| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 3.8256, |
| "eval_steps": 500, |
| "global_step": 600, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "entropy": 0.49680355116724967, |
| "epoch": 0.032, |
| "grad_norm": 1.9729487895965576, |
| "learning_rate": 2.2222222222222224e-07, |
| "logits/chosen": -21.962118183195834, |
| "logits/rejected": -22.031751048121173, |
| "logps/chosen": -20.207871508598327, |
| "logps/rejected": -21.45479338169098, |
| "loss": 0.690625, |
| "mean_token_accuracy": 0.8111328125, |
| "num_tokens": 150832.0, |
| "rewards/accuracies": 0.359375, |
| "rewards/chosen": 0.0033021977869793773, |
| "rewards/margins": 0.012731602950952946, |
| "rewards/rejected": -0.00942940522218123, |
| "step": 5 |
| }, |
| { |
| "entropy": 0.48400292098522185, |
| "epoch": 0.064, |
| "grad_norm": 2.1830124855041504, |
| "learning_rate": 5.000000000000001e-07, |
| "logits/chosen": -22.1027593799377, |
| "logits/rejected": -22.129805485015236, |
| "logps/chosen": -19.691568660736085, |
| "logps/rejected": -21.29295744895935, |
| "loss": 0.690625, |
| "mean_token_accuracy": 0.8115234375, |
| "num_tokens": 301843.0, |
| "rewards/accuracies": 0.55625, |
| "rewards/chosen": 0.013490626483689993, |
| "rewards/margins": 0.02202090589562431, |
| "rewards/rejected": -0.008530278934631496, |
| "step": 10 |
| }, |
| { |
| "entropy": 0.48046681731939317, |
| "epoch": 0.096, |
| "grad_norm": 1.807092308998108, |
| "learning_rate": 7.777777777777779e-07, |
| "logits/chosen": -22.18947406650365, |
| "logits/rejected": -22.187580812234224, |
| "logps/chosen": -19.324772119522095, |
| "logps/rejected": -19.67901744842529, |
| "loss": 0.703125, |
| "mean_token_accuracy": 0.8154296875, |
| "num_tokens": 452652.0, |
| "rewards/accuracies": 0.5375, |
| "rewards/chosen": -0.00036212803097441794, |
| "rewards/margins": 0.016041198861785233, |
| "rewards/rejected": -0.01640332759125158, |
| "step": 15 |
| }, |
| { |
| "entropy": 0.48490327373147013, |
| "epoch": 0.128, |
| "grad_norm": 2.1295247077941895, |
| "learning_rate": 1.0555555555555557e-06, |
| "logits/chosen": -22.082813541884143, |
| "logits/rejected": -22.093126968543437, |
| "logps/chosen": -19.756125855445863, |
| "logps/rejected": -20.089928770065306, |
| "loss": 0.70625, |
| "mean_token_accuracy": 0.81328125, |
| "num_tokens": 603458.0, |
| "rewards/accuracies": 0.50625, |
| "rewards/chosen": -0.0052990868804045025, |
| "rewards/margins": 0.004810371180064976, |
| "rewards/rejected": -0.010109457233920694, |
| "step": 20 |
| }, |
| { |
| "entropy": 0.48414079993963244, |
| "epoch": 0.16, |
| "grad_norm": 1.8482831716537476, |
| "learning_rate": 1.3333333333333334e-06, |
| "logits/chosen": -22.202938238943783, |
| "logits/rejected": -22.19768346061795, |
| "logps/chosen": -19.65427610874176, |
| "logps/rejected": -20.209699010849, |
| "loss": 0.7, |
| "mean_token_accuracy": 0.81201171875, |
| "num_tokens": 754241.0, |
| "rewards/accuracies": 0.509375, |
| "rewards/chosen": 0.01277370861498639, |
| "rewards/margins": 0.010195834364276379, |
| "rewards/rejected": 0.0025778753915801644, |
| "step": 25 |
| }, |
| { |
| "entropy": 0.484764963388443, |
| "epoch": 0.192, |
| "grad_norm": 1.9571627378463745, |
| "learning_rate": 1.6111111111111113e-06, |
| "logits/chosen": -22.104516369771968, |
| "logits/rejected": -22.102801838681415, |
| "logps/chosen": -19.943830585479738, |
| "logps/rejected": -20.305078315734864, |
| "loss": 0.703125, |
| "mean_token_accuracy": 0.8138671875, |
| "num_tokens": 905052.0, |
| "rewards/accuracies": 0.50625, |
| "rewards/chosen": 0.008338822494260966, |
| "rewards/margins": 0.005530653474852443, |
| "rewards/rejected": 0.0028081687516532837, |
| "step": 30 |
| }, |
| { |
| "entropy": 0.49777799248695376, |
| "epoch": 0.224, |
| "grad_norm": 2.0267391204833984, |
| "learning_rate": 1.888888888888889e-06, |
| "logits/chosen": -22.11836396257381, |
| "logits/rejected": -22.12094008856966, |
| "logps/chosen": -20.85767595767975, |
| "logps/rejected": -21.652758646011353, |
| "loss": 0.703125, |
| "mean_token_accuracy": 0.80615234375, |
| "num_tokens": 1055777.0, |
| "rewards/accuracies": 0.484375, |
| "rewards/chosen": -0.0003066336968913674, |
| "rewards/margins": -0.0023811923107132315, |
| "rewards/rejected": 0.0020745589630678297, |
| "step": 35 |
| }, |
| { |
| "entropy": 0.49067704007029533, |
| "epoch": 0.256, |
| "grad_norm": 2.076341390609741, |
| "learning_rate": 2.166666666666667e-06, |
| "logits/chosen": -22.263810676327857, |
| "logits/rejected": -22.261905347629245, |
| "logps/chosen": -19.70648772716522, |
| "logps/rejected": -20.510993862152098, |
| "loss": 0.709375, |
| "mean_token_accuracy": 0.8134765625, |
| "num_tokens": 1206478.0, |
| "rewards/accuracies": 0.51875, |
| "rewards/chosen": 0.006574534391984343, |
| "rewards/margins": -9.911635424941778e-05, |
| "rewards/rejected": 0.006673651188611985, |
| "step": 40 |
| }, |
| { |
| "entropy": 0.4931428723037243, |
| "epoch": 0.288, |
| "grad_norm": 1.9469462633132935, |
| "learning_rate": 2.4444444444444447e-06, |
| "logits/chosen": -22.04721452751252, |
| "logits/rejected": -22.085250743425505, |
| "logps/chosen": -19.638576602935792, |
| "logps/rejected": -20.6137903213501, |
| "loss": 0.7, |
| "mean_token_accuracy": 0.8125, |
| "num_tokens": 1357247.0, |
| "rewards/accuracies": 0.546875, |
| "rewards/chosen": 0.005889558521448635, |
| "rewards/margins": 0.015255183423869311, |
| "rewards/rejected": -0.009365626377984881, |
| "step": 45 |
| }, |
| { |
| "entropy": 0.4913199804723263, |
| "epoch": 0.32, |
| "grad_norm": 2.1868698596954346, |
| "learning_rate": 2.7222222222222224e-06, |
| "logits/chosen": -22.024245798033522, |
| "logits/rejected": -22.061843201087843, |
| "logps/chosen": -19.679979467391966, |
| "logps/rejected": -20.141611862182618, |
| "loss": 0.7125, |
| "mean_token_accuracy": 0.81142578125, |
| "num_tokens": 1508094.0, |
| "rewards/accuracies": 0.503125, |
| "rewards/chosen": -0.004460769856814295, |
| "rewards/margins": -0.00023858911590650677, |
| "rewards/rejected": -0.004222179052885622, |
| "step": 50 |
| }, |
| { |
| "entropy": 0.4987829826772213, |
| "epoch": 0.352, |
| "grad_norm": 1.8288943767547607, |
| "learning_rate": 3e-06, |
| "logits/chosen": -22.059007767463708, |
| "logits/rejected": -22.106842164056587, |
| "logps/chosen": -20.056474995613097, |
| "logps/rejected": -20.969518756866456, |
| "loss": 0.703125, |
| "mean_token_accuracy": 0.81142578125, |
| "num_tokens": 1658737.0, |
| "rewards/accuracies": 0.51875, |
| "rewards/chosen": 0.003189983183983713, |
| "rewards/margins": 0.009061455307528377, |
| "rewards/rejected": -0.005871471913997084, |
| "step": 55 |
| }, |
| { |
| "entropy": 0.49785798266530035, |
| "epoch": 0.384, |
| "grad_norm": 2.120474100112915, |
| "learning_rate": 3.277777777777778e-06, |
| "logits/chosen": -22.13838177852619, |
| "logits/rejected": -22.135088812769357, |
| "logps/chosen": -19.78649139404297, |
| "logps/rejected": -19.776062154769896, |
| "loss": 0.715625, |
| "mean_token_accuracy": 0.8115234375, |
| "num_tokens": 1809459.0, |
| "rewards/accuracies": 0.521875, |
| "rewards/chosen": 0.010127055528573691, |
| "rewards/margins": -0.0032248127507045867, |
| "rewards/rejected": 0.013351869536563754, |
| "step": 60 |
| }, |
| { |
| "entropy": 0.4799042470753193, |
| "epoch": 0.416, |
| "grad_norm": 2.173492908477783, |
| "learning_rate": 3.555555555555556e-06, |
| "logits/chosen": -22.193771188363314, |
| "logits/rejected": -22.25398577861252, |
| "logps/chosen": -20.205195355415345, |
| "logps/rejected": -20.15079708099365, |
| "loss": 0.7, |
| "mean_token_accuracy": 0.8076171875, |
| "num_tokens": 1960204.0, |
| "rewards/accuracies": 0.5375, |
| "rewards/chosen": 0.00017115732189267873, |
| "rewards/margins": 0.00880272446665913, |
| "rewards/rejected": -0.008631565980613232, |
| "step": 65 |
| }, |
| { |
| "entropy": 0.49170012697577475, |
| "epoch": 0.448, |
| "grad_norm": 2.054593086242676, |
| "learning_rate": 3.833333333333334e-06, |
| "logits/chosen": -22.02455824939806, |
| "logits/rejected": -22.104194944881176, |
| "logps/chosen": -20.092977499961854, |
| "logps/rejected": -20.04177131652832, |
| "loss": 0.7, |
| "mean_token_accuracy": 0.812109375, |
| "num_tokens": 2111028.0, |
| "rewards/accuracies": 0.509375, |
| "rewards/chosen": -0.007686619588639587, |
| "rewards/margins": 0.00808120525907725, |
| "rewards/rejected": -0.015767824207432567, |
| "step": 70 |
| }, |
| { |
| "entropy": 0.5046241872012616, |
| "epoch": 0.48, |
| "grad_norm": 2.0629024505615234, |
| "learning_rate": 4.111111111111111e-06, |
| "logits/chosen": -22.23320417771682, |
| "logits/rejected": -22.248176027217827, |
| "logps/chosen": -19.650324487686156, |
| "logps/rejected": -20.09525876045227, |
| "loss": 0.7, |
| "mean_token_accuracy": 0.809765625, |
| "num_tokens": 2261651.0, |
| "rewards/accuracies": 0.50625, |
| "rewards/chosen": 0.003265488403849304, |
| "rewards/margins": 0.009193273540586234, |
| "rewards/rejected": -0.005927785782841965, |
| "step": 75 |
| }, |
| { |
| "entropy": 0.4862229280173779, |
| "epoch": 0.512, |
| "grad_norm": 2.1042582988739014, |
| "learning_rate": 4.388888888888889e-06, |
| "logits/chosen": -22.160429093697452, |
| "logits/rejected": -22.1672622762626, |
| "logps/chosen": -19.360494756698607, |
| "logps/rejected": -20.44197759628296, |
| "loss": 0.69375, |
| "mean_token_accuracy": 0.814453125, |
| "num_tokens": 2412497.0, |
| "rewards/accuracies": 0.534375, |
| "rewards/chosen": 0.008373283955734223, |
| "rewards/margins": 0.017958746757358313, |
| "rewards/rejected": -0.009585461823735385, |
| "step": 80 |
| }, |
| { |
| "entropy": 0.5018395893275738, |
| "epoch": 0.544, |
| "grad_norm": 2.3265228271484375, |
| "learning_rate": 4.666666666666667e-06, |
| "logits/chosen": -22.239442740620493, |
| "logits/rejected": -22.282412534148044, |
| "logps/chosen": -19.815891814231872, |
| "logps/rejected": -21.30949659347534, |
| "loss": 0.69375, |
| "mean_token_accuracy": 0.8076171875, |
| "num_tokens": 2563322.0, |
| "rewards/accuracies": 0.540625, |
| "rewards/chosen": 0.013158861652482302, |
| "rewards/margins": 0.01510822488926351, |
| "rewards/rejected": -0.0019493644009344279, |
| "step": 85 |
| }, |
| { |
| "entropy": 0.48808433637022974, |
| "epoch": 0.576, |
| "grad_norm": 2.2838714122772217, |
| "learning_rate": 4.944444444444445e-06, |
| "logits/chosen": -22.12235499121274, |
| "logits/rejected": -22.14934863024158, |
| "logps/chosen": -20.513992166519166, |
| "logps/rejected": -20.16647353172302, |
| "loss": 0.690625, |
| "mean_token_accuracy": 0.80439453125, |
| "num_tokens": 2714203.0, |
| "rewards/accuracies": 0.540625, |
| "rewards/chosen": 0.0007259349338710308, |
| "rewards/margins": 0.019141921447589992, |
| "rewards/rejected": -0.01841598708415404, |
| "step": 90 |
| }, |
| { |
| "entropy": 0.4885494977235794, |
| "epoch": 0.608, |
| "grad_norm": 2.232922077178955, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.059512921091773, |
| "logits/rejected": -22.058729395108728, |
| "logps/chosen": -20.430141592025755, |
| "logps/rejected": -20.802348947525026, |
| "loss": 0.6875, |
| "mean_token_accuracy": 0.805859375, |
| "num_tokens": 2865115.0, |
| "rewards/accuracies": 0.58125, |
| "rewards/chosen": 0.0026725645060651005, |
| "rewards/margins": 0.032205825485289094, |
| "rewards/rejected": -0.029533259803429246, |
| "step": 95 |
| }, |
| { |
| "entropy": 0.4910079173743725, |
| "epoch": 0.64, |
| "grad_norm": 2.510983467102051, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.17147311839803, |
| "logits/rejected": -22.163630204938936, |
| "logps/chosen": -19.536120057106018, |
| "logps/rejected": -21.24213252067566, |
| "loss": 0.684375, |
| "mean_token_accuracy": 0.81708984375, |
| "num_tokens": 3015922.0, |
| "rewards/accuracies": 0.571875, |
| "rewards/chosen": -0.00933385007083416, |
| "rewards/margins": 0.02543017906136811, |
| "rewards/rejected": -0.034764029062353076, |
| "step": 100 |
| }, |
| { |
| "entropy": 0.4970912903547287, |
| "epoch": 0.672, |
| "grad_norm": 2.6018476486206055, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.280119252316823, |
| "logits/rejected": -22.257310908846645, |
| "logps/chosen": -19.652676916122438, |
| "logps/rejected": -21.870537567138673, |
| "loss": 0.68125, |
| "mean_token_accuracy": 0.81005859375, |
| "num_tokens": 3166667.0, |
| "rewards/accuracies": 0.575, |
| "rewards/chosen": -0.015339687454979867, |
| "rewards/margins": 0.04144090571207926, |
| "rewards/rejected": -0.056780594179872426, |
| "step": 105 |
| }, |
| { |
| "entropy": 0.47699177637696266, |
| "epoch": 0.704, |
| "grad_norm": 2.3537445068359375, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.11125563760093, |
| "logits/rejected": -22.160087128931593, |
| "logps/chosen": -19.682578992843627, |
| "logps/rejected": -20.84054789543152, |
| "loss": 0.690625, |
| "mean_token_accuracy": 0.8078125, |
| "num_tokens": 3317593.0, |
| "rewards/accuracies": 0.55, |
| "rewards/chosen": -0.03268031526822597, |
| "rewards/margins": 0.030540811363607645, |
| "rewards/rejected": -0.06322112721391022, |
| "step": 110 |
| }, |
| { |
| "entropy": 0.47647400125861167, |
| "epoch": 0.736, |
| "grad_norm": 2.464075803756714, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.225147799120446, |
| "logits/rejected": -22.265246610768973, |
| "logps/chosen": -20.503922939300537, |
| "logps/rejected": -20.839526748657228, |
| "loss": 0.696875, |
| "mean_token_accuracy": 0.80546875, |
| "num_tokens": 3468413.0, |
| "rewards/accuracies": 0.540625, |
| "rewards/chosen": -0.05282301185652614, |
| "rewards/margins": 0.024913082993589342, |
| "rewards/rejected": -0.07773609678260981, |
| "step": 115 |
| }, |
| { |
| "entropy": 0.4742073155939579, |
| "epoch": 0.768, |
| "grad_norm": 2.947187662124634, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.123511695150505, |
| "logits/rejected": -22.106136497661762, |
| "logps/chosen": -20.43842396736145, |
| "logps/rejected": -22.503919935226442, |
| "loss": 0.66875, |
| "mean_token_accuracy": 0.809375, |
| "num_tokens": 3619359.0, |
| "rewards/accuracies": 0.646875, |
| "rewards/chosen": -0.05819861753843725, |
| "rewards/margins": 0.0729756151791662, |
| "rewards/rejected": -0.13117423257790506, |
| "step": 120 |
| }, |
| { |
| "entropy": 0.48513352796435355, |
| "epoch": 0.8, |
| "grad_norm": 3.327395439147949, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.33309765822061, |
| "logits/rejected": -22.335630287901566, |
| "logps/chosen": -21.328091526031493, |
| "logps/rejected": -22.355806136131285, |
| "loss": 0.678125, |
| "mean_token_accuracy": 0.79921875, |
| "num_tokens": 3770090.0, |
| "rewards/accuracies": 0.584375, |
| "rewards/chosen": -0.08635386556852609, |
| "rewards/margins": 0.04549391316249966, |
| "rewards/rejected": -0.1318477805238217, |
| "step": 125 |
| }, |
| { |
| "entropy": 0.47725208178162576, |
| "epoch": 0.832, |
| "grad_norm": 3.1617085933685303, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.251881041175228, |
| "logits/rejected": -22.278605702271182, |
| "logps/chosen": -21.479117941856384, |
| "logps/rejected": -24.5278311252594, |
| "loss": 0.65625, |
| "mean_token_accuracy": 0.79921875, |
| "num_tokens": 3920936.0, |
| "rewards/accuracies": 0.646875, |
| "rewards/chosen": -0.07750612327363342, |
| "rewards/margins": 0.08405543565750122, |
| "rewards/rejected": -0.16156156021170318, |
| "step": 130 |
| }, |
| { |
| "entropy": 0.4841243714094162, |
| "epoch": 0.864, |
| "grad_norm": 2.883666515350342, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.419983687952598, |
| "logits/rejected": -22.410926742380024, |
| "logps/chosen": -23.444294929504395, |
| "logps/rejected": -23.638019704818724, |
| "loss": 0.703125, |
| "mean_token_accuracy": 0.78447265625, |
| "num_tokens": 4071660.0, |
| "rewards/accuracies": 0.51875, |
| "rewards/chosen": -0.1496299125254154, |
| "rewards/margins": 0.016162697970867158, |
| "rewards/rejected": -0.16579261235892773, |
| "step": 135 |
| }, |
| { |
| "entropy": 0.46966075524687767, |
| "epoch": 0.896, |
| "grad_norm": 3.2078323364257812, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.20780392528614, |
| "logits/rejected": -22.198430081641188, |
| "logps/chosen": -21.959484767913818, |
| "logps/rejected": -23.970044946670534, |
| "loss": 0.659375, |
| "mean_token_accuracy": 0.7978515625, |
| "num_tokens": 4222668.0, |
| "rewards/accuracies": 0.621875, |
| "rewards/chosen": -0.11726782361511141, |
| "rewards/margins": 0.08751830733381212, |
| "rewards/rejected": -0.20478613227605819, |
| "step": 140 |
| }, |
| { |
| "entropy": 0.4639404661953449, |
| "epoch": 0.928, |
| "grad_norm": 3.049100637435913, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.30407070812631, |
| "logits/rejected": -22.308133100070556, |
| "logps/chosen": -23.085755348205566, |
| "logps/rejected": -23.3309823513031, |
| "loss": 0.678125, |
| "mean_token_accuracy": 0.79306640625, |
| "num_tokens": 4373628.0, |
| "rewards/accuracies": 0.609375, |
| "rewards/chosen": -0.14029920333996415, |
| "rewards/margins": 0.0573499001795426, |
| "rewards/rejected": -0.19764910554513335, |
| "step": 145 |
| }, |
| { |
| "entropy": 0.4758478693664074, |
| "epoch": 0.96, |
| "grad_norm": 3.1753616333007812, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.2644950165088, |
| "logits/rejected": -22.290626382527574, |
| "logps/chosen": -21.532292556762695, |
| "logps/rejected": -24.28874592781067, |
| "loss": 0.665625, |
| "mean_token_accuracy": 0.80068359375, |
| "num_tokens": 4524625.0, |
| "rewards/accuracies": 0.64375, |
| "rewards/chosen": -0.10375504384282977, |
| "rewards/margins": 0.09158247401937843, |
| "rewards/rejected": -0.195337519608438, |
| "step": 150 |
| }, |
| { |
| "entropy": 0.47727343812584877, |
| "epoch": 0.992, |
| "grad_norm": 3.371931791305542, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.479302530855573, |
| "logits/rejected": -22.512217568177327, |
| "logps/chosen": -22.31882240772247, |
| "logps/rejected": -25.144134855270387, |
| "loss": 0.659375, |
| "mean_token_accuracy": 0.79658203125, |
| "num_tokens": 4675440.0, |
| "rewards/accuracies": 0.65, |
| "rewards/chosen": -0.1450558063108474, |
| "rewards/margins": 0.10969495181925595, |
| "rewards/rejected": -0.254750759806484, |
| "step": 155 |
| }, |
| { |
| "entropy": 0.4952109605073929, |
| "epoch": 1.0192, |
| "grad_norm": 3.1564483642578125, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.611952736373812, |
| "logits/rejected": -22.593619937797953, |
| "logps/chosen": -23.264904302709244, |
| "logps/rejected": -25.83666016073788, |
| "loss": 0.65625, |
| "mean_token_accuracy": 0.7868795955882353, |
| "num_tokens": 4803525.0, |
| "rewards/accuracies": 0.6360294117647058, |
| "rewards/chosen": -0.17913854242685964, |
| "rewards/margins": 0.09600255726014867, |
| "rewards/rejected": -0.27514110154965343, |
| "step": 160 |
| }, |
| { |
| "entropy": 0.48518902286887167, |
| "epoch": 1.0512, |
| "grad_norm": 3.362935781478882, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.55399976595566, |
| "logits/rejected": -22.575475620744648, |
| "logps/chosen": -23.573278665542603, |
| "logps/rejected": -28.34456162452698, |
| "loss": 0.625, |
| "mean_token_accuracy": 0.7837890625, |
| "num_tokens": 4954303.0, |
| "rewards/accuracies": 0.675, |
| "rewards/chosen": -0.20690367696806788, |
| "rewards/margins": 0.17821999453008175, |
| "rewards/rejected": -0.38512366600334647, |
| "step": 165 |
| }, |
| { |
| "entropy": 0.4884237475693226, |
| "epoch": 1.0832, |
| "grad_norm": 3.8610448837280273, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.67536226499875, |
| "logits/rejected": -22.665057944953894, |
| "logps/chosen": -26.298223447799682, |
| "logps/rejected": -29.633293247222902, |
| "loss": 0.646875, |
| "mean_token_accuracy": 0.76767578125, |
| "num_tokens": 5105160.0, |
| "rewards/accuracies": 0.64375, |
| "rewards/chosen": -0.2955171593464911, |
| "rewards/margins": 0.1583223425783217, |
| "rewards/rejected": -0.45383950099349024, |
| "step": 170 |
| }, |
| { |
| "entropy": 0.47854348421096804, |
| "epoch": 1.1152, |
| "grad_norm": 3.996198892593384, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.701960730557307, |
| "logits/rejected": -22.730720019025007, |
| "logps/chosen": -26.010914039611816, |
| "logps/rejected": -28.8562481880188, |
| "loss": 0.65625, |
| "mean_token_accuracy": 0.7728515625, |
| "num_tokens": 5255962.0, |
| "rewards/accuracies": 0.615625, |
| "rewards/chosen": -0.3181323821656406, |
| "rewards/margins": 0.14694206467829646, |
| "rewards/rejected": -0.46507445685565474, |
| "step": 175 |
| }, |
| { |
| "entropy": 0.4799050338566303, |
| "epoch": 1.1472, |
| "grad_norm": 3.878946542739868, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.599918768763693, |
| "logits/rejected": -22.591403243552286, |
| "logps/chosen": -25.680469036102295, |
| "logps/rejected": -29.167397260665894, |
| "loss": 0.64375, |
| "mean_token_accuracy": 0.76953125, |
| "num_tokens": 5406857.0, |
| "rewards/accuracies": 0.68125, |
| "rewards/chosen": -0.2874727608636022, |
| "rewards/margins": 0.15593584245070816, |
| "rewards/rejected": -0.4434086026623845, |
| "step": 180 |
| }, |
| { |
| "entropy": 0.4789674110710621, |
| "epoch": 1.1792, |
| "grad_norm": 4.02281379699707, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.7098062225217, |
| "logits/rejected": -22.758495942388805, |
| "logps/chosen": -26.099153232574462, |
| "logps/rejected": -29.699472284317018, |
| "loss": 0.64375, |
| "mean_token_accuracy": 0.76728515625, |
| "num_tokens": 5557673.0, |
| "rewards/accuracies": 0.675, |
| "rewards/chosen": -0.30420471653342246, |
| "rewards/margins": 0.16913537541404366, |
| "rewards/rejected": -0.47334009334445, |
| "step": 185 |
| }, |
| { |
| "entropy": 0.4524506017565727, |
| "epoch": 1.2112, |
| "grad_norm": 4.14384126663208, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.953460338188005, |
| "logits/rejected": -22.926287440218086, |
| "logps/chosen": -25.489940786361693, |
| "logps/rejected": -30.252872467041016, |
| "loss": 0.621875, |
| "mean_token_accuracy": 0.7759765625, |
| "num_tokens": 5708475.0, |
| "rewards/accuracies": 0.69375, |
| "rewards/chosen": -0.3161001421511173, |
| "rewards/margins": 0.21423221128061415, |
| "rewards/rejected": -0.5303323417901993, |
| "step": 190 |
| }, |
| { |
| "entropy": 0.4781226672232151, |
| "epoch": 1.2432, |
| "grad_norm": 3.518662214279175, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.807658279762883, |
| "logits/rejected": -22.8690476372636, |
| "logps/chosen": -27.676951026916505, |
| "logps/rejected": -33.29890904426575, |
| "loss": 0.6125, |
| "mean_token_accuracy": 0.766015625, |
| "num_tokens": 5859149.0, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": -0.3270489677786827, |
| "rewards/margins": 0.25331703033298253, |
| "rewards/rejected": -0.5803659938275814, |
| "step": 195 |
| }, |
| { |
| "entropy": 0.44460521563887595, |
| "epoch": 1.2752, |
| "grad_norm": 4.291970252990723, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.955977295729703, |
| "logits/rejected": -22.999704916990474, |
| "logps/chosen": -28.337022066116333, |
| "logps/rejected": -33.69034452438355, |
| "loss": 0.621875, |
| "mean_token_accuracy": 0.7611328125, |
| "num_tokens": 6010104.0, |
| "rewards/accuracies": 0.68125, |
| "rewards/chosen": -0.43169451467692854, |
| "rewards/margins": 0.24144710903055966, |
| "rewards/rejected": -0.6731416285037994, |
| "step": 200 |
| }, |
| { |
| "entropy": 0.4462406925857067, |
| "epoch": 1.3072, |
| "grad_norm": 3.7910988330841064, |
| "learning_rate": 5e-06, |
| "logits/chosen": -22.96260678784111, |
| "logits/rejected": -23.012315910957877, |
| "logps/chosen": -27.331138610839844, |
| "logps/rejected": -33.460497331619266, |
| "loss": 0.590625, |
| "mean_token_accuracy": 0.76396484375, |
| "num_tokens": 6160901.0, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": -0.38667107187211514, |
| "rewards/margins": 0.2907294140197337, |
| "rewards/rejected": -0.6774004846811295, |
| "step": 205 |
| }, |
| { |
| "entropy": 0.4538860201835632, |
| "epoch": 1.3392, |
| "grad_norm": 4.863774299621582, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.136923480368573, |
| "logits/rejected": -23.17609682910406, |
| "logps/chosen": -29.093460178375246, |
| "logps/rejected": -34.68924231529236, |
| "loss": 0.603125, |
| "mean_token_accuracy": 0.75537109375, |
| "num_tokens": 6311708.0, |
| "rewards/accuracies": 0.7, |
| "rewards/chosen": -0.4789995074272156, |
| "rewards/margins": 0.2709728317335248, |
| "rewards/rejected": -0.7499723382294178, |
| "step": 210 |
| }, |
| { |
| "entropy": 0.4540509045124054, |
| "epoch": 1.3712, |
| "grad_norm": 5.28690767288208, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.211061563532006, |
| "logits/rejected": -23.209657317324297, |
| "logps/chosen": -32.30440850257874, |
| "logps/rejected": -36.89135007858276, |
| "loss": 0.640625, |
| "mean_token_accuracy": 0.74501953125, |
| "num_tokens": 6462531.0, |
| "rewards/accuracies": 0.628125, |
| "rewards/chosen": -0.6221090689301491, |
| "rewards/margins": 0.2135551360435784, |
| "rewards/rejected": -0.8356641985476017, |
| "step": 215 |
| }, |
| { |
| "entropy": 0.46628558039665224, |
| "epoch": 1.4032, |
| "grad_norm": 4.862166404724121, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.125505146516055, |
| "logits/rejected": -23.140280610637664, |
| "logps/chosen": -32.66754117012024, |
| "logps/rejected": -37.275456285476686, |
| "loss": 0.61875, |
| "mean_token_accuracy": 0.7412109375, |
| "num_tokens": 6613262.0, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": -0.6013654366135597, |
| "rewards/margins": 0.25150299090892075, |
| "rewards/rejected": -0.8528684288263321, |
| "step": 220 |
| }, |
| { |
| "entropy": 0.4511294513940811, |
| "epoch": 1.4352, |
| "grad_norm": 4.880625247955322, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.099047661571763, |
| "logits/rejected": -23.113464852210036, |
| "logps/chosen": -31.25849027633667, |
| "logps/rejected": -35.820988941192624, |
| "loss": 0.6375, |
| "mean_token_accuracy": 0.747265625, |
| "num_tokens": 6764182.0, |
| "rewards/accuracies": 0.653125, |
| "rewards/chosen": -0.5661171391606331, |
| "rewards/margins": 0.21538838678970934, |
| "rewards/rejected": -0.7815055266022682, |
| "step": 225 |
| }, |
| { |
| "entropy": 0.4352009750902653, |
| "epoch": 1.4672, |
| "grad_norm": 4.09095573425293, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.31776165642016, |
| "logits/rejected": -23.295068592633335, |
| "logps/chosen": -27.809169149398805, |
| "logps/rejected": -33.70988502502441, |
| "loss": 0.60625, |
| "mean_token_accuracy": 0.76376953125, |
| "num_tokens": 6915081.0, |
| "rewards/accuracies": 0.7125, |
| "rewards/chosen": -0.43004697374999523, |
| "rewards/margins": 0.2860319158062339, |
| "rewards/rejected": -0.7160788841545582, |
| "step": 230 |
| }, |
| { |
| "entropy": 0.4373820908367634, |
| "epoch": 1.4992, |
| "grad_norm": 4.417893409729004, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.31567756588704, |
| "logits/rejected": -23.33086037977852, |
| "logps/chosen": -29.391495418548583, |
| "logps/rejected": -34.44461603164673, |
| "loss": 0.603125, |
| "mean_token_accuracy": 0.76357421875, |
| "num_tokens": 7065830.0, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": -0.4352296255528927, |
| "rewards/margins": 0.2669131119735539, |
| "rewards/rejected": -0.7021427422761917, |
| "step": 235 |
| }, |
| { |
| "entropy": 0.4345704182982445, |
| "epoch": 1.5312000000000001, |
| "grad_norm": 4.383578777313232, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.406836627793478, |
| "logits/rejected": -23.400130394754758, |
| "logps/chosen": -28.19483027458191, |
| "logps/rejected": -33.56565489768982, |
| "loss": 0.6125, |
| "mean_token_accuracy": 0.76123046875, |
| "num_tokens": 7216644.0, |
| "rewards/accuracies": 0.675, |
| "rewards/chosen": -0.4238139152526855, |
| "rewards/margins": 0.2505914020352066, |
| "rewards/rejected": -0.6744053177535534, |
| "step": 240 |
| }, |
| { |
| "entropy": 0.4306245781481266, |
| "epoch": 1.5632000000000001, |
| "grad_norm": 4.466887474060059, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.359672226421697, |
| "logits/rejected": -23.353592900207566, |
| "logps/chosen": -28.442044401168822, |
| "logps/rejected": -33.97496213912964, |
| "loss": 0.6125, |
| "mean_token_accuracy": 0.76689453125, |
| "num_tokens": 7367600.0, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": -0.42358348816633223, |
| "rewards/margins": 0.2510149137116969, |
| "rewards/rejected": -0.6745983995497227, |
| "step": 245 |
| }, |
| { |
| "entropy": 0.43906090036034584, |
| "epoch": 1.5952, |
| "grad_norm": 5.964234352111816, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.431583133766946, |
| "logits/rejected": -23.5323691640834, |
| "logps/chosen": -28.33654637336731, |
| "logps/rejected": -35.88848881721496, |
| "loss": 0.571875, |
| "mean_token_accuracy": 0.76513671875, |
| "num_tokens": 7518245.0, |
| "rewards/accuracies": 0.778125, |
| "rewards/chosen": -0.42815797589719296, |
| "rewards/margins": 0.34710690919309856, |
| "rewards/rejected": -0.7752648875117302, |
| "step": 250 |
| }, |
| { |
| "entropy": 0.44010329619050026, |
| "epoch": 1.6272, |
| "grad_norm": 4.843886375427246, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.34347842325234, |
| "logits/rejected": -23.363990905973242, |
| "logps/chosen": -30.963105154037475, |
| "logps/rejected": -36.999150133132936, |
| "loss": 0.584375, |
| "mean_token_accuracy": 0.75361328125, |
| "num_tokens": 7669165.0, |
| "rewards/accuracies": 0.75625, |
| "rewards/chosen": -0.4892259083688259, |
| "rewards/margins": 0.31986095774918794, |
| "rewards/rejected": -0.809086874872446, |
| "step": 255 |
| }, |
| { |
| "entropy": 0.40867728665471076, |
| "epoch": 1.6592, |
| "grad_norm": 4.955322742462158, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.493765758865464, |
| "logits/rejected": -23.537847630306974, |
| "logps/chosen": -31.53315477371216, |
| "logps/rejected": -38.73755555152893, |
| "loss": 0.5875, |
| "mean_token_accuracy": 0.7529296875, |
| "num_tokens": 7820058.0, |
| "rewards/accuracies": 0.728125, |
| "rewards/chosen": -0.5828495036810637, |
| "rewards/margins": 0.32151391603983936, |
| "rewards/rejected": -0.9043634235858917, |
| "step": 260 |
| }, |
| { |
| "entropy": 0.41678572744131087, |
| "epoch": 1.6912, |
| "grad_norm": 4.957008361816406, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.601254286378513, |
| "logits/rejected": -23.63242224733047, |
| "logps/chosen": -31.265224933624268, |
| "logps/rejected": -38.0910548210144, |
| "loss": 0.584375, |
| "mean_token_accuracy": 0.750390625, |
| "num_tokens": 7970901.0, |
| "rewards/accuracies": 0.740625, |
| "rewards/chosen": -0.5970524445176124, |
| "rewards/margins": 0.33683429267257453, |
| "rewards/rejected": -0.9338867455720902, |
| "step": 265 |
| }, |
| { |
| "entropy": 0.40209152102470397, |
| "epoch": 1.7231999999999998, |
| "grad_norm": 6.079311370849609, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.626515837323353, |
| "logits/rejected": -23.647804868067404, |
| "logps/chosen": -31.936182737350464, |
| "logps/rejected": -38.54815468788147, |
| "loss": 0.603125, |
| "mean_token_accuracy": 0.7517578125, |
| "num_tokens": 8121709.0, |
| "rewards/accuracies": 0.696875, |
| "rewards/chosen": -0.6252030048519372, |
| "rewards/margins": 0.3070303828455508, |
| "rewards/rejected": -0.9322333797812462, |
| "step": 270 |
| }, |
| { |
| "entropy": 0.41649706438183787, |
| "epoch": 1.7551999999999999, |
| "grad_norm": 4.9190754890441895, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.649771643838314, |
| "logits/rejected": -23.634129381604147, |
| "logps/chosen": -31.480729818344116, |
| "logps/rejected": -40.48275279998779, |
| "loss": 0.55, |
| "mean_token_accuracy": 0.7576171875, |
| "num_tokens": 8272547.0, |
| "rewards/accuracies": 0.76875, |
| "rewards/chosen": -0.587709330022335, |
| "rewards/margins": 0.4087777270004153, |
| "rewards/rejected": -0.9964870691299439, |
| "step": 275 |
| }, |
| { |
| "entropy": 0.39424559473991394, |
| "epoch": 1.7872, |
| "grad_norm": 4.63533353805542, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.807110448453876, |
| "logits/rejected": -23.871575425736623, |
| "logps/chosen": -31.923841047286988, |
| "logps/rejected": -40.58993577957153, |
| "loss": 0.571875, |
| "mean_token_accuracy": 0.75537109375, |
| "num_tokens": 8423430.0, |
| "rewards/accuracies": 0.76875, |
| "rewards/chosen": -0.6693632058799267, |
| "rewards/margins": 0.361633213609457, |
| "rewards/rejected": -1.0309964150190354, |
| "step": 280 |
| }, |
| { |
| "entropy": 0.3998428381979465, |
| "epoch": 1.8192, |
| "grad_norm": 4.982107639312744, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.800428666834968, |
| "logits/rejected": -23.827003697269177, |
| "logps/chosen": -33.038239574432374, |
| "logps/rejected": -42.419326591491696, |
| "loss": 0.55625, |
| "mean_token_accuracy": 0.745703125, |
| "num_tokens": 8574325.0, |
| "rewards/accuracies": 0.790625, |
| "rewards/chosen": -0.6933430828154087, |
| "rewards/margins": 0.43362359665334227, |
| "rewards/rejected": -1.1269666835665704, |
| "step": 285 |
| }, |
| { |
| "entropy": 0.4138856314122677, |
| "epoch": 1.8512, |
| "grad_norm": 4.876551151275635, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.858734396678237, |
| "logits/rejected": -23.815305281175295, |
| "logps/chosen": -34.78357348442078, |
| "logps/rejected": -43.67720069885254, |
| "loss": 0.553125, |
| "mean_token_accuracy": 0.73447265625, |
| "num_tokens": 8725022.0, |
| "rewards/accuracies": 0.753125, |
| "rewards/chosen": -0.7325984433293342, |
| "rewards/margins": 0.44189396053552626, |
| "rewards/rejected": -1.1744924068450928, |
| "step": 290 |
| }, |
| { |
| "entropy": 0.4119267471134663, |
| "epoch": 1.8832, |
| "grad_norm": 5.815010070800781, |
| "learning_rate": 5e-06, |
| "logits/chosen": -23.9709202373874, |
| "logits/rejected": -23.972968525587, |
| "logps/chosen": -35.91455888748169, |
| "logps/rejected": -46.68118925094605, |
| "loss": 0.55, |
| "mean_token_accuracy": 0.726953125, |
| "num_tokens": 8875875.0, |
| "rewards/accuracies": 0.74375, |
| "rewards/chosen": -0.8230781421065331, |
| "rewards/margins": 0.4535051151178777, |
| "rewards/rejected": -1.276583270728588, |
| "step": 295 |
| }, |
| { |
| "entropy": 0.4055053681135178, |
| "epoch": 1.9152, |
| "grad_norm": 5.092159271240234, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.129554617680718, |
| "logits/rejected": -24.17981090238822, |
| "logps/chosen": -36.72642884254456, |
| "logps/rejected": -47.98744974136353, |
| "loss": 0.528125, |
| "mean_token_accuracy": 0.73037109375, |
| "num_tokens": 9026632.0, |
| "rewards/accuracies": 0.75625, |
| "rewards/chosen": -0.8602523311972619, |
| "rewards/margins": 0.5361747246235609, |
| "rewards/rejected": -1.3964270666241645, |
| "step": 300 |
| }, |
| { |
| "entropy": 0.3946771830320358, |
| "epoch": 1.9472, |
| "grad_norm": 5.5546746253967285, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.133076721043466, |
| "logits/rejected": -24.1340199217744, |
| "logps/chosen": -38.450519704818724, |
| "logps/rejected": -50.3550687789917, |
| "loss": 0.515625, |
| "mean_token_accuracy": 0.72431640625, |
| "num_tokens": 9177411.0, |
| "rewards/accuracies": 0.790625, |
| "rewards/chosen": -0.9096269533038139, |
| "rewards/margins": 0.5736862637102604, |
| "rewards/rejected": -1.483313226699829, |
| "step": 305 |
| }, |
| { |
| "entropy": 0.3926751546561718, |
| "epoch": 1.9792, |
| "grad_norm": 5.562469482421875, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.051962810738857, |
| "logits/rejected": -24.06771397104591, |
| "logps/chosen": -39.958859825134276, |
| "logps/rejected": -49.82817888259888, |
| "loss": 0.55625, |
| "mean_token_accuracy": 0.72001953125, |
| "num_tokens": 9328310.0, |
| "rewards/accuracies": 0.7625, |
| "rewards/chosen": -0.9941851183772087, |
| "rewards/margins": 0.47167541990056633, |
| "rewards/rejected": -1.4658605247735976, |
| "step": 310 |
| }, |
| { |
| "entropy": 0.3922588790164274, |
| "epoch": 2.0064, |
| "grad_norm": 5.53153133392334, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.149734036674772, |
| "logits/rejected": -24.194233471953602, |
| "logps/chosen": -38.37042169009938, |
| "logps/rejected": -50.03265594033634, |
| "loss": 0.525, |
| "mean_token_accuracy": 0.7310431985294118, |
| "num_tokens": 9456436.0, |
| "rewards/accuracies": 0.7830882352941176, |
| "rewards/chosen": -0.9160057078389561, |
| "rewards/margins": 0.5664738780435394, |
| "rewards/rejected": -1.482479586320765, |
| "step": 315 |
| }, |
| { |
| "entropy": 0.3935938835144043, |
| "epoch": 2.0384, |
| "grad_norm": 5.870213031768799, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.206608410762072, |
| "logits/rejected": -24.217967269516983, |
| "logps/chosen": -37.65716261863709, |
| "logps/rejected": -50.2130352973938, |
| "loss": 0.496875, |
| "mean_token_accuracy": 0.733203125, |
| "num_tokens": 9607271.0, |
| "rewards/accuracies": 0.825, |
| "rewards/chosen": -0.8895024463534356, |
| "rewards/margins": 0.6046069579198956, |
| "rewards/rejected": -1.49410939514637, |
| "step": 320 |
| }, |
| { |
| "entropy": 0.38338030204176904, |
| "epoch": 2.0704, |
| "grad_norm": 5.663544654846191, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.089506385461014, |
| "logits/rejected": -24.158271784630394, |
| "logps/chosen": -37.66941123008728, |
| "logps/rejected": -51.0740345954895, |
| "loss": 0.509375, |
| "mean_token_accuracy": 0.73798828125, |
| "num_tokens": 9758188.0, |
| "rewards/accuracies": 0.796875, |
| "rewards/chosen": -0.9006093442440033, |
| "rewards/margins": 0.6113568138331175, |
| "rewards/rejected": -1.5119661852717399, |
| "step": 325 |
| }, |
| { |
| "entropy": 0.3914147213101387, |
| "epoch": 2.1024, |
| "grad_norm": 5.537415027618408, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.259089032721228, |
| "logits/rejected": -24.298371268830355, |
| "logps/chosen": -37.91493172645569, |
| "logps/rejected": -50.96251525878906, |
| "loss": 0.496875, |
| "mean_token_accuracy": 0.7294921875, |
| "num_tokens": 9908881.0, |
| "rewards/accuracies": 0.79375, |
| "rewards/chosen": -0.9152560964226722, |
| "rewards/margins": 0.6253166727721691, |
| "rewards/rejected": -1.5405727803707123, |
| "step": 330 |
| }, |
| { |
| "entropy": 0.3846339240670204, |
| "epoch": 2.1344, |
| "grad_norm": 6.073315143585205, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.268502183075707, |
| "logits/rejected": -24.253160218586256, |
| "logps/chosen": -37.65253806114197, |
| "logps/rejected": -52.06829042434693, |
| "loss": 0.4875, |
| "mean_token_accuracy": 0.73662109375, |
| "num_tokens": 10059595.0, |
| "rewards/accuracies": 0.828125, |
| "rewards/chosen": -0.8810642987489701, |
| "rewards/margins": 0.6875064481049776, |
| "rewards/rejected": -1.5685707360506058, |
| "step": 335 |
| }, |
| { |
| "entropy": 0.3822611324489117, |
| "epoch": 2.1664, |
| "grad_norm": 6.016086101531982, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.16974293764945, |
| "logits/rejected": -24.218686101364717, |
| "logps/chosen": -38.29234871864319, |
| "logps/rejected": -51.30879316329956, |
| "loss": 0.5, |
| "mean_token_accuracy": 0.73017578125, |
| "num_tokens": 10210466.0, |
| "rewards/accuracies": 0.821875, |
| "rewards/chosen": -0.9170552954077721, |
| "rewards/margins": 0.6155369689688086, |
| "rewards/rejected": -1.5325922697782517, |
| "step": 340 |
| }, |
| { |
| "entropy": 0.37890644297003745, |
| "epoch": 2.1984, |
| "grad_norm": 6.27139949798584, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.287747322942067, |
| "logits/rejected": -24.297362990626294, |
| "logps/chosen": -38.42732787132263, |
| "logps/rejected": -51.74098300933838, |
| "loss": 0.484375, |
| "mean_token_accuracy": 0.72900390625, |
| "num_tokens": 10361331.0, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": -0.9211938828229904, |
| "rewards/margins": 0.6350689101964235, |
| "rewards/rejected": -1.5562627956271171, |
| "step": 345 |
| }, |
| { |
| "entropy": 0.370716817677021, |
| "epoch": 2.2304, |
| "grad_norm": 6.485368251800537, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.28075639658453, |
| "logits/rejected": -24.284953170889242, |
| "logps/chosen": -40.34224147796631, |
| "logps/rejected": -54.90016975402832, |
| "loss": 0.4875, |
| "mean_token_accuracy": 0.72353515625, |
| "num_tokens": 10512141.0, |
| "rewards/accuracies": 0.821875, |
| "rewards/chosen": -1.054071944952011, |
| "rewards/margins": 0.6897623263299465, |
| "rewards/rejected": -1.7438342660665511, |
| "step": 350 |
| }, |
| { |
| "entropy": 0.3734958402812481, |
| "epoch": 2.2624, |
| "grad_norm": 6.332367420196533, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.383364754370497, |
| "logits/rejected": -24.377376980533874, |
| "logps/chosen": -41.408950853347775, |
| "logps/rejected": -55.78951148986816, |
| "loss": 0.484375, |
| "mean_token_accuracy": 0.716015625, |
| "num_tokens": 10662935.0, |
| "rewards/accuracies": 0.840625, |
| "rewards/chosen": -1.05933967679739, |
| "rewards/margins": 0.7176455095410347, |
| "rewards/rejected": -1.7769851803779602, |
| "step": 355 |
| }, |
| { |
| "entropy": 0.37396894916892054, |
| "epoch": 2.2944, |
| "grad_norm": 6.250120639801025, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.3733864432035, |
| "logits/rejected": -24.383639435413762, |
| "logps/chosen": -42.31962890625, |
| "logps/rejected": -57.560160636901855, |
| "loss": 0.471875, |
| "mean_token_accuracy": 0.71494140625, |
| "num_tokens": 10813756.0, |
| "rewards/accuracies": 0.796875, |
| "rewards/chosen": -1.1360176876187325, |
| "rewards/margins": 0.754439314827323, |
| "rewards/rejected": -1.8904570013284683, |
| "step": 360 |
| }, |
| { |
| "entropy": 0.3651010349392891, |
| "epoch": 2.3264, |
| "grad_norm": 6.565396308898926, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.386384145031254, |
| "logits/rejected": -24.41246286580034, |
| "logps/chosen": -41.985077953338624, |
| "logps/rejected": -57.587640571594235, |
| "loss": 0.475, |
| "mean_token_accuracy": 0.72265625, |
| "num_tokens": 10964688.0, |
| "rewards/accuracies": 0.828125, |
| "rewards/chosen": -1.120146569609642, |
| "rewards/margins": 0.7471146807074547, |
| "rewards/rejected": -1.8672612398862838, |
| "step": 365 |
| }, |
| { |
| "entropy": 0.36412821859121325, |
| "epoch": 2.3584, |
| "grad_norm": 6.449187278747559, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.49703348425451, |
| "logits/rejected": -24.49014196170392, |
| "logps/chosen": -40.445683002471924, |
| "logps/rejected": -57.763357162475586, |
| "loss": 0.45, |
| "mean_token_accuracy": 0.724609375, |
| "num_tokens": 11115515.0, |
| "rewards/accuracies": 0.84375, |
| "rewards/chosen": -1.0540059104561805, |
| "rewards/margins": 0.7839702643454075, |
| "rewards/rejected": -1.8379761695861816, |
| "step": 370 |
| }, |
| { |
| "entropy": 0.3677647367119789, |
| "epoch": 2.3904, |
| "grad_norm": 6.950628280639648, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.444762077242242, |
| "logits/rejected": -24.471458427477053, |
| "logps/chosen": -42.84081106185913, |
| "logps/rejected": -59.76805982589722, |
| "loss": 0.446875, |
| "mean_token_accuracy": 0.718359375, |
| "num_tokens": 11266282.0, |
| "rewards/accuracies": 0.840625, |
| "rewards/chosen": -1.1263013422489165, |
| "rewards/margins": 0.8474021539092064, |
| "rewards/rejected": -1.9737034916877747, |
| "step": 375 |
| }, |
| { |
| "entropy": 0.3636451981961727, |
| "epoch": 2.4224, |
| "grad_norm": 6.976022720336914, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.562865267832557, |
| "logits/rejected": -24.58747823277994, |
| "logps/chosen": -43.619415283203125, |
| "logps/rejected": -56.828695106506345, |
| "loss": 0.5125, |
| "mean_token_accuracy": 0.71640625, |
| "num_tokens": 11417029.0, |
| "rewards/accuracies": 0.7625, |
| "rewards/chosen": -1.2213508561253548, |
| "rewards/margins": 0.6378849178552628, |
| "rewards/rejected": -1.8592357635498047, |
| "step": 380 |
| }, |
| { |
| "entropy": 0.3667745187878609, |
| "epoch": 2.4544, |
| "grad_norm": 5.500574588775635, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.51088276918522, |
| "logits/rejected": -24.597310246261294, |
| "logps/chosen": -43.10864315032959, |
| "logps/rejected": -59.85773401260376, |
| "loss": 0.4375, |
| "mean_token_accuracy": 0.7189453125, |
| "num_tokens": 11567842.0, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": -1.1550978109240533, |
| "rewards/margins": 0.8523115241900087, |
| "rewards/rejected": -2.0074093639850616, |
| "step": 385 |
| }, |
| { |
| "entropy": 0.36734507903456687, |
| "epoch": 2.4864, |
| "grad_norm": 7.42901611328125, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.530855877820905, |
| "logits/rejected": -24.51802270001329, |
| "logps/chosen": -46.009543895721436, |
| "logps/rejected": -62.626318550109865, |
| "loss": 0.453125, |
| "mean_token_accuracy": 0.706640625, |
| "num_tokens": 11718639.0, |
| "rewards/accuracies": 0.840625, |
| "rewards/chosen": -1.2961086645722388, |
| "rewards/margins": 0.8299951769411564, |
| "rewards/rejected": -2.126103848218918, |
| "step": 390 |
| }, |
| { |
| "entropy": 0.3617811560630798, |
| "epoch": 2.5183999999999997, |
| "grad_norm": 7.369869709014893, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.629660384837155, |
| "logits/rejected": -24.63559780051194, |
| "logps/chosen": -47.18812131881714, |
| "logps/rejected": -66.1587459564209, |
| "loss": 0.43125, |
| "mean_token_accuracy": 0.70830078125, |
| "num_tokens": 11869456.0, |
| "rewards/accuracies": 0.846875, |
| "rewards/chosen": -1.349448035657406, |
| "rewards/margins": 0.9228871289640665, |
| "rewards/rejected": -2.272335159778595, |
| "step": 395 |
| }, |
| { |
| "entropy": 0.3719852648675442, |
| "epoch": 2.5504, |
| "grad_norm": 7.120789051055908, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.650620443016937, |
| "logits/rejected": -24.676608398171574, |
| "logps/chosen": -48.31374540328979, |
| "logps/rejected": -66.53067092895508, |
| "loss": 0.440625, |
| "mean_token_accuracy": 0.7046875, |
| "num_tokens": 12020422.0, |
| "rewards/accuracies": 0.840625, |
| "rewards/chosen": -1.4188964247703553, |
| "rewards/margins": 0.8901826798915863, |
| "rewards/rejected": -2.309079107642174, |
| "step": 400 |
| }, |
| { |
| "entropy": 0.3593819744884968, |
| "epoch": 2.5824, |
| "grad_norm": 7.498592376708984, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.574948779008928, |
| "logits/rejected": -24.60266122265609, |
| "logps/chosen": -50.710439109802245, |
| "logps/rejected": -69.25852966308594, |
| "loss": 0.440625, |
| "mean_token_accuracy": 0.6947265625, |
| "num_tokens": 12171340.0, |
| "rewards/accuracies": 0.853125, |
| "rewards/chosen": -1.523548685014248, |
| "rewards/margins": 0.940007409453392, |
| "rewards/rejected": -2.463556116819382, |
| "step": 405 |
| }, |
| { |
| "entropy": 0.3592563413083553, |
| "epoch": 2.6144, |
| "grad_norm": 6.778897762298584, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.64159831164012, |
| "logits/rejected": -24.678670204455102, |
| "logps/chosen": -49.8059003829956, |
| "logps/rejected": -67.01162490844726, |
| "loss": 0.465625, |
| "mean_token_accuracy": 0.69541015625, |
| "num_tokens": 12322068.0, |
| "rewards/accuracies": 0.815625, |
| "rewards/chosen": -1.4885912179946899, |
| "rewards/margins": 0.8339855313301087, |
| "rewards/rejected": -2.322576770186424, |
| "step": 410 |
| }, |
| { |
| "entropy": 0.3432131476700306, |
| "epoch": 2.6464, |
| "grad_norm": 6.680260181427002, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.627115599935195, |
| "logits/rejected": -24.639765450921853, |
| "logps/chosen": -45.41163873672485, |
| "logps/rejected": -64.65480012893677, |
| "loss": 0.421875, |
| "mean_token_accuracy": 0.7158203125, |
| "num_tokens": 12472938.0, |
| "rewards/accuracies": 0.84375, |
| "rewards/chosen": -1.2895102351903915, |
| "rewards/margins": 0.9426323518157005, |
| "rewards/rejected": -2.2321425855159758, |
| "step": 415 |
| }, |
| { |
| "entropy": 0.35311873704195024, |
| "epoch": 2.6784, |
| "grad_norm": 6.916601657867432, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.66342320712633, |
| "logits/rejected": -24.659541341480256, |
| "logps/chosen": -45.28245210647583, |
| "logps/rejected": -62.9191234588623, |
| "loss": 0.446875, |
| "mean_token_accuracy": 0.71318359375, |
| "num_tokens": 12623749.0, |
| "rewards/accuracies": 0.821875, |
| "rewards/chosen": -1.2987738519906997, |
| "rewards/margins": 0.8664737090468406, |
| "rewards/rejected": -2.1652475744485855, |
| "step": 420 |
| }, |
| { |
| "entropy": 0.35651103407144547, |
| "epoch": 2.7104, |
| "grad_norm": 6.897266864776611, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.64527581046648, |
| "logits/rejected": -24.724301579765644, |
| "logps/chosen": -47.109345245361325, |
| "logps/rejected": -65.39630517959594, |
| "loss": 0.440625, |
| "mean_token_accuracy": 0.70966796875, |
| "num_tokens": 12774554.0, |
| "rewards/accuracies": 0.83125, |
| "rewards/chosen": -1.348003403842449, |
| "rewards/margins": 0.9132076129317284, |
| "rewards/rejected": -2.2612110167741775, |
| "step": 425 |
| }, |
| { |
| "entropy": 0.3578870095312595, |
| "epoch": 2.7424, |
| "grad_norm": 8.14350700378418, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.786621900320974, |
| "logits/rejected": -24.867742053670764, |
| "logps/chosen": -49.39408779144287, |
| "logps/rejected": -71.00293645858764, |
| "loss": 0.40625, |
| "mean_token_accuracy": 0.7052734375, |
| "num_tokens": 12925356.0, |
| "rewards/accuracies": 0.878125, |
| "rewards/chosen": -1.4575166389346124, |
| "rewards/margins": 1.0326086923480033, |
| "rewards/rejected": -2.4901253134012222, |
| "step": 430 |
| }, |
| { |
| "entropy": 0.35039815977215766, |
| "epoch": 2.7744, |
| "grad_norm": 7.449344158172607, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.739288614190578, |
| "logits/rejected": -24.755637191371683, |
| "logps/chosen": -52.73114595413208, |
| "logps/rejected": -70.80000200271607, |
| "loss": 0.44375, |
| "mean_token_accuracy": 0.68798828125, |
| "num_tokens": 13076169.0, |
| "rewards/accuracies": 0.80625, |
| "rewards/chosen": -1.6145994812250137, |
| "rewards/margins": 0.9192891106009483, |
| "rewards/rejected": -2.5338885694742204, |
| "step": 435 |
| }, |
| { |
| "entropy": 0.34932322055101395, |
| "epoch": 2.8064, |
| "grad_norm": 6.517420291900635, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.740115388769667, |
| "logits/rejected": -24.8271093904833, |
| "logps/chosen": -53.61327123641968, |
| "logps/rejected": -77.15019493103027, |
| "loss": 0.390625, |
| "mean_token_accuracy": 0.68828125, |
| "num_tokens": 13226894.0, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": -1.6991019412875175, |
| "rewards/margins": 1.1353167057037354, |
| "rewards/rejected": -2.834418645501137, |
| "step": 440 |
| }, |
| { |
| "entropy": 0.3466518945991993, |
| "epoch": 2.8384, |
| "grad_norm": 6.438902378082275, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.829863713219332, |
| "logits/rejected": -24.815264142521432, |
| "logps/chosen": -51.42146377563476, |
| "logps/rejected": -73.04679698944092, |
| "loss": 0.403125, |
| "mean_token_accuracy": 0.69423828125, |
| "num_tokens": 13377645.0, |
| "rewards/accuracies": 0.8625, |
| "rewards/chosen": -1.6007766336202622, |
| "rewards/margins": 1.0688797503709793, |
| "rewards/rejected": -2.669656378030777, |
| "step": 445 |
| }, |
| { |
| "entropy": 0.3434488832950592, |
| "epoch": 2.8704, |
| "grad_norm": 6.985738754272461, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.70450278087584, |
| "logits/rejected": -24.740914714346296, |
| "logps/chosen": -50.460514545440674, |
| "logps/rejected": -71.26909322738648, |
| "loss": 0.4125, |
| "mean_token_accuracy": 0.70185546875, |
| "num_tokens": 13528479.0, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": -1.520570918917656, |
| "rewards/margins": 1.012993934750557, |
| "rewards/rejected": -2.5335648715496064, |
| "step": 450 |
| }, |
| { |
| "entropy": 0.3436164803802967, |
| "epoch": 2.9024, |
| "grad_norm": 6.937602996826172, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.814630885961577, |
| "logits/rejected": -24.845365227427735, |
| "logps/chosen": -48.49508094787598, |
| "logps/rejected": -70.10940446853638, |
| "loss": 0.403125, |
| "mean_token_accuracy": 0.7091796875, |
| "num_tokens": 13679288.0, |
| "rewards/accuracies": 0.878125, |
| "rewards/chosen": -1.417954184114933, |
| "rewards/margins": 1.084460623562336, |
| "rewards/rejected": -2.50241479575634, |
| "step": 455 |
| }, |
| { |
| "entropy": 0.3391841046512127, |
| "epoch": 2.9344, |
| "grad_norm": 7.7364020347595215, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.68823858488307, |
| "logits/rejected": -24.70028430740195, |
| "logps/chosen": -49.72757091522217, |
| "logps/rejected": -69.60700139999389, |
| "loss": 0.425, |
| "mean_token_accuracy": 0.7052734375, |
| "num_tokens": 13830207.0, |
| "rewards/accuracies": 0.846875, |
| "rewards/chosen": -1.5200662434101104, |
| "rewards/margins": 0.9510445635765791, |
| "rewards/rejected": -2.471110835671425, |
| "step": 460 |
| }, |
| { |
| "entropy": 0.3532839730381966, |
| "epoch": 2.9664, |
| "grad_norm": 7.29717493057251, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.842374894226815, |
| "logits/rejected": -24.872690353131624, |
| "logps/chosen": -50.464571952819824, |
| "logps/rejected": -71.67424020767211, |
| "loss": 0.390625, |
| "mean_token_accuracy": 0.6990234375, |
| "num_tokens": 13981057.0, |
| "rewards/accuracies": 0.903125, |
| "rewards/chosen": -1.5262503534555436, |
| "rewards/margins": 1.034103088080883, |
| "rewards/rejected": -2.560353475809097, |
| "step": 465 |
| }, |
| { |
| "entropy": 0.3396706894040108, |
| "epoch": 2.9984, |
| "grad_norm": 8.289803504943848, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.926769393836334, |
| "logits/rejected": -24.903967938070597, |
| "logps/chosen": -53.15957546234131, |
| "logps/rejected": -75.00491743087768, |
| "loss": 0.409375, |
| "mean_token_accuracy": 0.6978515625, |
| "num_tokens": 14131863.0, |
| "rewards/accuracies": 0.85625, |
| "rewards/chosen": -1.6725902140140534, |
| "rewards/margins": 1.0837561517953873, |
| "rewards/rejected": -2.756346347928047, |
| "step": 470 |
| }, |
| { |
| "entropy": 0.3428618434597464, |
| "epoch": 3.0256, |
| "grad_norm": 7.934013843536377, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.822954903884387, |
| "logits/rejected": -24.88707891713786, |
| "logps/chosen": -52.54689508325913, |
| "logps/rejected": -77.82823820675121, |
| "loss": 0.35, |
| "mean_token_accuracy": 0.6963465073529411, |
| "num_tokens": 14260035.0, |
| "rewards/accuracies": 0.8933823529411765, |
| "rewards/chosen": -1.6242689770810745, |
| "rewards/margins": 1.2483787597978817, |
| "rewards/rejected": -2.8726477412616505, |
| "step": 475 |
| }, |
| { |
| "entropy": 0.33350130543112755, |
| "epoch": 3.0576, |
| "grad_norm": 7.3017354011535645, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.75946102517421, |
| "logits/rejected": -24.778833539928517, |
| "logps/chosen": -52.98688735961914, |
| "logps/rejected": -78.90101699829101, |
| "loss": 0.35, |
| "mean_token_accuracy": 0.6919921875, |
| "num_tokens": 14410949.0, |
| "rewards/accuracies": 0.89375, |
| "rewards/chosen": -1.653100284934044, |
| "rewards/margins": 1.277121265977621, |
| "rewards/rejected": -2.9302215576171875, |
| "step": 480 |
| }, |
| { |
| "entropy": 0.3268673948943615, |
| "epoch": 3.0896, |
| "grad_norm": 7.105314254760742, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.844564110145257, |
| "logits/rejected": -24.877093765222117, |
| "logps/chosen": -54.47879152297973, |
| "logps/rejected": -81.64413318634033, |
| "loss": 0.3296875, |
| "mean_token_accuracy": 0.69140625, |
| "num_tokens": 14561801.0, |
| "rewards/accuracies": 0.91875, |
| "rewards/chosen": -1.724043107032776, |
| "rewards/margins": 1.362267342209816, |
| "rewards/rejected": -3.08631049990654, |
| "step": 485 |
| }, |
| { |
| "entropy": 0.3224384166300297, |
| "epoch": 3.1216, |
| "grad_norm": 7.154715538024902, |
| "learning_rate": 5e-06, |
| "logits/chosen": -24.941539107370637, |
| "logits/rejected": -24.91892394649721, |
| "logps/chosen": -56.65810012817383, |
| "logps/rejected": -84.36843948364258, |
| "loss": 0.35, |
| "mean_token_accuracy": 0.69736328125, |
| "num_tokens": 14712650.0, |
| "rewards/accuracies": 0.89375, |
| "rewards/chosen": -1.8231250077486039, |
| "rewards/margins": 1.3463428869843483, |
| "rewards/rejected": -3.1694678962230682, |
| "step": 490 |
| }, |
| { |
| "entropy": 0.32237044870853426, |
| "epoch": 3.1536, |
| "grad_norm": 8.422739028930664, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.014691972510292, |
| "logits/rejected": -25.008551911392267, |
| "logps/chosen": -58.14810581207276, |
| "logps/rejected": -83.99305610656738, |
| "loss": 0.35625, |
| "mean_token_accuracy": 0.6900390625, |
| "num_tokens": 14863323.0, |
| "rewards/accuracies": 0.903125, |
| "rewards/chosen": -1.906025806069374, |
| "rewards/margins": 1.2657862961292268, |
| "rewards/rejected": -3.171812105178833, |
| "step": 495 |
| }, |
| { |
| "entropy": 0.33439541682600976, |
| "epoch": 3.1856, |
| "grad_norm": 6.615303993225098, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.034348427126798, |
| "logits/rejected": -25.089082352716524, |
| "logps/chosen": -57.907769203186035, |
| "logps/rejected": -87.38656406402588, |
| "loss": 0.321875, |
| "mean_token_accuracy": 0.68671875, |
| "num_tokens": 15014022.0, |
| "rewards/accuracies": 0.89375, |
| "rewards/chosen": -1.8971118003129959, |
| "rewards/margins": 1.419259013235569, |
| "rewards/rejected": -3.316370838880539, |
| "step": 500 |
| }, |
| { |
| "entropy": 0.33486936390399935, |
| "epoch": 3.2176, |
| "grad_norm": 6.856691360473633, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.097309766039402, |
| "logits/rejected": -25.123050733765176, |
| "logps/chosen": -55.20230054855347, |
| "logps/rejected": -82.01931171417236, |
| "loss": 0.3328125, |
| "mean_token_accuracy": 0.69267578125, |
| "num_tokens": 15164814.0, |
| "rewards/accuracies": 0.9125, |
| "rewards/chosen": -1.7272155970335006, |
| "rewards/margins": 1.358732697367668, |
| "rewards/rejected": -3.0859483003616335, |
| "step": 505 |
| }, |
| { |
| "entropy": 0.32465014010667803, |
| "epoch": 3.2496, |
| "grad_norm": 7.708802700042725, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.06178075046449, |
| "logits/rejected": -25.06139493606153, |
| "logps/chosen": -56.790830516815184, |
| "logps/rejected": -83.27829141616822, |
| "loss": 0.34375, |
| "mean_token_accuracy": 0.69052734375, |
| "num_tokens": 15315747.0, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -1.8488321423530578, |
| "rewards/margins": 1.30717381387949, |
| "rewards/rejected": -3.1560059666633604, |
| "step": 510 |
| }, |
| { |
| "entropy": 0.33148474991321564, |
| "epoch": 3.2816, |
| "grad_norm": 9.32934284210205, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.02186053936112, |
| "logits/rejected": -25.043419792451697, |
| "logps/chosen": -58.26601972579956, |
| "logps/rejected": -87.74777336120606, |
| "loss": 0.3328125, |
| "mean_token_accuracy": 0.682421875, |
| "num_tokens": 15466608.0, |
| "rewards/accuracies": 0.909375, |
| "rewards/chosen": -1.9069230437278748, |
| "rewards/margins": 1.486403262615204, |
| "rewards/rejected": -3.3933263242244722, |
| "step": 515 |
| }, |
| { |
| "entropy": 0.3297711081802845, |
| "epoch": 3.3136, |
| "grad_norm": 7.6144633293151855, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.07154103969037, |
| "logits/rejected": -25.093085303334863, |
| "logps/chosen": -60.708700466156, |
| "logps/rejected": -89.8111494064331, |
| "loss": 0.3328125, |
| "mean_token_accuracy": 0.67763671875, |
| "num_tokens": 15617463.0, |
| "rewards/accuracies": 0.896875, |
| "rewards/chosen": -2.035364729166031, |
| "rewards/margins": 1.4433226779103279, |
| "rewards/rejected": -3.478687381744385, |
| "step": 520 |
| }, |
| { |
| "entropy": 0.32690444886684417, |
| "epoch": 3.3456, |
| "grad_norm": 8.8988037109375, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.081575792922685, |
| "logits/rejected": -25.06044757607366, |
| "logps/chosen": -58.68674058914185, |
| "logps/rejected": -85.75738296508788, |
| "loss": 0.35, |
| "mean_token_accuracy": 0.6822265625, |
| "num_tokens": 15768323.0, |
| "rewards/accuracies": 0.878125, |
| "rewards/chosen": -1.950838530063629, |
| "rewards/margins": 1.3486662358045578, |
| "rewards/rejected": -3.2995047569274902, |
| "step": 525 |
| }, |
| { |
| "entropy": 0.32144315764307974, |
| "epoch": 3.3776, |
| "grad_norm": 9.414474487304688, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.02346376332278, |
| "logits/rejected": -25.132050288779, |
| "logps/chosen": -58.32930955886841, |
| "logps/rejected": -86.2393642425537, |
| "loss": 0.33125, |
| "mean_token_accuracy": 0.69326171875, |
| "num_tokens": 15919237.0, |
| "rewards/accuracies": 0.896875, |
| "rewards/chosen": -1.9309983760118485, |
| "rewards/margins": 1.3938798882067203, |
| "rewards/rejected": -3.324878269433975, |
| "step": 530 |
| }, |
| { |
| "entropy": 0.3253258664160967, |
| "epoch": 3.4096, |
| "grad_norm": 7.656866073608398, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.15629693879466, |
| "logits/rejected": -25.2110073823901, |
| "logps/chosen": -61.19778699874878, |
| "logps/rejected": -90.46648483276367, |
| "loss": 0.3203125, |
| "mean_token_accuracy": 0.678125, |
| "num_tokens": 16069990.0, |
| "rewards/accuracies": 0.896875, |
| "rewards/chosen": -2.075758436322212, |
| "rewards/margins": 1.4620672211050987, |
| "rewards/rejected": -3.537825679779053, |
| "step": 535 |
| }, |
| { |
| "entropy": 0.32369369864463804, |
| "epoch": 3.4416, |
| "grad_norm": 7.227489948272705, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.136269851969193, |
| "logits/rejected": -25.166459494806855, |
| "logps/chosen": -60.808385181427, |
| "logps/rejected": -87.06750965118408, |
| "loss": 0.3546875, |
| "mean_token_accuracy": 0.67626953125, |
| "num_tokens": 16220682.0, |
| "rewards/accuracies": 0.884375, |
| "rewards/chosen": -2.0209580093622206, |
| "rewards/margins": 1.3187703728675841, |
| "rewards/rejected": -3.339728432893753, |
| "step": 540 |
| }, |
| { |
| "entropy": 0.32080022245645523, |
| "epoch": 3.4736000000000002, |
| "grad_norm": 7.4668474197387695, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.337290093071932, |
| "logits/rejected": -25.359987287706826, |
| "logps/chosen": -63.02305011749267, |
| "logps/rejected": -92.37039718627929, |
| "loss": 0.3421875, |
| "mean_token_accuracy": 0.67353515625, |
| "num_tokens": 16371280.0, |
| "rewards/accuracies": 0.8875, |
| "rewards/chosen": -2.110487860441208, |
| "rewards/margins": 1.4733156695961953, |
| "rewards/rejected": -3.5838034987449645, |
| "step": 545 |
| }, |
| { |
| "entropy": 0.32314006239175797, |
| "epoch": 3.5056000000000003, |
| "grad_norm": 8.896626472473145, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.10484775634213, |
| "logits/rejected": -25.159798916176396, |
| "logps/chosen": -60.16801671981811, |
| "logps/rejected": -89.1381483078003, |
| "loss": 0.3390625, |
| "mean_token_accuracy": 0.6794921875, |
| "num_tokens": 16522051.0, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": -1.9921035408973693, |
| "rewards/margins": 1.406488086283207, |
| "rewards/rejected": -3.398591637611389, |
| "step": 550 |
| }, |
| { |
| "entropy": 0.3151465982198715, |
| "epoch": 3.5376, |
| "grad_norm": 7.762248992919922, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.075146768990493, |
| "logits/rejected": -25.127884761279446, |
| "logps/chosen": -57.92060632705689, |
| "logps/rejected": -86.7011022567749, |
| "loss": 0.3171875, |
| "mean_token_accuracy": 0.6916015625, |
| "num_tokens": 16672850.0, |
| "rewards/accuracies": 0.93125, |
| "rewards/chosen": -1.9231851696968079, |
| "rewards/margins": 1.4329466834664344, |
| "rewards/rejected": -3.3561318635940554, |
| "step": 555 |
| }, |
| { |
| "entropy": 0.3160481035709381, |
| "epoch": 3.5696, |
| "grad_norm": 7.261136531829834, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.211173508583453, |
| "logits/rejected": -25.228998300221427, |
| "logps/chosen": -58.394444179534915, |
| "logps/rejected": -89.74018287658691, |
| "loss": 0.30625, |
| "mean_token_accuracy": 0.6841796875, |
| "num_tokens": 16823589.0, |
| "rewards/accuracies": 0.934375, |
| "rewards/chosen": -1.9537742137908936, |
| "rewards/margins": 1.522391778230667, |
| "rewards/rejected": -3.4761659860610963, |
| "step": 560 |
| }, |
| { |
| "entropy": 0.3107485048472881, |
| "epoch": 3.6016, |
| "grad_norm": 7.515143394470215, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.08429607725777, |
| "logits/rejected": -25.10101329558092, |
| "logps/chosen": -62.87925596237183, |
| "logps/rejected": -95.88719673156739, |
| "loss": 0.3140625, |
| "mean_token_accuracy": 0.6806640625, |
| "num_tokens": 16974554.0, |
| "rewards/accuracies": 0.9, |
| "rewards/chosen": -2.1517376840114593, |
| "rewards/margins": 1.6111235454678536, |
| "rewards/rejected": -3.762861263751984, |
| "step": 565 |
| }, |
| { |
| "entropy": 0.2989758800715208, |
| "epoch": 3.6336, |
| "grad_norm": 9.089282989501953, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.123257466264384, |
| "logits/rejected": -25.134205914781283, |
| "logps/chosen": -62.33095741271973, |
| "logps/rejected": -97.47971630096436, |
| "loss": 0.2875, |
| "mean_token_accuracy": 0.6837890625, |
| "num_tokens": 17125568.0, |
| "rewards/accuracies": 0.91875, |
| "rewards/chosen": -2.143625110387802, |
| "rewards/margins": 1.7342963308095931, |
| "rewards/rejected": -3.8779214560985564, |
| "step": 570 |
| }, |
| { |
| "entropy": 0.29984469451010226, |
| "epoch": 3.6656, |
| "grad_norm": 7.57747745513916, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.276228115803153, |
| "logits/rejected": -25.229910348304784, |
| "logps/chosen": -61.946348667144775, |
| "logps/rejected": -97.51498222351074, |
| "loss": 0.290625, |
| "mean_token_accuracy": 0.68447265625, |
| "num_tokens": 17276477.0, |
| "rewards/accuracies": 0.915625, |
| "rewards/chosen": -2.14150555729866, |
| "rewards/margins": 1.7264192268252372, |
| "rewards/rejected": -3.86792476773262, |
| "step": 575 |
| }, |
| { |
| "entropy": 0.30276297554373743, |
| "epoch": 3.6976, |
| "grad_norm": 6.707726955413818, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.266017213675017, |
| "logits/rejected": -25.284176210194275, |
| "logps/chosen": -60.62504739761353, |
| "logps/rejected": -92.99531497955323, |
| "loss": 0.2984375, |
| "mean_token_accuracy": 0.685546875, |
| "num_tokens": 17427266.0, |
| "rewards/accuracies": 0.915625, |
| "rewards/chosen": -2.0443687409162523, |
| "rewards/margins": 1.5940889507532119, |
| "rewards/rejected": -3.638457703590393, |
| "step": 580 |
| }, |
| { |
| "entropy": 0.2994834594428539, |
| "epoch": 3.7296, |
| "grad_norm": 7.014945983886719, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.149203888177357, |
| "logits/rejected": -25.167344175057433, |
| "logps/chosen": -62.56181554794311, |
| "logps/rejected": -93.96750469207764, |
| "loss": 0.3, |
| "mean_token_accuracy": 0.684765625, |
| "num_tokens": 17578075.0, |
| "rewards/accuracies": 0.91875, |
| "rewards/chosen": -2.152925172448158, |
| "rewards/margins": 1.5519817426800728, |
| "rewards/rejected": -3.7049069881439207, |
| "step": 585 |
| }, |
| { |
| "entropy": 0.30284112840890887, |
| "epoch": 3.7616, |
| "grad_norm": 7.132539749145508, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.235278837475853, |
| "logits/rejected": -25.308976446774704, |
| "logps/chosen": -61.15845012664795, |
| "logps/rejected": -95.02275524139404, |
| "loss": 0.26875, |
| "mean_token_accuracy": 0.68876953125, |
| "num_tokens": 17728849.0, |
| "rewards/accuracies": 0.95625, |
| "rewards/chosen": -2.060811784863472, |
| "rewards/margins": 1.7082012817263603, |
| "rewards/rejected": -3.7690130889415743, |
| "step": 590 |
| }, |
| { |
| "entropy": 0.2905466418713331, |
| "epoch": 3.7936, |
| "grad_norm": 9.337401390075684, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.275419498284357, |
| "logits/rejected": -25.260414372518646, |
| "logps/chosen": -61.83354110717774, |
| "logps/rejected": -93.50354328155518, |
| "loss": 0.309375, |
| "mean_token_accuracy": 0.68740234375, |
| "num_tokens": 17879646.0, |
| "rewards/accuracies": 0.903125, |
| "rewards/chosen": -2.1254778146743774, |
| "rewards/margins": 1.5528875276446343, |
| "rewards/rejected": -3.6783653497695923, |
| "step": 595 |
| }, |
| { |
| "entropy": 0.299504354968667, |
| "epoch": 3.8256, |
| "grad_norm": 9.183058738708496, |
| "learning_rate": 5e-06, |
| "logits/chosen": -25.322890498889848, |
| "logits/rejected": -25.36737138493695, |
| "logps/chosen": -62.754973602294925, |
| "logps/rejected": -96.71400203704835, |
| "loss": 0.3015625, |
| "mean_token_accuracy": 0.68720703125, |
| "num_tokens": 18030499.0, |
| "rewards/accuracies": 0.93125, |
| "rewards/chosen": -2.14866643846035, |
| "rewards/margins": 1.659352433681488, |
| "rewards/rejected": -3.8080188810825346, |
| "step": 600 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 3000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 20, |
| "save_steps": 200, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.7689503349597537e+18, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
