| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 312, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.009615384615384616, | |
| "grad_norm": 28.194065437278258, | |
| "learning_rate": 1.5625e-08, | |
| "logits/chosen": -1.609375, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -152.0, | |
| "logps/rejected": -127.5, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.09615384615384616, | |
| "grad_norm": 26.916436077863562, | |
| "learning_rate": 1.5624999999999999e-07, | |
| "logits/chosen": -1.6015625, | |
| "logits/rejected": -1.484375, | |
| "logps/chosen": -149.0, | |
| "logps/rejected": -128.0, | |
| "loss": 0.6868, | |
| "rewards/accuracies": 0.3194444477558136, | |
| "rewards/chosen": -0.0045166015625, | |
| "rewards/margins": 0.01080322265625, | |
| "rewards/rejected": -0.0152587890625, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.19230769230769232, | |
| "grad_norm": 20.740344035629946, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.4765625, | |
| "logps/chosen": -160.0, | |
| "logps/rejected": -133.0, | |
| "loss": 0.5799, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.03662109375, | |
| "rewards/margins": 0.2734375, | |
| "rewards/rejected": -0.2373046875, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.28846153846153844, | |
| "grad_norm": 6.570831756440352, | |
| "learning_rate": 4.6874999999999996e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.484375, | |
| "logps/chosen": -142.0, | |
| "logps/rejected": -142.0, | |
| "loss": 0.254, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.0072021484375, | |
| "rewards/margins": 1.390625, | |
| "rewards/rejected": -1.3828125, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 0.3430324547403658, | |
| "learning_rate": 4.857142857142857e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.546875, | |
| "logps/chosen": -169.0, | |
| "logps/rejected": -180.0, | |
| "loss": 0.0296, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.341796875, | |
| "rewards/margins": 4.8125, | |
| "rewards/rejected": -5.15625, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4807692307692308, | |
| "grad_norm": 1.0998802903857687, | |
| "learning_rate": 4.6785714285714283e-07, | |
| "logits/chosen": -1.6484375, | |
| "logits/rejected": -1.5234375, | |
| "logps/chosen": -165.0, | |
| "logps/rejected": -200.0, | |
| "loss": 0.0098, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.7265625, | |
| "rewards/margins": 6.625, | |
| "rewards/rejected": -7.34375, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5769230769230769, | |
| "grad_norm": 2.398247571338227, | |
| "learning_rate": 4.5e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.5703125, | |
| "logps/chosen": -173.0, | |
| "logps/rejected": -213.0, | |
| "loss": 0.0084, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.9609375, | |
| "rewards/margins": 7.4375, | |
| "rewards/rejected": -8.375, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6730769230769231, | |
| "grad_norm": 3.272238721827958, | |
| "learning_rate": 4.3214285714285713e-07, | |
| "logits/chosen": -1.640625, | |
| "logits/rejected": -1.5625, | |
| "logps/chosen": -169.0, | |
| "logps/rejected": -219.0, | |
| "loss": 0.0037, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.796875, | |
| "rewards/margins": 8.125, | |
| "rewards/rejected": -8.875, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 0.03234130751760337, | |
| "learning_rate": 4.142857142857143e-07, | |
| "logits/chosen": -1.65625, | |
| "logits/rejected": -1.609375, | |
| "logps/chosen": -151.0, | |
| "logps/rejected": -229.0, | |
| "loss": 0.0027, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.90625, | |
| "rewards/margins": 9.0, | |
| "rewards/rejected": -9.875, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.8653846153846154, | |
| "grad_norm": 2.22694231928167, | |
| "learning_rate": 3.9642857142857137e-07, | |
| "logits/chosen": -1.640625, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -154.0, | |
| "logps/rejected": -229.0, | |
| "loss": 0.0023, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.0390625, | |
| "rewards/margins": 8.75, | |
| "rewards/rejected": -9.8125, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.9615384615384616, | |
| "grad_norm": 0.04456450837390773, | |
| "learning_rate": 3.785714285714285e-07, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": -1.6328125, | |
| "logps/chosen": -174.0, | |
| "logps/rejected": -232.0, | |
| "loss": 0.0029, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.87109375, | |
| "rewards/margins": 9.0625, | |
| "rewards/rejected": -9.9375, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -1.4765625, | |
| "eval_logits/rejected": -1.4140625, | |
| "eval_logps/chosen": -119.0, | |
| "eval_logps/rejected": -235.0, | |
| "eval_loss": 0.02374856546521187, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.83203125, | |
| "eval_rewards/margins": 9.8125, | |
| "eval_rewards/rejected": -10.625, | |
| "eval_runtime": 5.2091, | |
| "eval_samples_per_second": 13.054, | |
| "eval_steps_per_second": 0.576, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.0576923076923077, | |
| "grad_norm": 0.020791139211903984, | |
| "learning_rate": 3.607142857142857e-07, | |
| "logits/chosen": -1.6875, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -148.0, | |
| "logps/rejected": -226.0, | |
| "loss": 0.0017, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.83203125, | |
| "rewards/margins": 9.25, | |
| "rewards/rejected": -10.0625, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.1538461538461537, | |
| "grad_norm": 0.04771399447611739, | |
| "learning_rate": 3.4285714285714286e-07, | |
| "logits/chosen": -1.6484375, | |
| "logits/rejected": -1.5859375, | |
| "logps/chosen": -167.0, | |
| "logps/rejected": -234.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.3125, | |
| "rewards/margins": 9.5625, | |
| "rewards/rejected": -10.875, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 0.2534459160228267, | |
| "learning_rate": 3.25e-07, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -165.0, | |
| "logps/rejected": -234.0, | |
| "loss": 0.001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.3359375, | |
| "rewards/margins": 9.3125, | |
| "rewards/rejected": -10.6875, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.3461538461538463, | |
| "grad_norm": 0.07091308610265967, | |
| "learning_rate": 3.0714285714285716e-07, | |
| "logits/chosen": -1.6640625, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -179.0, | |
| "logps/rejected": -245.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.25, | |
| "rewards/margins": 10.125, | |
| "rewards/rejected": -11.375, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.4423076923076923, | |
| "grad_norm": 0.01427238285225871, | |
| "learning_rate": 2.892857142857143e-07, | |
| "logits/chosen": -1.734375, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -176.0, | |
| "logps/rejected": -240.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.25, | |
| "rewards/margins": 9.75, | |
| "rewards/rejected": -11.0625, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 0.06865294866224332, | |
| "learning_rate": 2.714285714285714e-07, | |
| "logits/chosen": -1.7578125, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -183.0, | |
| "logps/rejected": -241.0, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.140625, | |
| "rewards/margins": 10.25, | |
| "rewards/rejected": -11.375, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.6346153846153846, | |
| "grad_norm": 0.07896118186411183, | |
| "learning_rate": 2.5357142857142855e-07, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.5234375, | |
| "logps/chosen": -160.0, | |
| "logps/rejected": -246.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.2109375, | |
| "rewards/margins": 10.5625, | |
| "rewards/rejected": -11.75, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.7307692307692308, | |
| "grad_norm": 0.026222949044449873, | |
| "learning_rate": 2.357142857142857e-07, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": -1.6484375, | |
| "logps/chosen": -178.0, | |
| "logps/rejected": -245.0, | |
| "loss": 0.0015, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.3984375, | |
| "rewards/margins": 10.25, | |
| "rewards/rejected": -11.625, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.8269230769230769, | |
| "grad_norm": 0.011793924192378881, | |
| "learning_rate": 2.1785714285714284e-07, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": -1.71875, | |
| "logps/chosen": -159.0, | |
| "logps/rejected": -245.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.25, | |
| "rewards/margins": 10.4375, | |
| "rewards/rejected": -11.6875, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.9230769230769231, | |
| "grad_norm": 0.060978420362998054, | |
| "learning_rate": 2e-07, | |
| "logits/chosen": -1.703125, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -158.0, | |
| "logps/rejected": -243.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.0859375, | |
| "rewards/margins": 10.5, | |
| "rewards/rejected": -11.5625, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -1.5, | |
| "eval_logits/rejected": -1.4453125, | |
| "eval_logps/chosen": -121.0, | |
| "eval_logps/rejected": -248.0, | |
| "eval_loss": 0.006824689917266369, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -1.0078125, | |
| "eval_rewards/margins": 10.875, | |
| "eval_rewards/rejected": -11.875, | |
| "eval_runtime": 6.628, | |
| "eval_samples_per_second": 10.26, | |
| "eval_steps_per_second": 0.453, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 2.019230769230769, | |
| "grad_norm": 0.025694364273425255, | |
| "learning_rate": 1.8214285714285714e-07, | |
| "logits/chosen": -1.640625, | |
| "logits/rejected": -1.640625, | |
| "logps/chosen": -162.0, | |
| "logps/rejected": -244.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.015625, | |
| "rewards/margins": 10.625, | |
| "rewards/rejected": -11.625, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.1153846153846154, | |
| "grad_norm": 1.318722077689268, | |
| "learning_rate": 1.6428571428571429e-07, | |
| "logits/chosen": -1.71875, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -165.0, | |
| "logps/rejected": -250.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.03125, | |
| "rewards/margins": 10.8125, | |
| "rewards/rejected": -11.8125, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.2115384615384617, | |
| "grad_norm": 0.15842660516208706, | |
| "learning_rate": 1.4642857142857143e-07, | |
| "logits/chosen": -1.734375, | |
| "logits/rejected": -1.7109375, | |
| "logps/chosen": -179.0, | |
| "logps/rejected": -255.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.5234375, | |
| "rewards/margins": 10.75, | |
| "rewards/rejected": -12.25, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.3076923076923075, | |
| "grad_norm": 0.1644318912700695, | |
| "learning_rate": 1.2857142857142855e-07, | |
| "logits/chosen": -1.6953125, | |
| "logits/rejected": -1.734375, | |
| "logps/chosen": -161.0, | |
| "logps/rejected": -251.0, | |
| "loss": 0.0008, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.1484375, | |
| "rewards/margins": 10.5625, | |
| "rewards/rejected": -11.6875, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.4038461538461537, | |
| "grad_norm": 0.1117902717935036, | |
| "learning_rate": 1.107142857142857e-07, | |
| "logits/chosen": -1.703125, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -174.0, | |
| "logps/rejected": -249.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.25, | |
| "rewards/margins": 10.9375, | |
| "rewards/rejected": -12.1875, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.04593607447714544, | |
| "learning_rate": 9.285714285714286e-08, | |
| "logits/chosen": -1.6484375, | |
| "logits/rejected": -1.6328125, | |
| "logps/chosen": -166.0, | |
| "logps/rejected": -251.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.1328125, | |
| "rewards/margins": 11.1875, | |
| "rewards/rejected": -12.3125, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.5961538461538463, | |
| "grad_norm": 0.08723297585467851, | |
| "learning_rate": 7.5e-08, | |
| "logits/chosen": -1.7109375, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -175.0, | |
| "logps/rejected": -250.0, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.328125, | |
| "rewards/margins": 10.4375, | |
| "rewards/rejected": -11.8125, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.6923076923076925, | |
| "grad_norm": 0.16735539683605713, | |
| "learning_rate": 5.714285714285714e-08, | |
| "logits/chosen": -1.71875, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -181.0, | |
| "logps/rejected": -252.0, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.2734375, | |
| "rewards/margins": 10.9375, | |
| "rewards/rejected": -12.25, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.7884615384615383, | |
| "grad_norm": 0.10696117602485962, | |
| "learning_rate": 3.9285714285714285e-08, | |
| "logits/chosen": -1.625, | |
| "logits/rejected": -1.6640625, | |
| "logps/chosen": -150.0, | |
| "logps/rejected": -248.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.109375, | |
| "rewards/margins": 11.0, | |
| "rewards/rejected": -12.0625, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.8846153846153846, | |
| "grad_norm": 0.00967600454300797, | |
| "learning_rate": 2.142857142857143e-08, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -171.0, | |
| "logps/rejected": -258.0, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.171875, | |
| "rewards/margins": 11.25, | |
| "rewards/rejected": -12.4375, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.980769230769231, | |
| "grad_norm": 0.12453557861807049, | |
| "learning_rate": 3.571428571428571e-09, | |
| "logits/chosen": -1.6640625, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -164.0, | |
| "logps/rejected": -245.0, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -1.1796875, | |
| "rewards/margins": 10.8125, | |
| "rewards/rejected": -12.0, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_logits/chosen": -1.5, | |
| "eval_logits/rejected": -1.453125, | |
| "eval_logps/chosen": -121.5, | |
| "eval_logps/rejected": -253.0, | |
| "eval_loss": 0.004145369865000248, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -1.0859375, | |
| "eval_rewards/margins": 11.25, | |
| "eval_rewards/rejected": -12.3125, | |
| "eval_runtime": 6.3498, | |
| "eval_samples_per_second": 10.709, | |
| "eval_steps_per_second": 0.472, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 312, | |
| "total_flos": 0.0, | |
| "train_loss": 0.051078493730762065, | |
| "train_runtime": 2008.9755, | |
| "train_samples_per_second": 4.959, | |
| "train_steps_per_second": 0.155 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 312, | |
| "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 | |
| } | |