{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9874476987447699, "eval_steps": 500, "global_step": 59, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.016736401673640166, "grad_norm": 6.686057910691401, "learning_rate": 1.6666666666666664e-08, "logits/chosen": -2.7900264263153076, "logits/rejected": -2.8420324325561523, "logps/chosen": -162.99993896484375, "logps/pi_response": -170.2482147216797, "logps/ref_response": -170.2482147216797, "logps/rejected": -172.05972290039062, "loss": 0.6947, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.16736401673640167, "grad_norm": 6.423095508046628, "learning_rate": 9.860114570402053e-08, "logits/chosen": -2.730027198791504, "logits/rejected": -2.7307584285736084, "logps/chosen": -147.63504028320312, "logps/pi_response": -153.25831604003906, "logps/ref_response": -153.30535888671875, "logps/rejected": -149.9705047607422, "loss": 0.6946, "rewards/accuracies": 0.4375, "rewards/chosen": 0.00022389186779037118, "rewards/margins": -0.00020425634284038097, "rewards/rejected": 0.0004281482251826674, "step": 10 }, { "epoch": 0.33472803347280333, "grad_norm": 6.824808764101014, "learning_rate": 8.374915007591053e-08, "logits/chosen": -2.700606107711792, "logits/rejected": -2.6993486881256104, "logps/chosen": -155.84555053710938, "logps/pi_response": -155.1090850830078, "logps/ref_response": -155.0887451171875, "logps/rejected": -152.13174438476562, "loss": 0.6946, "rewards/accuracies": 0.5, "rewards/chosen": -6.415335519704968e-05, "rewards/margins": -2.050258990493603e-05, "rewards/rejected": -4.3650768930092454e-05, "step": 20 }, { "epoch": 0.502092050209205, "grad_norm": 6.476727877615816, "learning_rate": 5.738232820012406e-08, "logits/chosen": -2.7094643115997314, "logits/rejected": -2.6976773738861084, "logps/chosen": -149.053466796875, "logps/pi_response": -150.77244567871094, "logps/ref_response": -150.64227294921875, "logps/rejected": -151.15628051757812, "loss": 0.6946, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -0.0011378342751413584, "rewards/margins": 0.0005736011080443859, "rewards/rejected": -0.0017114352667704225, "step": 30 }, { "epoch": 0.6694560669456067, "grad_norm": 6.389121900535284, "learning_rate": 2.8496739886173992e-08, "logits/chosen": -2.7291507720947266, "logits/rejected": -2.7260355949401855, "logps/chosen": -160.88548278808594, "logps/pi_response": -155.79739379882812, "logps/ref_response": -155.7581329345703, "logps/rejected": -158.36874389648438, "loss": 0.6946, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -0.0001912559091579169, "rewards/margins": -8.280497422674671e-05, "rewards/rejected": -0.00010845092765521258, "step": 40 }, { "epoch": 0.8368200836820083, "grad_norm": 6.430706683984891, "learning_rate": 6.947819411632222e-09, "logits/chosen": -2.7263987064361572, "logits/rejected": -2.7108066082000732, "logps/chosen": -151.0059051513672, "logps/pi_response": -149.9954071044922, "logps/ref_response": -150.04244995117188, "logps/rejected": -152.29220581054688, "loss": 0.6944, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": 0.001144839683547616, "rewards/margins": 0.000988497631624341, "rewards/rejected": 0.0001563422556500882, "step": 50 }, { "epoch": 0.9874476987447699, "step": 59, "total_flos": 0.0, "train_loss": 0.6946261302899506, "train_runtime": 1308.9042, "train_samples_per_second": 11.676, "train_steps_per_second": 0.045 } ], "logging_steps": 10, "max_steps": 59, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "total_flos": 0.0, "train_batch_size": 4, "trial_name": null, "trial_params": null }