| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 478, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0020920502092050207, | |
| "grad_norm": 9.710838317871094, | |
| "learning_rate": 1.0416666666666666e-08, | |
| "logits/chosen": -3.161454916000366, | |
| "logits/rejected": -3.0680501461029053, | |
| "logps/chosen": -437.20782470703125, | |
| "logps/rejected": -343.8380432128906, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.05230125523012552, | |
| "grad_norm": 8.52570629119873, | |
| "learning_rate": 2.604166666666667e-07, | |
| "logits/chosen": -3.30265212059021, | |
| "logits/rejected": -3.2195777893066406, | |
| "logps/chosen": -432.7400207519531, | |
| "logps/rejected": -391.2707824707031, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.5234375, | |
| "rewards/chosen": 0.0006945470231585205, | |
| "rewards/margins": 0.0007254942320287228, | |
| "rewards/rejected": -3.094731437158771e-05, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10460251046025104, | |
| "grad_norm": 9.973247528076172, | |
| "learning_rate": 4.999733114418725e-07, | |
| "logits/chosen": -3.299149751663208, | |
| "logits/rejected": -3.2438504695892334, | |
| "logps/chosen": -446.8968811035156, | |
| "logps/rejected": -404.587158203125, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.5799999833106995, | |
| "rewards/chosen": 0.005268932785838842, | |
| "rewards/margins": 0.002896952675655484, | |
| "rewards/rejected": 0.002371980343014002, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15690376569037656, | |
| "grad_norm": 9.357086181640625, | |
| "learning_rate": 4.951516761176343e-07, | |
| "logits/chosen": -3.302896738052368, | |
| "logits/rejected": -3.225714683532715, | |
| "logps/chosen": -480.6836853027344, | |
| "logps/rejected": -436.4853515625, | |
| "loss": 0.6871, | |
| "rewards/accuracies": 0.5975000262260437, | |
| "rewards/chosen": 0.013074235059320927, | |
| "rewards/margins": 0.012715624645352364, | |
| "rewards/rejected": 0.0003586093371268362, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "grad_norm": 8.75236988067627, | |
| "learning_rate": 4.821741763807186e-07, | |
| "logits/chosen": -3.2598507404327393, | |
| "logits/rejected": -3.2092175483703613, | |
| "logps/chosen": -426.1629638671875, | |
| "logps/rejected": -389.50421142578125, | |
| "loss": 0.6795, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.003723819274455309, | |
| "rewards/margins": 0.02804265171289444, | |
| "rewards/rejected": -0.02431883104145527, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "eval_logits/chosen": -3.313567638397217, | |
| "eval_logits/rejected": -3.2565112113952637, | |
| "eval_logps/chosen": -423.0196533203125, | |
| "eval_logps/rejected": -407.8036804199219, | |
| "eval_loss": 0.6759119629859924, | |
| "eval_rewards/accuracies": 0.62890625, | |
| "eval_rewards/chosen": 0.0017230990342795849, | |
| "eval_rewards/margins": 0.0345395989716053, | |
| "eval_rewards/rejected": -0.03281649947166443, | |
| "eval_runtime": 8.2733, | |
| "eval_samples_per_second": 241.742, | |
| "eval_steps_per_second": 3.868, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2615062761506276, | |
| "grad_norm": 8.889654159545898, | |
| "learning_rate": 4.614725560802639e-07, | |
| "logits/chosen": -3.2867064476013184, | |
| "logits/rejected": -3.2023117542266846, | |
| "logps/chosen": -430.924560546875, | |
| "logps/rejected": -382.1640319824219, | |
| "loss": 0.6731, | |
| "rewards/accuracies": 0.6474999785423279, | |
| "rewards/chosen": -0.01004675030708313, | |
| "rewards/margins": 0.04363078624010086, | |
| "rewards/rejected": -0.05367753654718399, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3138075313807531, | |
| "grad_norm": 9.882610321044922, | |
| "learning_rate": 4.337355301007335e-07, | |
| "logits/chosen": -3.222916841506958, | |
| "logits/rejected": -3.1899585723876953, | |
| "logps/chosen": -443.3583679199219, | |
| "logps/rejected": -414.344970703125, | |
| "loss": 0.667, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.017106110230088234, | |
| "rewards/margins": 0.06052257865667343, | |
| "rewards/rejected": -0.07762870192527771, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.36610878661087864, | |
| "grad_norm": 9.968969345092773, | |
| "learning_rate": 3.9988587174999306e-07, | |
| "logits/chosen": -3.2201168537139893, | |
| "logits/rejected": -3.131910800933838, | |
| "logps/chosen": -477.7035217285156, | |
| "logps/rejected": -405.7104797363281, | |
| "loss": 0.6574, | |
| "rewards/accuracies": 0.6399999856948853, | |
| "rewards/chosen": -0.03587143123149872, | |
| "rewards/margins": 0.09045815467834473, | |
| "rewards/rejected": -0.12632958590984344, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "grad_norm": 9.230700492858887, | |
| "learning_rate": 3.610497133404795e-07, | |
| "logits/chosen": -3.23995304107666, | |
| "logits/rejected": -3.1550867557525635, | |
| "logps/chosen": -431.6617126464844, | |
| "logps/rejected": -396.62774658203125, | |
| "loss": 0.6584, | |
| "rewards/accuracies": 0.6349999904632568, | |
| "rewards/chosen": -0.06991340965032578, | |
| "rewards/margins": 0.08443903923034668, | |
| "rewards/rejected": -0.15435244143009186, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "eval_logits/chosen": -3.2767982482910156, | |
| "eval_logits/rejected": -3.2240023612976074, | |
| "eval_logps/chosen": -429.85614013671875, | |
| "eval_logps/rejected": -420.6952209472656, | |
| "eval_loss": 0.6533502340316772, | |
| "eval_rewards/accuracies": 0.64453125, | |
| "eval_rewards/chosen": -0.06664139777421951, | |
| "eval_rewards/margins": 0.09509073942899704, | |
| "eval_rewards/rejected": -0.16173213720321655, | |
| "eval_runtime": 8.2763, | |
| "eval_samples_per_second": 241.653, | |
| "eval_steps_per_second": 3.866, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4707112970711297, | |
| "grad_norm": 10.79430103302002, | |
| "learning_rate": 3.185190812915646e-07, | |
| "logits/chosen": -3.1671783924102783, | |
| "logits/rejected": -3.118861436843872, | |
| "logps/chosen": -446.6968994140625, | |
| "logps/rejected": -410.1864013671875, | |
| "loss": 0.6518, | |
| "rewards/accuracies": 0.6449999809265137, | |
| "rewards/chosen": -0.08074235171079636, | |
| "rewards/margins": 0.10100732743740082, | |
| "rewards/rejected": -0.18174967169761658, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5230125523012552, | |
| "grad_norm": 10.086767196655273, | |
| "learning_rate": 2.7370891215954565e-07, | |
| "logits/chosen": -3.1980080604553223, | |
| "logits/rejected": -3.1623446941375732, | |
| "logps/chosen": -440.4261474609375, | |
| "logps/rejected": -439.9627685546875, | |
| "loss": 0.6476, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.07635506987571716, | |
| "rewards/margins": 0.11709018051624298, | |
| "rewards/rejected": -0.19344525039196014, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5753138075313807, | |
| "grad_norm": 9.709197044372559, | |
| "learning_rate": 2.2810997961375938e-07, | |
| "logits/chosen": -3.216128349304199, | |
| "logits/rejected": -3.1448960304260254, | |
| "logps/chosen": -425.5062561035156, | |
| "logps/rejected": -391.5084228515625, | |
| "loss": 0.6461, | |
| "rewards/accuracies": 0.6800000071525574, | |
| "rewards/chosen": -0.072402223944664, | |
| "rewards/margins": 0.1306913197040558, | |
| "rewards/rejected": -0.2030935436487198, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "grad_norm": 10.980072975158691, | |
| "learning_rate": 1.8323929841460178e-07, | |
| "logits/chosen": -3.2006824016571045, | |
| "logits/rejected": -3.1383109092712402, | |
| "logps/chosen": -443.0799865722656, | |
| "logps/rejected": -436.2774658203125, | |
| "loss": 0.6494, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.1245601624250412, | |
| "rewards/margins": 0.10581608861684799, | |
| "rewards/rejected": -0.2303762584924698, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "eval_logits/chosen": -3.2553329467773438, | |
| "eval_logits/rejected": -3.2049574851989746, | |
| "eval_logps/chosen": -433.9639892578125, | |
| "eval_logps/rejected": -428.6236572265625, | |
| "eval_loss": 0.6438009142875671, | |
| "eval_rewards/accuracies": 0.62109375, | |
| "eval_rewards/chosen": -0.10771973431110382, | |
| "eval_rewards/margins": 0.13329659402370453, | |
| "eval_rewards/rejected": -0.24101632833480835, | |
| "eval_runtime": 8.3559, | |
| "eval_samples_per_second": 239.351, | |
| "eval_steps_per_second": 3.83, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6799163179916318, | |
| "grad_norm": 9.95361614227295, | |
| "learning_rate": 1.4058965538597032e-07, | |
| "logits/chosen": -3.2326276302337646, | |
| "logits/rejected": -3.198971748352051, | |
| "logps/chosen": -443.37371826171875, | |
| "logps/rejected": -433.8561706542969, | |
| "loss": 0.6392, | |
| "rewards/accuracies": 0.6399999856948853, | |
| "rewards/chosen": -0.11981040984392166, | |
| "rewards/margins": 0.12150833755731583, | |
| "rewards/rejected": -0.2413187474012375, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7322175732217573, | |
| "grad_norm": 10.580938339233398, | |
| "learning_rate": 1.0157994641835734e-07, | |
| "logits/chosen": -3.1840951442718506, | |
| "logits/rejected": -3.0981223583221436, | |
| "logps/chosen": -430.7939453125, | |
| "logps/rejected": -403.5060119628906, | |
| "loss": 0.6447, | |
| "rewards/accuracies": 0.6225000023841858, | |
| "rewards/chosen": -0.15974169969558716, | |
| "rewards/margins": 0.09658970683813095, | |
| "rewards/rejected": -0.2563314139842987, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7845188284518828, | |
| "grad_norm": 8.326279640197754, | |
| "learning_rate": 6.75079717232744e-08, | |
| "logits/chosen": -3.2011187076568604, | |
| "logits/rejected": -3.0969460010528564, | |
| "logps/chosen": -465.22344970703125, | |
| "logps/rejected": -392.5793762207031, | |
| "loss": 0.6411, | |
| "rewards/accuracies": 0.6549999713897705, | |
| "rewards/chosen": -0.11683624982833862, | |
| "rewards/margins": 0.14286838471889496, | |
| "rewards/rejected": -0.2597046196460724, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "grad_norm": 8.851872444152832, | |
| "learning_rate": 3.9507259776993954e-08, | |
| "logits/chosen": -3.221069097518921, | |
| "logits/rejected": -3.1744155883789062, | |
| "logps/chosen": -449.8016052246094, | |
| "logps/rejected": -409.0178527832031, | |
| "loss": 0.6428, | |
| "rewards/accuracies": 0.6524999737739563, | |
| "rewards/chosen": -0.08884063363075256, | |
| "rewards/margins": 0.13940051198005676, | |
| "rewards/rejected": -0.22824116051197052, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "eval_logits/chosen": -3.254263401031494, | |
| "eval_logits/rejected": -3.204622507095337, | |
| "eval_logps/chosen": -433.20001220703125, | |
| "eval_logps/rejected": -428.88836669921875, | |
| "eval_loss": 0.6415477395057678, | |
| "eval_rewards/accuracies": 0.62109375, | |
| "eval_rewards/chosen": -0.10007989406585693, | |
| "eval_rewards/margins": 0.14358317852020264, | |
| "eval_rewards/rejected": -0.24366310238838196, | |
| "eval_runtime": 8.7287, | |
| "eval_samples_per_second": 229.13, | |
| "eval_steps_per_second": 3.666, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8891213389121339, | |
| "grad_norm": 10.138018608093262, | |
| "learning_rate": 1.850935636255496e-08, | |
| "logits/chosen": -3.1941397190093994, | |
| "logits/rejected": -3.1335413455963135, | |
| "logps/chosen": -463.3765869140625, | |
| "logps/rejected": -418.0171203613281, | |
| "loss": 0.6393, | |
| "rewards/accuracies": 0.6399999856948853, | |
| "rewards/chosen": -0.11999661475419998, | |
| "rewards/margins": 0.13007262349128723, | |
| "rewards/rejected": -0.2500692307949066, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9414225941422594, | |
| "grad_norm": 9.687002182006836, | |
| "learning_rate": 5.212833302556258e-09, | |
| "logits/chosen": -3.202920913696289, | |
| "logits/rejected": -3.14504337310791, | |
| "logps/chosen": -442.259765625, | |
| "logps/rejected": -406.7275085449219, | |
| "loss": 0.6419, | |
| "rewards/accuracies": 0.6050000190734863, | |
| "rewards/chosen": -0.13094988465309143, | |
| "rewards/margins": 0.1017264574766159, | |
| "rewards/rejected": -0.23267632722854614, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9937238493723849, | |
| "grad_norm": 8.862220764160156, | |
| "learning_rate": 6.004792024680294e-11, | |
| "logits/chosen": -3.1527392864227295, | |
| "logits/rejected": -3.1062843799591064, | |
| "logps/chosen": -445.8639831542969, | |
| "logps/rejected": -418.4268493652344, | |
| "loss": 0.6377, | |
| "rewards/accuracies": 0.6575000286102295, | |
| "rewards/chosen": -0.11145105212926865, | |
| "rewards/margins": 0.15232053399085999, | |
| "rewards/rejected": -0.26377159357070923, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 478, | |
| "total_flos": 0.0, | |
| "train_loss": 0.6571792745689967, | |
| "train_runtime": 784.6622, | |
| "train_samples_per_second": 77.913, | |
| "train_steps_per_second": 0.609 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 478, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |