| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 93, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03225806451612903, | |
| "grad_norm": 147.81363301892154, | |
| "learning_rate": 5e-08, | |
| "logits/chosen": -0.828125, | |
| "logits/rejected": -0.64453125, | |
| "logps/chosen": -1360.0, | |
| "logps/rejected": -1192.0, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.3225806451612903, | |
| "grad_norm": 98.08774887191912, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": -0.4921875, | |
| "logits/rejected": -0.8515625, | |
| "logps/chosen": -672.0, | |
| "logps/rejected": -572.0, | |
| "loss": 0.6069, | |
| "rewards/accuracies": 0.3888888955116272, | |
| "rewards/chosen": 0.1279296875, | |
| "rewards/margins": 0.22265625, | |
| "rewards/rejected": -0.0947265625, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.6451612903225806, | |
| "grad_norm": 73.53014763743471, | |
| "learning_rate": 4.3975903614457827e-07, | |
| "logits/chosen": -0.263671875, | |
| "logits/rejected": -0.36328125, | |
| "logps/chosen": -456.0, | |
| "logps/rejected": -552.0, | |
| "loss": 0.2511, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 2.03125, | |
| "rewards/margins": 2.234375, | |
| "rewards/rejected": -0.2099609375, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.967741935483871, | |
| "grad_norm": 46.884348451479816, | |
| "learning_rate": 3.795180722891566e-07, | |
| "logits/chosen": -0.8671875, | |
| "logits/rejected": -1.2265625, | |
| "logps/chosen": -362.0, | |
| "logps/rejected": -428.0, | |
| "loss": 0.1771, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 2.734375, | |
| "rewards/margins": 3.9375, | |
| "rewards/rejected": -1.1953125, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -0.73046875, | |
| "eval_logits/rejected": -1.2109375, | |
| "eval_logps/chosen": -400.0, | |
| "eval_logps/rejected": -392.0, | |
| "eval_loss": 0.06396484375, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": 1.8046875, | |
| "eval_rewards/margins": 3.25, | |
| "eval_rewards/rejected": -1.453125, | |
| "eval_runtime": 0.399, | |
| "eval_samples_per_second": 5.012, | |
| "eval_steps_per_second": 2.506, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 1.2903225806451613, | |
| "grad_norm": 3.0375537356928777, | |
| "learning_rate": 3.192771084337349e-07, | |
| "logits/chosen": -0.59765625, | |
| "logits/rejected": -0.5859375, | |
| "logps/chosen": -454.0, | |
| "logps/rejected": -454.0, | |
| "loss": 0.0471, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 4.34375, | |
| "rewards/margins": 6.125, | |
| "rewards/rejected": -1.7890625, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.6129032258064515, | |
| "grad_norm": 10.062564968864535, | |
| "learning_rate": 2.5903614457831324e-07, | |
| "logits/chosen": -0.349609375, | |
| "logits/rejected": -0.51953125, | |
| "logps/chosen": -552.0, | |
| "logps/rejected": -640.0, | |
| "loss": 0.0393, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 4.25, | |
| "rewards/margins": 7.40625, | |
| "rewards/rejected": -3.15625, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.935483870967742, | |
| "grad_norm": 9.004337079403212, | |
| "learning_rate": 1.9879518072289155e-07, | |
| "logits/chosen": -0.58203125, | |
| "logits/rejected": -0.8125, | |
| "logps/chosen": -352.0, | |
| "logps/rejected": -436.0, | |
| "loss": 0.03, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 3.4375, | |
| "rewards/margins": 6.6875, | |
| "rewards/rejected": -3.265625, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -0.8203125, | |
| "eval_logits/rejected": -1.3671875, | |
| "eval_logps/chosen": -400.0, | |
| "eval_logps/rejected": -398.0, | |
| "eval_loss": 0.060791015625, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": 1.6015625, | |
| "eval_rewards/margins": 3.65625, | |
| "eval_rewards/rejected": -2.046875, | |
| "eval_runtime": 0.4056, | |
| "eval_samples_per_second": 4.932, | |
| "eval_steps_per_second": 2.466, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 2.258064516129032, | |
| "grad_norm": 1.1747278729225081, | |
| "learning_rate": 1.3855421686746987e-07, | |
| "logits/chosen": -0.58203125, | |
| "logits/rejected": -0.8125, | |
| "logps/chosen": -388.0, | |
| "logps/rejected": -484.0, | |
| "loss": 0.0093, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 5.25, | |
| "rewards/margins": 9.5, | |
| "rewards/rejected": -4.25, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.5806451612903225, | |
| "grad_norm": 1.0498376354880619, | |
| "learning_rate": 7.83132530120482e-08, | |
| "logits/chosen": -0.6171875, | |
| "logits/rejected": -0.78515625, | |
| "logps/chosen": -412.0, | |
| "logps/rejected": -496.0, | |
| "loss": 0.0057, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 3.515625, | |
| "rewards/margins": 7.59375, | |
| "rewards/rejected": -4.09375, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.903225806451613, | |
| "grad_norm": 1.068907016463518, | |
| "learning_rate": 1.8072289156626504e-08, | |
| "logits/chosen": -0.90234375, | |
| "logits/rejected": -0.875, | |
| "logps/chosen": -478.0, | |
| "logps/rejected": -616.0, | |
| "loss": 0.0037, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 5.28125, | |
| "rewards/margins": 9.1875, | |
| "rewards/rejected": -3.875, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_logits/chosen": -0.8828125, | |
| "eval_logits/rejected": -1.453125, | |
| "eval_logps/chosen": -406.0, | |
| "eval_logps/rejected": -408.0, | |
| "eval_loss": 0.040771484375, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": 1.09375, | |
| "eval_rewards/margins": 4.25, | |
| "eval_rewards/rejected": -3.15625, | |
| "eval_runtime": 0.404, | |
| "eval_samples_per_second": 4.95, | |
| "eval_steps_per_second": 2.475, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 93, | |
| "total_flos": 0.0, | |
| "train_loss": 0.1267821729503652, | |
| "train_runtime": 288.8037, | |
| "train_samples_per_second": 5.08, | |
| "train_steps_per_second": 0.322 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 93, | |
| "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": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |