08, |
|
"logits/chosen": 2.1837244033813477, |
|
"logits/rejected": 4.996502876281738, |
|
"logps/chosen": -797.1134643554688, |
|
"logps/rejected": -1595.029541015625, |
|
"loss": 0.1469, |
|
"rewards/accuracies": 0.934374988079071, |
|
"rewards/chosen": -4.397826671600342, |
|
"rewards/margins": 7.9338555335998535, |
|
"rewards/rejected": -12.331681251525879, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 1.8043025676613462, |
|
"grad_norm": 22.286605985901687, |
|
"learning_rate": 1.4258820954781037e-08, |
|
"logits/chosen": 2.1744418144226074, |
|
"logits/rejected": 5.030113697052002, |
|
"logps/chosen": -767.9185180664062, |
|
"logps/rejected": -1573.216552734375, |
|
"loss": 0.1686, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": -4.429121971130371, |
|
"rewards/margins": 7.945558071136475, |
|
"rewards/rejected": -12.37468147277832, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 1.8181818181818183, |
|
"grad_norm": 25.01416006471772, |
|
"learning_rate": 1.2310890517958389e-08, |
|
"logits/chosen": 2.1604392528533936, |
|
"logits/rejected": 5.2982378005981445, |
|
"logps/chosen": -799.98046875, |
|
"logps/rejected": -1684.6910400390625, |
|
"loss": 0.1451, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": -4.48465633392334, |
|
"rewards/margins": 8.867886543273926, |
|
"rewards/rejected": -13.352543830871582, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 1.83206106870229, |
|
"grad_norm": 22.26904532535015, |
|
"learning_rate": 1.0502621921127774e-08, |
|
"logits/chosen": 2.3922975063323975, |
|
"logits/rejected": 5.1035685539245605, |
|
"logps/chosen": -778.9490966796875, |
|
"logps/rejected": -1598.516845703125, |
|
"loss": 0.1596, |
|
"rewards/accuracies": 0.909375011920929, |
|
"rewards/chosen": -4.48614501953125, |
|
"rewards/margins": 8.004111289978027, |
|
"rewards/rejected": -12.490256309509277, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 1.845940319222762, |
|
"grad_norm": 20.889996927509948, |
|
"learning_rate": 8.83507767035016e-09, |
|
"logits/chosen": 2.3073320388793945, |
|
"logits/rejected": 4.977648735046387, |
|
"logps/chosen": -797.042236328125, |
|
"logps/rejected": -1626.1595458984375, |
|
"loss": 0.1657, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": -4.542098045349121, |
|
"rewards/margins": 8.27629280090332, |
|
"rewards/rejected": -12.818391799926758, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 1.8598195697432338, |
|
"grad_norm": 26.353814591753384, |
|
"learning_rate": 7.309237584595007e-09, |
|
"logits/chosen": 2.40840220451355, |
|
"logits/rejected": 5.275250434875488, |
|
"logps/chosen": -784.8826904296875, |
|
"logps/rejected": -1593.8211669921875, |
|
"loss": 0.1604, |
|
"rewards/accuracies": 0.934374988079071, |
|
"rewards/chosen": -4.507115364074707, |
|
"rewards/margins": 8.047722816467285, |
|
"rewards/rejected": -12.554839134216309, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 1.8736988202637057, |
|
"grad_norm": 27.69480128946826, |
|
"learning_rate": 5.925998220016659e-09, |
|
"logits/chosen": 2.0922293663024902, |
|
"logits/rejected": 5.0452680587768555, |
|
"logps/chosen": -795.8992919921875, |
|
"logps/rejected": -1614.81298828125, |
|
"loss": 0.1657, |
|
"rewards/accuracies": 0.940625011920929, |
|
"rewards/chosen": -4.478159427642822, |
|
"rewards/margins": 8.272706985473633, |
|
"rewards/rejected": -12.75086498260498, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 1.8875780707841776, |
|
"grad_norm": 21.698890899471028, |
|
"learning_rate": 4.6861723431538265e-09, |
|
"logits/chosen": 2.527801275253296, |
|
"logits/rejected": 5.509335517883301, |
|
"logps/chosen": -762.1702270507812, |
|
"logps/rejected": -1630.4300537109375, |
|
"loss": 0.16, |
|
"rewards/accuracies": 0.9281250238418579, |
|
"rewards/chosen": -4.3918232917785645, |
|
"rewards/margins": 8.475740432739258, |
|
"rewards/rejected": -12.867563247680664, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 1.9014573213046495, |
|
"grad_norm": 23.606565024567274, |
|
"learning_rate": 3.5904884533627113e-09, |
|
"logits/chosen": 2.040306568145752, |
|
"logits/rejected": 5.0976457595825195, |
|
"logps/chosen": -809.3186645507812, |
|
"logps/rejected": -1723.1839599609375, |
|
"loss": 0.1661, |
|
"rewards/accuracies": 0.956250011920929, |
|
"rewards/chosen": -4.630982398986816, |
|
"rewards/margins": 9.00139331817627, |
|
"rewards/rejected": -13.63237476348877, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 1.9153365718251214, |
|
"grad_norm": 23.741237778017844, |
|
"learning_rate": 2.639590354763882e-09, |
|
"logits/chosen": 2.073582410812378, |
|
"logits/rejected": 4.958869934082031, |
|
"logps/chosen": -795.91162109375, |
|
"logps/rejected": -1612.056396484375, |
|
"loss": 0.1613, |
|
"rewards/accuracies": 0.940625011920929, |
|
"rewards/chosen": -4.596648693084717, |
|
"rewards/margins": 8.110963821411133, |
|
"rewards/rejected": -12.707611083984375, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 1.9292158223455933, |
|
"grad_norm": 18.578608062996363, |
|
"learning_rate": 1.8340367779545452e-09, |
|
"logits/chosen": 2.297240972518921, |
|
"logits/rejected": 5.08568811416626, |
|
"logps/chosen": -794.4322509765625, |
|
"logps/rejected": -1613.2269287109375, |
|
"loss": 0.1486, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": -4.632008075714111, |
|
"rewards/margins": 8.13344955444336, |
|
"rewards/rejected": -12.765457153320312, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 1.9430950728660652, |
|
"grad_norm": 18.67065103551978, |
|
"learning_rate": 1.1743010517085427e-09, |
|
"logits/chosen": 2.2125773429870605, |
|
"logits/rejected": 5.183515548706055, |
|
"logps/chosen": -828.9417724609375, |
|
"logps/rejected": -1649.573974609375, |
|
"loss": 0.1548, |
|
"rewards/accuracies": 0.9437500238418579, |
|
"rewards/chosen": -4.853725433349609, |
|
"rewards/margins": 8.179043769836426, |
|
"rewards/rejected": -13.032768249511719, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 1.956974323386537, |
|
"grad_norm": 22.95032078584552, |
|
"learning_rate": 6.607708248569377e-10, |
|
"logits/chosen": 2.1667537689208984, |
|
"logits/rejected": 4.85316276550293, |
|
"logps/chosen": -793.8030395507812, |
|
"logps/rejected": -1551.1268310546875, |
|
"loss": 0.1672, |
|
"rewards/accuracies": 0.918749988079071, |
|
"rewards/chosen": -4.6431074142456055, |
|
"rewards/margins": 7.463442325592041, |
|
"rewards/rejected": -12.106550216674805, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 1.970853573907009, |
|
"grad_norm": 18.974426521602602, |
|
"learning_rate": 2.9374783851240923e-10, |
|
"logits/chosen": 2.2130208015441895, |
|
"logits/rejected": 5.2726945877075195, |
|
"logps/chosen": -810.5875244140625, |
|
"logps/rejected": -1586.91552734375, |
|
"loss": 0.1663, |
|
"rewards/accuracies": 0.9156249761581421, |
|
"rewards/chosen": -4.543966770172119, |
|
"rewards/margins": 7.790687561035156, |
|
"rewards/rejected": -12.334654808044434, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 1.984732824427481, |
|
"grad_norm": 22.46417636362631, |
|
"learning_rate": 7.34477487716878e-11, |
|
"logits/chosen": 1.7181364297866821, |
|
"logits/rejected": 4.7832465171813965, |
|
"logps/chosen": -800.3704223632812, |
|
"logps/rejected": -1517.8531494140625, |
|
"loss": 0.1627, |
|
"rewards/accuracies": 0.9375, |
|
"rewards/chosen": -4.434643745422363, |
|
"rewards/margins": 7.220560550689697, |
|
"rewards/rejected": -11.655204772949219, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 1.9986120749479528, |
|
"grad_norm": 31.109687143592964, |
|
"learning_rate": 0.0, |
|
"logits/chosen": 2.243917942047119, |
|
"logits/rejected": 5.361481666564941, |
|
"logps/chosen": -816.5538940429688, |
|
"logps/rejected": -1620.42529296875, |
|
"loss": 0.1599, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": -4.733948707580566, |
|
"rewards/margins": 8.088279724121094, |
|
"rewards/rejected": -12.822227478027344, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 1.9986120749479528, |
|
"step": 1440, |
|
"total_flos": 0.0, |
|
"train_loss": 0.25082930790053476, |
|
"train_runtime": 42170.4729, |
|
"train_samples_per_second": 8.747, |
|
"train_steps_per_second": 0.034 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 1440, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 2, |
|
"save_steps": 100, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": true |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 0.0, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|