| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 52.806314549108166, | |
| "learning_rate": 3.3333333333333334e-08, | |
| "logits/chosen": -3.515625, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -154.0, | |
| "logps/rejected": -102.0, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 56.240893408179446, | |
| "learning_rate": 3.333333333333333e-07, | |
| "logits/chosen": -3.546875, | |
| "logits/rejected": -3.53125, | |
| "logps/chosen": -187.0, | |
| "logps/rejected": -165.0, | |
| "loss": 0.6884, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.0167236328125, | |
| "rewards/margins": 0.005889892578125, | |
| "rewards/rejected": 0.01080322265625, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 44.13028314484008, | |
| "learning_rate": 4.814814814814814e-07, | |
| "logits/chosen": -3.640625, | |
| "logits/rejected": -3.578125, | |
| "logps/chosen": -166.0, | |
| "logps/rejected": -150.0, | |
| "loss": 0.6584, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.087890625, | |
| "rewards/margins": 0.1142578125, | |
| "rewards/rejected": -0.0262451171875, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 48.03945494362009, | |
| "learning_rate": 4.444444444444444e-07, | |
| "logits/chosen": -3.59375, | |
| "logits/rejected": -3.671875, | |
| "logps/chosen": -176.0, | |
| "logps/rejected": -143.0, | |
| "loss": 0.6756, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.00087738037109375, | |
| "rewards/margins": 0.099609375, | |
| "rewards/rejected": -0.10009765625, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 56.29734383934514, | |
| "learning_rate": 4.0740740740740737e-07, | |
| "logits/chosen": -3.671875, | |
| "logits/rejected": -3.65625, | |
| "logps/chosen": -161.0, | |
| "logps/rejected": -152.0, | |
| "loss": 0.6825, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.017333984375, | |
| "rewards/margins": 0.07568359375, | |
| "rewards/rejected": -0.09326171875, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 50.25615937451418, | |
| "learning_rate": 3.703703703703703e-07, | |
| "logits/chosen": -3.640625, | |
| "logits/rejected": -3.640625, | |
| "logps/chosen": -159.0, | |
| "logps/rejected": -147.0, | |
| "loss": 0.6643, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.0257568359375, | |
| "rewards/margins": 0.08740234375, | |
| "rewards/rejected": -0.0615234375, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -3.703125, | |
| "eval_logits/rejected": -3.734375, | |
| "eval_logps/chosen": -157.0, | |
| "eval_logps/rejected": -141.0, | |
| "eval_loss": 0.6264062523841858, | |
| "eval_rewards/accuracies": 0.625, | |
| "eval_rewards/chosen": 0.1142578125, | |
| "eval_rewards/margins": 0.2001953125, | |
| "eval_rewards/rejected": -0.08544921875, | |
| "eval_runtime": 12.4892, | |
| "eval_samples_per_second": 16.014, | |
| "eval_steps_per_second": 0.56, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 12.074720568581668, | |
| "learning_rate": 3.333333333333333e-07, | |
| "logits/chosen": -3.625, | |
| "logits/rejected": -3.59375, | |
| "logps/chosen": -139.0, | |
| "logps/rejected": -160.0, | |
| "loss": 0.1625, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 1.1953125, | |
| "rewards/margins": 2.859375, | |
| "rewards/rejected": -1.671875, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 13.74152867856679, | |
| "learning_rate": 2.962962962962963e-07, | |
| "logits/chosen": -3.671875, | |
| "logits/rejected": -3.640625, | |
| "logps/chosen": -153.0, | |
| "logps/rejected": -161.0, | |
| "loss": 0.169, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 1.53125, | |
| "rewards/margins": 3.25, | |
| "rewards/rejected": -1.71875, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 34.103416975119806, | |
| "learning_rate": 2.5925925925925923e-07, | |
| "logits/chosen": -3.625, | |
| "logits/rejected": -3.671875, | |
| "logps/chosen": -154.0, | |
| "logps/rejected": -189.0, | |
| "loss": 0.1765, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 1.3125, | |
| "rewards/margins": 3.203125, | |
| "rewards/rejected": -1.890625, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 26.20584063624067, | |
| "learning_rate": 2.222222222222222e-07, | |
| "logits/chosen": -3.640625, | |
| "logits/rejected": -3.65625, | |
| "logps/chosen": -160.0, | |
| "logps/rejected": -173.0, | |
| "loss": 0.17, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 1.2890625, | |
| "rewards/margins": 3.390625, | |
| "rewards/rejected": -2.109375, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 18.8256623968867, | |
| "learning_rate": 1.8518518518518516e-07, | |
| "logits/chosen": -3.734375, | |
| "logits/rejected": -3.703125, | |
| "logps/chosen": -154.0, | |
| "logps/rejected": -157.0, | |
| "loss": 0.1878, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 1.21875, | |
| "rewards/margins": 3.140625, | |
| "rewards/rejected": -1.921875, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -3.875, | |
| "eval_logits/rejected": -3.90625, | |
| "eval_logps/chosen": -163.0, | |
| "eval_logps/rejected": -148.0, | |
| "eval_loss": 0.619921863079071, | |
| "eval_rewards/accuracies": 0.6964285969734192, | |
| "eval_rewards/chosen": -0.478515625, | |
| "eval_rewards/margins": 0.283203125, | |
| "eval_rewards/rejected": -0.76171875, | |
| "eval_runtime": 15.6192, | |
| "eval_samples_per_second": 12.805, | |
| "eval_steps_per_second": 0.448, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 24.45339417020041, | |
| "learning_rate": 1.4814814814814815e-07, | |
| "logits/chosen": -3.984375, | |
| "logits/rejected": -3.921875, | |
| "logps/chosen": -151.0, | |
| "logps/rejected": -176.0, | |
| "loss": 0.0629, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.85546875, | |
| "rewards/margins": 4.875, | |
| "rewards/rejected": -4.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 4.297431326133669, | |
| "learning_rate": 1.111111111111111e-07, | |
| "logits/chosen": -3.8125, | |
| "logits/rejected": -3.78125, | |
| "logps/chosen": -169.0, | |
| "logps/rejected": -206.0, | |
| "loss": 0.0346, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.47265625, | |
| "rewards/margins": 5.8125, | |
| "rewards/rejected": -5.34375, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 10.550256730002323, | |
| "learning_rate": 7.407407407407407e-08, | |
| "logits/chosen": -3.796875, | |
| "logits/rejected": -3.765625, | |
| "logps/chosen": -173.0, | |
| "logps/rejected": -198.0, | |
| "loss": 0.0839, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.35546875, | |
| "rewards/margins": 5.59375, | |
| "rewards/rejected": -5.25, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 7.609442555366527, | |
| "learning_rate": 3.7037037037037036e-08, | |
| "logits/chosen": -3.984375, | |
| "logits/rejected": -3.96875, | |
| "logps/chosen": -158.0, | |
| "logps/rejected": -205.0, | |
| "loss": 0.0544, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 0.5859375, | |
| "rewards/margins": 5.65625, | |
| "rewards/rejected": -5.0625, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 6.620137166944363, | |
| "learning_rate": 0.0, | |
| "logits/chosen": -3.96875, | |
| "logits/rejected": -4.0, | |
| "logps/chosen": -154.0, | |
| "logps/rejected": -178.0, | |
| "loss": 0.0766, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.376953125, | |
| "rewards/margins": 4.53125, | |
| "rewards/rejected": -4.15625, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_logits/chosen": -4.03125, | |
| "eval_logits/rejected": -4.0625, | |
| "eval_logps/chosen": -179.0, | |
| "eval_logps/rejected": -162.0, | |
| "eval_loss": 0.7985156178474426, | |
| "eval_rewards/accuracies": 0.5535714030265808, | |
| "eval_rewards/chosen": -2.046875, | |
| "eval_rewards/margins": 0.185546875, | |
| "eval_rewards/rejected": -2.21875, | |
| "eval_runtime": 14.95, | |
| "eval_samples_per_second": 13.378, | |
| "eval_steps_per_second": 0.468, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 150, | |
| "total_flos": 0.0, | |
| "train_loss": 0.30317095438639324, | |
| "train_runtime": 1362.0859, | |
| "train_samples_per_second": 3.522, | |
| "train_steps_per_second": 0.11 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 150, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "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 | |
| } | |