| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9921671018276762, | |
| "eval_steps": 500, | |
| "global_step": 95, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 6.531418223090284, | |
| "learning_rate": 1e-07, | |
| "logits/chosen": -2.851747512817383, | |
| "logits/rejected": -2.833996534347534, | |
| "logps/chosen": -165.70089721679688, | |
| "logps/rejected": -198.857666015625, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 5.924500505195678, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": -2.7712948322296143, | |
| "logits/rejected": -2.7740774154663086, | |
| "logps/chosen": -171.32467651367188, | |
| "logps/rejected": -172.57489013671875, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.0003833131631836295, | |
| "rewards/margins": -8.183407771866769e-06, | |
| "rewards/rejected": 0.00039149660733528435, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 7.115083584877395, | |
| "learning_rate": 1e-06, | |
| "logits/chosen": -2.787468433380127, | |
| "logits/rejected": -2.796309471130371, | |
| "logps/chosen": -188.18690490722656, | |
| "logps/rejected": -193.19281005859375, | |
| "loss": 0.6903, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.021850628778338432, | |
| "rewards/margins": 0.00575407687574625, | |
| "rewards/rejected": 0.016096554696559906, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 7.576406474777614, | |
| "learning_rate": 9.91486549841951e-07, | |
| "logits/chosen": -2.84570574760437, | |
| "logits/rejected": -2.855926513671875, | |
| "logps/chosen": -193.6645050048828, | |
| "logps/rejected": -185.57162475585938, | |
| "loss": 0.681, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.06298746168613434, | |
| "rewards/margins": 0.029651161283254623, | |
| "rewards/rejected": 0.03333630412817001, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 9.329171862378226, | |
| "learning_rate": 9.66236114702178e-07, | |
| "logits/chosen": -2.831686496734619, | |
| "logits/rejected": -2.8417012691497803, | |
| "logps/chosen": -170.6667938232422, | |
| "logps/rejected": -189.2131805419922, | |
| "loss": 0.6564, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.019239652901887894, | |
| "rewards/margins": 0.0761820375919342, | |
| "rewards/rejected": -0.0954216942191124, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 11.230824794807015, | |
| "learning_rate": 9.251085678648071e-07, | |
| "logits/chosen": -2.7868192195892334, | |
| "logits/rejected": -2.7850959300994873, | |
| "logps/chosen": -136.22763061523438, | |
| "logps/rejected": -169.01756286621094, | |
| "loss": 0.6363, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.04755619913339615, | |
| "rewards/margins": 0.1481182873249054, | |
| "rewards/rejected": -0.19567449390888214, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 10.213782257588267, | |
| "learning_rate": 8.695044586103295e-07, | |
| "logits/chosen": -2.74267840385437, | |
| "logits/rejected": -2.7413182258605957, | |
| "logps/chosen": -163.45603942871094, | |
| "logps/rejected": -190.14236450195312, | |
| "loss": 0.6341, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.039765361696481705, | |
| "rewards/margins": 0.20882920920848846, | |
| "rewards/rejected": -0.24859456717967987, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 9.724737450156574, | |
| "learning_rate": 8.013173181896282e-07, | |
| "logits/chosen": -2.7809696197509766, | |
| "logits/rejected": -2.774864435195923, | |
| "logps/chosen": -203.69174194335938, | |
| "logps/rejected": -228.673828125, | |
| "loss": 0.623, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.12420445680618286, | |
| "rewards/margins": 0.1601138412952423, | |
| "rewards/rejected": -0.28431832790374756, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 12.552857274756276, | |
| "learning_rate": 7.228691778882692e-07, | |
| "logits/chosen": -2.6985127925872803, | |
| "logits/rejected": -2.7104218006134033, | |
| "logps/chosen": -158.4896240234375, | |
| "logps/rejected": -199.87753295898438, | |
| "loss": 0.6049, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.25630590319633484, | |
| "rewards/margins": 0.26447391510009766, | |
| "rewards/rejected": -0.5207797884941101, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 12.147016576418242, | |
| "learning_rate": 6.368314950360415e-07, | |
| "logits/chosen": -2.6584348678588867, | |
| "logits/rejected": -2.6736438274383545, | |
| "logps/chosen": -232.7853240966797, | |
| "logps/rejected": -262.41015625, | |
| "loss": 0.6157, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.4413759112358093, | |
| "rewards/margins": 0.3659912645816803, | |
| "rewards/rejected": -0.807367205619812, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 13.949220920878545, | |
| "learning_rate": 5.46134179731651e-07, | |
| "logits/chosen": -2.61830735206604, | |
| "logits/rejected": -2.6202476024627686, | |
| "logps/chosen": -231.3226776123047, | |
| "logps/rejected": -291.72406005859375, | |
| "loss": 0.588, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.5699625015258789, | |
| "rewards/margins": 0.44822096824645996, | |
| "rewards/rejected": -1.0181834697723389, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 15.029123695933512, | |
| "learning_rate": 4.5386582026834904e-07, | |
| "logits/chosen": -2.4635961055755615, | |
| "logits/rejected": -2.4716153144836426, | |
| "logps/chosen": -202.64266967773438, | |
| "logps/rejected": -248.4220733642578, | |
| "loss": 0.5839, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.5452982783317566, | |
| "rewards/margins": 0.4111364483833313, | |
| "rewards/rejected": -0.9564347267150879, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 17.205678814880788, | |
| "learning_rate": 3.6316850496395855e-07, | |
| "logits/chosen": -2.5053164958953857, | |
| "logits/rejected": -2.5017011165618896, | |
| "logps/chosen": -228.46469116210938, | |
| "logps/rejected": -310.7025451660156, | |
| "loss": 0.5648, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.5650664567947388, | |
| "rewards/margins": 0.5835850238800049, | |
| "rewards/rejected": -1.148651361465454, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 19.89901816550521, | |
| "learning_rate": 2.771308221117309e-07, | |
| "logits/chosen": -2.4981369972229004, | |
| "logits/rejected": -2.5108680725097656, | |
| "logps/chosen": -236.54653930664062, | |
| "logps/rejected": -310.23162841796875, | |
| "loss": 0.563, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.7446938753128052, | |
| "rewards/margins": 0.6594551801681519, | |
| "rewards/rejected": -1.4041489362716675, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 23.0785172466263, | |
| "learning_rate": 1.9868268181037184e-07, | |
| "logits/chosen": -2.4181766510009766, | |
| "logits/rejected": -2.4266486167907715, | |
| "logps/chosen": -264.23138427734375, | |
| "logps/rejected": -309.7364807128906, | |
| "loss": 0.5533, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.8709458112716675, | |
| "rewards/margins": 0.48874396085739136, | |
| "rewards/rejected": -1.3596898317337036, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 20.36959281359639, | |
| "learning_rate": 1.304955413896705e-07, | |
| "logits/chosen": -2.3758950233459473, | |
| "logits/rejected": -2.3715245723724365, | |
| "logps/chosen": -251.191650390625, | |
| "logps/rejected": -353.01251220703125, | |
| "loss": 0.5328, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.7933205366134644, | |
| "rewards/margins": 0.9006916880607605, | |
| "rewards/rejected": -1.6940120458602905, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 32.00854391073857, | |
| "learning_rate": 7.4891432135193e-08, | |
| "logits/chosen": -2.346649646759033, | |
| "logits/rejected": -2.3456640243530273, | |
| "logps/chosen": -242.99374389648438, | |
| "logps/rejected": -310.0765075683594, | |
| "loss": 0.5629, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7500452995300293, | |
| "rewards/margins": 0.5974142551422119, | |
| "rewards/rejected": -1.3474594354629517, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 24.468901281465246, | |
| "learning_rate": 3.376388529782215e-08, | |
| "logits/chosen": -2.396498918533325, | |
| "logits/rejected": -2.398857593536377, | |
| "logps/chosen": -232.2967987060547, | |
| "logps/rejected": -311.627197265625, | |
| "loss": 0.5814, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.7211315035820007, | |
| "rewards/margins": 0.70616614818573, | |
| "rewards/rejected": -1.427297830581665, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 21.76700663596853, | |
| "learning_rate": 8.513450158049106e-09, | |
| "logits/chosen": -2.337101459503174, | |
| "logits/rejected": -2.347583770751953, | |
| "logps/chosen": -246.55062866210938, | |
| "logps/rejected": -331.9063415527344, | |
| "loss": 0.5556, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.7761114835739136, | |
| "rewards/margins": 0.7871293425559998, | |
| "rewards/rejected": -1.5632407665252686, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 19.722062441035273, | |
| "learning_rate": 0.0, | |
| "logits/chosen": -2.403937339782715, | |
| "logits/rejected": -2.4172253608703613, | |
| "logps/chosen": -263.65972900390625, | |
| "logps/rejected": -332.18292236328125, | |
| "loss": 0.5584, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7805107831954956, | |
| "rewards/margins": 0.713537335395813, | |
| "rewards/rejected": -1.4940482378005981, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "step": 95, | |
| "total_flos": 0.0, | |
| "train_loss": 0.6041482373287803, | |
| "train_runtime": 2420.5057, | |
| "train_samples_per_second": 5.051, | |
| "train_steps_per_second": 0.039 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 95, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |