| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 313, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003194888178913738, | |
| "grad_norm": 32.14257471485931, | |
| "learning_rate": 5.3191489361702125e-09, | |
| "logits/chosen": -1.7265625, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -249.0, | |
| "logps/rejected": -242.0, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03194888178913738, | |
| "grad_norm": 33.76096895998352, | |
| "learning_rate": 5.3191489361702123e-08, | |
| "logits/chosen": -1.765625, | |
| "logits/rejected": -1.765625, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -244.0, | |
| "loss": 0.6926, | |
| "rewards/accuracies": 0.2083333283662796, | |
| "rewards/chosen": 0.00555419921875, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.00555419921875, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06389776357827476, | |
| "grad_norm": 31.259900823289186, | |
| "learning_rate": 1.0638297872340425e-07, | |
| "logits/chosen": -1.71875, | |
| "logits/rejected": -1.75, | |
| "logps/chosen": -237.0, | |
| "logps/rejected": -241.0, | |
| "loss": 0.6916, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.006256103515625, | |
| "rewards/rejected": -0.006256103515625, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09584664536741214, | |
| "grad_norm": 32.366987695897166, | |
| "learning_rate": 1.5957446808510638e-07, | |
| "logits/chosen": -1.75, | |
| "logits/rejected": -1.75, | |
| "logps/chosen": -245.0, | |
| "logps/rejected": -246.0, | |
| "loss": 0.6946, | |
| "rewards/accuracies": 0.22499999403953552, | |
| "rewards/chosen": 6.109476089477539e-06, | |
| "rewards/margins": -0.0012359619140625, | |
| "rewards/rejected": 0.0012664794921875, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12779552715654952, | |
| "grad_norm": 30.81764382136956, | |
| "learning_rate": 2.127659574468085e-07, | |
| "logits/chosen": -1.703125, | |
| "logits/rejected": -1.734375, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -240.0, | |
| "loss": 0.6818, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.003753662109375, | |
| "rewards/margins": 0.032470703125, | |
| "rewards/rejected": -0.036376953125, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1597444089456869, | |
| "grad_norm": 31.214992328374773, | |
| "learning_rate": 2.659574468085106e-07, | |
| "logits/chosen": -1.7109375, | |
| "logits/rejected": -1.7265625, | |
| "logps/chosen": -242.0, | |
| "logps/rejected": -243.0, | |
| "loss": 0.673, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.0400390625, | |
| "rewards/margins": 0.033935546875, | |
| "rewards/rejected": -0.07373046875, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.19169329073482427, | |
| "grad_norm": 28.88257684001896, | |
| "learning_rate": 3.1914893617021275e-07, | |
| "logits/chosen": -1.7265625, | |
| "logits/rejected": -1.75, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -243.0, | |
| "loss": 0.6296, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.0625, | |
| "rewards/margins": 0.111328125, | |
| "rewards/rejected": -0.173828125, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.22364217252396165, | |
| "grad_norm": 30.584634735099684, | |
| "learning_rate": 3.7234042553191484e-07, | |
| "logits/chosen": -1.7109375, | |
| "logits/rejected": -1.71875, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -250.0, | |
| "loss": 0.6439, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.080078125, | |
| "rewards/margins": 0.19140625, | |
| "rewards/rejected": -0.271484375, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.25559105431309903, | |
| "grad_norm": 29.751500863088314, | |
| "learning_rate": 4.25531914893617e-07, | |
| "logits/chosen": -1.703125, | |
| "logits/rejected": -1.75, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -251.0, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.2060546875, | |
| "rewards/margins": 0.2099609375, | |
| "rewards/rejected": -0.41796875, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.28753993610223644, | |
| "grad_norm": 32.267981389042255, | |
| "learning_rate": 4.787234042553192e-07, | |
| "logits/chosen": -1.7109375, | |
| "logits/rejected": -1.765625, | |
| "logps/chosen": -245.0, | |
| "logps/rejected": -247.0, | |
| "loss": 0.6458, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.349609375, | |
| "rewards/margins": 0.2060546875, | |
| "rewards/rejected": -0.5546875, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3194888178913738, | |
| "grad_norm": 31.916118030442767, | |
| "learning_rate": 4.964497041420119e-07, | |
| "logits/chosen": -1.6953125, | |
| "logits/rejected": -1.7109375, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -243.0, | |
| "loss": 0.6301, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.29296875, | |
| "rewards/margins": 0.1767578125, | |
| "rewards/rejected": -0.470703125, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3514376996805112, | |
| "grad_norm": 30.563502769937987, | |
| "learning_rate": 4.905325443786982e-07, | |
| "logits/chosen": -1.6796875, | |
| "logits/rejected": -1.734375, | |
| "logps/chosen": -247.0, | |
| "logps/rejected": -250.0, | |
| "loss": 0.6481, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.23046875, | |
| "rewards/margins": 0.2421875, | |
| "rewards/rejected": -0.47265625, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.38338658146964855, | |
| "grad_norm": 26.847013088556583, | |
| "learning_rate": 4.846153846153846e-07, | |
| "logits/chosen": -1.7109375, | |
| "logits/rejected": -1.7109375, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -251.0, | |
| "loss": 0.6152, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2734375, | |
| "rewards/margins": 0.306640625, | |
| "rewards/rejected": -0.578125, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.41533546325878595, | |
| "grad_norm": 31.074353820130803, | |
| "learning_rate": 4.78698224852071e-07, | |
| "logits/chosen": -1.703125, | |
| "logits/rejected": -1.71875, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -260.0, | |
| "loss": 0.604, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.47265625, | |
| "rewards/margins": 0.291015625, | |
| "rewards/rejected": -0.76171875, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.4472843450479233, | |
| "grad_norm": 32.192668717979046, | |
| "learning_rate": 4.727810650887574e-07, | |
| "logits/chosen": -1.6796875, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -254.0, | |
| "loss": 0.5609, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.416015625, | |
| "rewards/margins": 0.5234375, | |
| "rewards/rejected": -0.9375, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4792332268370607, | |
| "grad_norm": 32.155266134791546, | |
| "learning_rate": 4.668639053254438e-07, | |
| "logits/chosen": -1.6953125, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -251.0, | |
| "loss": 0.5987, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.45703125, | |
| "rewards/margins": 0.453125, | |
| "rewards/rejected": -0.91015625, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5111821086261981, | |
| "grad_norm": 30.837637394035802, | |
| "learning_rate": 4.6094674556213014e-07, | |
| "logits/chosen": -1.6875, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -248.0, | |
| "logps/rejected": -252.0, | |
| "loss": 0.6033, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.333984375, | |
| "rewards/margins": 0.373046875, | |
| "rewards/rejected": -0.70703125, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5431309904153354, | |
| "grad_norm": 25.59680166981086, | |
| "learning_rate": 4.5502958579881655e-07, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": -1.6796875, | |
| "logps/chosen": -238.0, | |
| "logps/rejected": -248.0, | |
| "loss": 0.5817, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.322265625, | |
| "rewards/margins": 0.30078125, | |
| "rewards/rejected": -0.62109375, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5750798722044729, | |
| "grad_norm": 29.648466125367857, | |
| "learning_rate": 4.491124260355029e-07, | |
| "logits/chosen": -1.6875, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -245.0, | |
| "logps/rejected": -251.0, | |
| "loss": 0.5874, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.38671875, | |
| "rewards/margins": 0.50390625, | |
| "rewards/rejected": -0.890625, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6070287539936102, | |
| "grad_norm": 28.442764104750204, | |
| "learning_rate": 4.4319526627218936e-07, | |
| "logits/chosen": -1.6875, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -249.0, | |
| "loss": 0.5737, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.294921875, | |
| "rewards/margins": 0.5078125, | |
| "rewards/rejected": -0.80078125, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6389776357827476, | |
| "grad_norm": 26.370679072337964, | |
| "learning_rate": 4.372781065088757e-07, | |
| "logits/chosen": -1.71875, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -256.0, | |
| "loss": 0.5867, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.3828125, | |
| "rewards/margins": 0.3671875, | |
| "rewards/rejected": -0.75, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.670926517571885, | |
| "grad_norm": 24.556255560376357, | |
| "learning_rate": 4.313609467455621e-07, | |
| "logits/chosen": -1.703125, | |
| "logits/rejected": -1.6953125, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -252.0, | |
| "loss": 0.5853, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.349609375, | |
| "rewards/margins": 0.3671875, | |
| "rewards/rejected": -0.71875, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.7028753993610224, | |
| "grad_norm": 29.98817140815093, | |
| "learning_rate": 4.2544378698224847e-07, | |
| "logits/chosen": -1.6796875, | |
| "logits/rejected": -1.71875, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -256.0, | |
| "loss": 0.5872, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.39453125, | |
| "rewards/margins": 0.55078125, | |
| "rewards/rejected": -0.9453125, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.7348242811501597, | |
| "grad_norm": 27.478577985922634, | |
| "learning_rate": 4.195266272189349e-07, | |
| "logits/chosen": -1.671875, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -249.0, | |
| "loss": 0.5549, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3203125, | |
| "rewards/margins": 0.50390625, | |
| "rewards/rejected": -0.82421875, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.7667731629392971, | |
| "grad_norm": 26.416091326825903, | |
| "learning_rate": 4.1360946745562133e-07, | |
| "logits/chosen": -1.703125, | |
| "logits/rejected": -1.71875, | |
| "logps/chosen": -240.0, | |
| "logps/rejected": -262.0, | |
| "loss": 0.538, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.32421875, | |
| "rewards/margins": 0.55078125, | |
| "rewards/rejected": -0.875, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7987220447284346, | |
| "grad_norm": 28.30204123280793, | |
| "learning_rate": 4.076923076923077e-07, | |
| "logits/chosen": -1.6796875, | |
| "logits/rejected": -1.6875, | |
| "logps/chosen": -239.0, | |
| "logps/rejected": -260.0, | |
| "loss": 0.5507, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.38671875, | |
| "rewards/margins": 0.5546875, | |
| "rewards/rejected": -0.9453125, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.8306709265175719, | |
| "grad_norm": 33.17216428250153, | |
| "learning_rate": 4.017751479289941e-07, | |
| "logits/chosen": -1.6953125, | |
| "logits/rejected": -1.7421875, | |
| "logps/chosen": -244.0, | |
| "logps/rejected": -256.0, | |
| "loss": 0.5584, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.27734375, | |
| "rewards/margins": 0.78125, | |
| "rewards/rejected": -1.0625, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.8626198083067093, | |
| "grad_norm": 27.41764842123467, | |
| "learning_rate": 3.9585798816568044e-07, | |
| "logits/chosen": -1.6640625, | |
| "logits/rejected": -1.65625, | |
| "logps/chosen": -238.0, | |
| "logps/rejected": -252.0, | |
| "loss": 0.5742, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.24609375, | |
| "rewards/margins": 0.6015625, | |
| "rewards/rejected": -0.84765625, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.8945686900958466, | |
| "grad_norm": 34.061232014315166, | |
| "learning_rate": 3.8994082840236685e-07, | |
| "logits/chosen": -1.6953125, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -242.0, | |
| "logps/rejected": -251.0, | |
| "loss": 0.5279, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.3203125, | |
| "rewards/margins": 0.65234375, | |
| "rewards/rejected": -0.97265625, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.9265175718849841, | |
| "grad_norm": 28.160658079870018, | |
| "learning_rate": 3.840236686390532e-07, | |
| "logits/chosen": -1.6796875, | |
| "logits/rejected": -1.6953125, | |
| "logps/chosen": -248.0, | |
| "logps/rejected": -252.0, | |
| "loss": 0.5804, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.4296875, | |
| "rewards/margins": 0.45703125, | |
| "rewards/rejected": -0.88671875, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.9584664536741214, | |
| "grad_norm": 29.60484241313221, | |
| "learning_rate": 3.7810650887573966e-07, | |
| "logits/chosen": -1.65625, | |
| "logits/rejected": -1.671875, | |
| "logps/chosen": -247.0, | |
| "logps/rejected": -255.0, | |
| "loss": 0.5635, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.35546875, | |
| "rewards/margins": 0.400390625, | |
| "rewards/rejected": -0.75390625, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.9904153354632588, | |
| "grad_norm": 33.368421615399974, | |
| "learning_rate": 3.72189349112426e-07, | |
| "logits/chosen": -1.7109375, | |
| "logits/rejected": -1.703125, | |
| "logps/chosen": -244.0, | |
| "logps/rejected": -250.0, | |
| "loss": 0.5843, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.2578125, | |
| "rewards/margins": 0.49609375, | |
| "rewards/rejected": -0.75390625, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -1.6953125, | |
| "eval_logits/rejected": -1.7109375, | |
| "eval_logps/chosen": -246.0, | |
| "eval_logps/rejected": -248.0, | |
| "eval_loss": 0.6162499785423279, | |
| "eval_rewards/accuracies": 0.5714285969734192, | |
| "eval_rewards/chosen": -0.380859375, | |
| "eval_rewards/margins": 0.296875, | |
| "eval_rewards/rejected": -0.6796875, | |
| "eval_runtime": 12.6011, | |
| "eval_samples_per_second": 15.872, | |
| "eval_steps_per_second": 0.556, | |
| "step": 313 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 939, | |
| "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": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |