| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0695187165775402, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.053475935828877004, | |
| "grad_norm": 54.05966214679458, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": -2.740234851837158, | |
| "logits/rejected": -2.7408976554870605, | |
| "logps/chosen": -260.93475341796875, | |
| "logps/rejected": -211.115966796875, | |
| "loss": 0.6911, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.019142702221870422, | |
| "rewards/margins": 0.0086396224796772, | |
| "rewards/rejected": 0.010503082536160946, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.10695187165775401, | |
| "grad_norm": 48.125498996039966, | |
| "learning_rate": 1e-06, | |
| "logits/chosen": -2.709315776824951, | |
| "logits/rejected": -2.7182488441467285, | |
| "logps/chosen": -258.59228515625, | |
| "logps/rejected": -225.8120880126953, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.472653865814209, | |
| "rewards/margins": 0.14731308817863464, | |
| "rewards/rejected": 0.32534077763557434, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16042780748663102, | |
| "grad_norm": 44.380678098165866, | |
| "learning_rate": 9.991477798614637e-07, | |
| "logits/chosen": -2.6404635906219482, | |
| "logits/rejected": -2.6321334838867188, | |
| "logps/chosen": -245.1427764892578, | |
| "logps/rejected": -204.20162963867188, | |
| "loss": 0.6283, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 1.1205874681472778, | |
| "rewards/margins": 0.548652172088623, | |
| "rewards/rejected": 0.5719352960586548, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.21390374331550802, | |
| "grad_norm": 45.00568883373122, | |
| "learning_rate": 9.965940245625131e-07, | |
| "logits/chosen": -2.5277099609375, | |
| "logits/rejected": -2.532583236694336, | |
| "logps/chosen": -265.33990478515625, | |
| "logps/rejected": -236.9266815185547, | |
| "loss": 0.6455, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 1.2614521980285645, | |
| "rewards/margins": 0.9182946085929871, | |
| "rewards/rejected": 0.34315773844718933, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.26737967914438504, | |
| "grad_norm": 49.76731551004973, | |
| "learning_rate": 9.923474395499264e-07, | |
| "logits/chosen": -2.448563814163208, | |
| "logits/rejected": -2.4464802742004395, | |
| "logps/chosen": -234.9072723388672, | |
| "logps/rejected": -208.8161163330078, | |
| "loss": 0.6152, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 1.3159387111663818, | |
| "rewards/margins": 1.1293729543685913, | |
| "rewards/rejected": 0.18656572699546814, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.32085561497326204, | |
| "grad_norm": 37.50219389396827, | |
| "learning_rate": 9.86422500924775e-07, | |
| "logits/chosen": -2.470647096633911, | |
| "logits/rejected": -2.449976682662964, | |
| "logps/chosen": -256.57049560546875, | |
| "logps/rejected": -202.13229370117188, | |
| "loss": 0.5704, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 1.158319354057312, | |
| "rewards/margins": 0.9538521766662598, | |
| "rewards/rejected": 0.20446734130382538, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.37433155080213903, | |
| "grad_norm": 39.982844211127215, | |
| "learning_rate": 9.788394060951227e-07, | |
| "logits/chosen": -2.488635540008545, | |
| "logits/rejected": -2.4961345195770264, | |
| "logps/chosen": -247.16757202148438, | |
| "logps/rejected": -212.7868194580078, | |
| "loss": 0.6044, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": 0.8941676020622253, | |
| "rewards/margins": 0.9727466702461243, | |
| "rewards/rejected": -0.07857902348041534, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.42780748663101603, | |
| "grad_norm": 43.02526224638259, | |
| "learning_rate": 9.696240049254742e-07, | |
| "logits/chosen": -2.5331573486328125, | |
| "logits/rejected": -2.5155434608459473, | |
| "logps/chosen": -243.4714813232422, | |
| "logps/rejected": -212.7275848388672, | |
| "loss": 0.567, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.5864531397819519, | |
| "rewards/margins": 1.3161426782608032, | |
| "rewards/rejected": -0.7296894788742065, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.48128342245989303, | |
| "grad_norm": 38.62113257038537, | |
| "learning_rate": 9.588077116176756e-07, | |
| "logits/chosen": -2.5625014305114746, | |
| "logits/rejected": -2.5336291790008545, | |
| "logps/chosen": -257.4891662597656, | |
| "logps/rejected": -241.340576171875, | |
| "loss": 0.5311, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": 0.603995144367218, | |
| "rewards/margins": 1.318690538406372, | |
| "rewards/rejected": -0.7146955132484436, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.5347593582887701, | |
| "grad_norm": 37.205626252672076, | |
| "learning_rate": 9.464273976236516e-07, | |
| "logits/chosen": -2.555994749069214, | |
| "logits/rejected": -2.5348007678985596, | |
| "logps/chosen": -246.0524444580078, | |
| "logps/rejected": -224.68038940429688, | |
| "loss": 0.5372, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.7058802843093872, | |
| "rewards/margins": 1.3246526718139648, | |
| "rewards/rejected": -0.6187724471092224, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5347593582887701, | |
| "eval_logits/chosen": -2.565211772918701, | |
| "eval_logits/rejected": -2.5557875633239746, | |
| "eval_logps/chosen": -242.1534881591797, | |
| "eval_logps/rejected": -226.5532989501953, | |
| "eval_loss": 0.5589047074317932, | |
| "eval_rewards/accuracies": 0.7708333134651184, | |
| "eval_rewards/chosen": 0.888508141040802, | |
| "eval_rewards/margins": 1.6572500467300415, | |
| "eval_rewards/rejected": -0.76874178647995, | |
| "eval_runtime": 179.6866, | |
| "eval_samples_per_second": 14.804, | |
| "eval_steps_per_second": 0.234, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5882352941176471, | |
| "grad_norm": 41.38413864195239, | |
| "learning_rate": 9.325252659550308e-07, | |
| "logits/chosen": -2.564159393310547, | |
| "logits/rejected": -2.562894105911255, | |
| "logps/chosen": -239.49169921875, | |
| "logps/rejected": -221.1046905517578, | |
| "loss": 0.5664, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.8376470804214478, | |
| "rewards/margins": 1.171051263809204, | |
| "rewards/rejected": -0.3334041237831116, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.6417112299465241, | |
| "grad_norm": 38.183548449406196, | |
| "learning_rate": 9.171487073181197e-07, | |
| "logits/chosen": -2.5205636024475098, | |
| "logits/rejected": -2.5134239196777344, | |
| "logps/chosen": -247.12869262695312, | |
| "logps/rejected": -223.0773468017578, | |
| "loss": 0.5445, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": 0.6957036256790161, | |
| "rewards/margins": 1.3918036222457886, | |
| "rewards/rejected": -0.6960999965667725, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6951871657754011, | |
| "grad_norm": 38.105501822781726, | |
| "learning_rate": 9.003501385646448e-07, | |
| "logits/chosen": -2.5033719539642334, | |
| "logits/rejected": -2.473064422607422, | |
| "logps/chosen": -230.86160278320312, | |
| "logps/rejected": -228.2729949951172, | |
| "loss": 0.5983, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.8065773248672485, | |
| "rewards/margins": 1.1710764169692993, | |
| "rewards/rejected": -0.3644990026950836, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.7486631016042781, | |
| "grad_norm": 32.7431204523646, | |
| "learning_rate": 8.821868240089676e-07, | |
| "logits/chosen": -2.4544219970703125, | |
| "logits/rejected": -2.4576900005340576, | |
| "logps/chosen": -241.304443359375, | |
| "logps/rejected": -234.3035125732422, | |
| "loss": 0.5341, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": 1.1656383275985718, | |
| "rewards/margins": 2.016019582748413, | |
| "rewards/rejected": -0.8503812551498413, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.8021390374331551, | |
| "grad_norm": 35.17657050274663, | |
| "learning_rate": 8.62720680220876e-07, | |
| "logits/chosen": -2.4114229679107666, | |
| "logits/rejected": -2.374790906906128, | |
| "logps/chosen": -244.16860961914062, | |
| "logps/rejected": -230.75845336914062, | |
| "loss": 0.5453, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.4619104862213135, | |
| "rewards/margins": 1.3655352592468262, | |
| "rewards/rejected": -0.9036247134208679, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.8556149732620321, | |
| "grad_norm": 41.012118211896635, | |
| "learning_rate": 8.420180649593929e-07, | |
| "logits/chosen": -2.25986647605896, | |
| "logits/rejected": -2.228325366973877, | |
| "logps/chosen": -249.28677368164062, | |
| "logps/rejected": -226.74453735351562, | |
| "loss": 0.5689, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.25673240423202515, | |
| "rewards/margins": 1.7667404413223267, | |
| "rewards/rejected": -1.5100078582763672, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 32.649741196796036, | |
| "learning_rate": 8.201495509671036e-07, | |
| "logits/chosen": -2.2017948627471924, | |
| "logits/rejected": -2.1529741287231445, | |
| "logps/chosen": -272.0859069824219, | |
| "logps/rejected": -241.6033172607422, | |
| "loss": 0.4923, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.13092926144599915, | |
| "rewards/margins": 1.9662458896636963, | |
| "rewards/rejected": -1.8353164196014404, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.9625668449197861, | |
| "grad_norm": 36.664818682351935, | |
| "learning_rate": 7.971896853961042e-07, | |
| "logits/chosen": -2.264296531677246, | |
| "logits/rejected": -2.21388840675354, | |
| "logps/chosen": -250.2943572998047, | |
| "logps/rejected": -235.5312042236328, | |
| "loss": 0.4863, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.07481655478477478, | |
| "rewards/margins": 1.6215366125106812, | |
| "rewards/rejected": -1.546720027923584, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.0160427807486632, | |
| "grad_norm": 17.846722548016714, | |
| "learning_rate": 7.732167356856654e-07, | |
| "logits/chosen": -2.2805099487304688, | |
| "logits/rejected": -2.2655348777770996, | |
| "logps/chosen": -253.1170196533203, | |
| "logps/rejected": -233.9530487060547, | |
| "loss": 0.4497, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.8429386019706726, | |
| "rewards/margins": 2.112081289291382, | |
| "rewards/rejected": -1.269142746925354, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.0695187165775402, | |
| "grad_norm": 23.335980228800327, | |
| "learning_rate": 7.48312422757881e-07, | |
| "logits/chosen": -2.319995164871216, | |
| "logits/rejected": -2.2926743030548096, | |
| "logps/chosen": -255.7274627685547, | |
| "logps/rejected": -237.66000366210938, | |
| "loss": 0.2535, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.8762633204460144, | |
| "rewards/margins": 2.882870674133301, | |
| "rewards/rejected": -2.0066072940826416, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.0695187165775402, | |
| "eval_logits/chosen": -2.3488855361938477, | |
| "eval_logits/rejected": -2.315967559814453, | |
| "eval_logps/chosen": -244.4204864501953, | |
| "eval_logps/rejected": -233.458984375, | |
| "eval_loss": 0.5071699619293213, | |
| "eval_rewards/accuracies": 0.8154761791229248, | |
| "eval_rewards/chosen": 0.6618079543113708, | |
| "eval_rewards/margins": 2.1211202144622803, | |
| "eval_rewards/rejected": -1.459312081336975, | |
| "eval_runtime": 179.4652, | |
| "eval_samples_per_second": 14.822, | |
| "eval_steps_per_second": 0.234, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 279, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1178822762299392.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |