| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.2461059190031152, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06230529595015576, | |
| "grad_norm": 64.78196225600344, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": -2.7243542671203613, | |
| "logits/rejected": -2.7354743480682373, | |
| "logps/chosen": -260.3916320800781, | |
| "logps/rejected": -244.31298828125, | |
| "loss": 0.6904, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.012873289175331593, | |
| "rewards/margins": 0.0037455155979841948, | |
| "rewards/rejected": 0.009127774275839329, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.12461059190031153, | |
| "grad_norm": 45.8994374201786, | |
| "learning_rate": 1e-06, | |
| "logits/chosen": -2.6493823528289795, | |
| "logits/rejected": -2.650038957595825, | |
| "logps/chosen": -234.55868530273438, | |
| "logps/rejected": -202.4860076904297, | |
| "loss": 0.6405, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.4499489665031433, | |
| "rewards/margins": 0.15348409116268158, | |
| "rewards/rejected": 0.29646486043930054, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.18691588785046728, | |
| "grad_norm": 39.36900659105447, | |
| "learning_rate": 9.988343845952696e-07, | |
| "logits/chosen": -2.487607479095459, | |
| "logits/rejected": -2.481687068939209, | |
| "logps/chosen": -227.55001831054688, | |
| "logps/rejected": -217.2289581298828, | |
| "loss": 0.6294, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 1.2727148532867432, | |
| "rewards/margins": 0.4717913269996643, | |
| "rewards/rejected": 0.8009236454963684, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.24922118380062305, | |
| "grad_norm": 48.3442406976696, | |
| "learning_rate": 9.953429730181652e-07, | |
| "logits/chosen": -2.409071683883667, | |
| "logits/rejected": -2.3922438621520996, | |
| "logps/chosen": -244.6020965576172, | |
| "logps/rejected": -230.30615234375, | |
| "loss": 0.6128, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 1.3646974563598633, | |
| "rewards/margins": 0.8445035815238953, | |
| "rewards/rejected": 0.5201937556266785, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3115264797507788, | |
| "grad_norm": 43.25375347405899, | |
| "learning_rate": 9.895420438411615e-07, | |
| "logits/chosen": -2.404625654220581, | |
| "logits/rejected": -2.380873203277588, | |
| "logps/chosen": -260.6168212890625, | |
| "logps/rejected": -226.8410186767578, | |
| "loss": 0.5858, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": 1.0510780811309814, | |
| "rewards/margins": 1.0232232809066772, | |
| "rewards/rejected": 0.027854669839143753, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.37383177570093457, | |
| "grad_norm": 35.13451219232732, | |
| "learning_rate": 9.814586436738997e-07, | |
| "logits/chosen": -2.457035779953003, | |
| "logits/rejected": -2.448774576187134, | |
| "logps/chosen": -257.907470703125, | |
| "logps/rejected": -218.4612579345703, | |
| "loss": 0.5578, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 1.1478369235992432, | |
| "rewards/margins": 1.3836512565612793, | |
| "rewards/rejected": -0.23581421375274658, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.43613707165109034, | |
| "grad_norm": 41.912297815106236, | |
| "learning_rate": 9.711304610594102e-07, | |
| "logits/chosen": -2.485905408859253, | |
| "logits/rejected": -2.459043025970459, | |
| "logps/chosen": -246.7182159423828, | |
| "logps/rejected": -219.68246459960938, | |
| "loss": 0.593, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.6787748336791992, | |
| "rewards/margins": 1.0257090330123901, | |
| "rewards/rejected": -0.34693413972854614, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.4984423676012461, | |
| "grad_norm": 47.725973440299406, | |
| "learning_rate": 9.586056507527264e-07, | |
| "logits/chosen": -2.4728102684020996, | |
| "logits/rejected": -2.4659764766693115, | |
| "logps/chosen": -242.521728515625, | |
| "logps/rejected": -244.7397003173828, | |
| "loss": 0.5851, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.7883692979812622, | |
| "rewards/margins": 1.2785449028015137, | |
| "rewards/rejected": -0.4901755452156067, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5607476635514018, | |
| "grad_norm": 36.45457422056591, | |
| "learning_rate": 9.439426092011875e-07, | |
| "logits/chosen": -2.435615301132202, | |
| "logits/rejected": -2.4443929195404053, | |
| "logps/chosen": -280.24755859375, | |
| "logps/rejected": -212.96292114257812, | |
| "loss": 0.558, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 1.2600438594818115, | |
| "rewards/margins": 1.4364185333251953, | |
| "rewards/rejected": -0.17637479305267334, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.6230529595015576, | |
| "grad_norm": 37.12133742483065, | |
| "learning_rate": 9.272097022732443e-07, | |
| "logits/chosen": -2.447756052017212, | |
| "logits/rejected": -2.4266324043273926, | |
| "logps/chosen": -246.95840454101562, | |
| "logps/rejected": -214.29959106445312, | |
| "loss": 0.5451, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 1.2870022058486938, | |
| "rewards/margins": 1.596084475517273, | |
| "rewards/rejected": -0.3090822994709015, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6230529595015576, | |
| "eval_logits/chosen": -2.4445273876190186, | |
| "eval_logits/rejected": -2.4265356063842773, | |
| "eval_logps/chosen": -268.5102844238281, | |
| "eval_logps/rejected": -221.56869506835938, | |
| "eval_loss": 0.5847126245498657, | |
| "eval_rewards/accuracies": 0.7465277910232544, | |
| "eval_rewards/chosen": 1.2566064596176147, | |
| "eval_rewards/margins": 1.209053635597229, | |
| "eval_rewards/rejected": 0.04755274951457977, | |
| "eval_runtime": 151.9316, | |
| "eval_samples_per_second": 15.007, | |
| "eval_steps_per_second": 0.237, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6853582554517134, | |
| "grad_norm": 40.79262464200207, | |
| "learning_rate": 9.084849465052209e-07, | |
| "logits/chosen": -2.3776774406433105, | |
| "logits/rejected": -2.3956453800201416, | |
| "logps/chosen": -257.2913513183594, | |
| "logps/rejected": -197.3280029296875, | |
| "loss": 0.5601, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 1.311511516571045, | |
| "rewards/margins": 1.5678449869155884, | |
| "rewards/rejected": -0.2563334107398987, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.7476635514018691, | |
| "grad_norm": 48.34269415680465, | |
| "learning_rate": 8.878556453522099e-07, | |
| "logits/chosen": -2.3521735668182373, | |
| "logits/rejected": -2.3181591033935547, | |
| "logps/chosen": -233.2803497314453, | |
| "logps/rejected": -218.5789031982422, | |
| "loss": 0.5833, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.7265979647636414, | |
| "rewards/margins": 1.4349805116653442, | |
| "rewards/rejected": -0.7083825469017029, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.8099688473520249, | |
| "grad_norm": 37.4927506245011, | |
| "learning_rate": 8.654179821390621e-07, | |
| "logits/chosen": -2.362112522125244, | |
| "logits/rejected": -2.3338117599487305, | |
| "logps/chosen": -267.119140625, | |
| "logps/rejected": -216.8646240234375, | |
| "loss": 0.5886, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": 0.6950536966323853, | |
| "rewards/margins": 1.5640451908111572, | |
| "rewards/rejected": -0.8689913749694824, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.8722741433021807, | |
| "grad_norm": 39.43481962342951, | |
| "learning_rate": 8.41276571609327e-07, | |
| "logits/chosen": -2.3775908946990967, | |
| "logits/rejected": -2.365400552749634, | |
| "logps/chosen": -277.951904296875, | |
| "logps/rejected": -225.5980682373047, | |
| "loss": 0.5456, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 1.0310388803482056, | |
| "rewards/margins": 1.5813945531845093, | |
| "rewards/rejected": -0.5503557324409485, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.9345794392523364, | |
| "grad_norm": 41.2092804317616, | |
| "learning_rate": 8.155439721630264e-07, | |
| "logits/chosen": -2.337092638015747, | |
| "logits/rejected": -2.31592059135437, | |
| "logps/chosen": -252.63143920898438, | |
| "logps/rejected": -201.25433349609375, | |
| "loss": 0.5874, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.8362587690353394, | |
| "rewards/margins": 1.4375728368759155, | |
| "rewards/rejected": -0.6013139486312866, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.9968847352024922, | |
| "grad_norm": 32.58419185513781, | |
| "learning_rate": 7.883401610574336e-07, | |
| "logits/chosen": -2.255605459213257, | |
| "logits/rejected": -2.23878812789917, | |
| "logps/chosen": -257.6908264160156, | |
| "logps/rejected": -198.56063842773438, | |
| "loss": 0.5088, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.8411666750907898, | |
| "rewards/margins": 1.4409376382827759, | |
| "rewards/rejected": -0.5997709035873413, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.0591900311526479, | |
| "grad_norm": 19.724965519418905, | |
| "learning_rate": 7.597919750177168e-07, | |
| "logits/chosen": -2.2557170391082764, | |
| "logits/rejected": -2.2346577644348145, | |
| "logps/chosen": -261.6788330078125, | |
| "logps/rejected": -238.0162353515625, | |
| "loss": 0.2444, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 1.7993049621582031, | |
| "rewards/margins": 3.0407943725585938, | |
| "rewards/rejected": -1.2414895296096802, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.1214953271028036, | |
| "grad_norm": 22.70642349887756, | |
| "learning_rate": 7.30032518865576e-07, | |
| "logits/chosen": -2.3579375743865967, | |
| "logits/rejected": -2.339137554168701, | |
| "logps/chosen": -234.2419891357422, | |
| "logps/rejected": -222.1098175048828, | |
| "loss": 0.2253, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 2.0569443702697754, | |
| "rewards/margins": 3.1539688110351562, | |
| "rewards/rejected": -1.0970245599746704, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.1838006230529594, | |
| "grad_norm": 26.513380455541846, | |
| "learning_rate": 6.992005449231207e-07, | |
| "logits/chosen": -2.4564146995544434, | |
| "logits/rejected": -2.480577230453491, | |
| "logps/chosen": -241.68600463867188, | |
| "logps/rejected": -214.6504364013672, | |
| "loss": 0.2597, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 1.810720682144165, | |
| "rewards/margins": 3.2071242332458496, | |
| "rewards/rejected": -1.3964035511016846, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.2461059190031152, | |
| "grad_norm": 25.112538207360036, | |
| "learning_rate": 6.67439806085493e-07, | |
| "logits/chosen": -2.5166354179382324, | |
| "logits/rejected": -2.513838291168213, | |
| "logps/chosen": -245.987548828125, | |
| "logps/rejected": -252.9072265625, | |
| "loss": 0.2411, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 1.6626945734024048, | |
| "rewards/margins": 3.801987409591675, | |
| "rewards/rejected": -2.1392927169799805, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.2461059190031152, | |
| "eval_logits/chosen": -2.5858781337738037, | |
| "eval_logits/rejected": -2.563420534133911, | |
| "eval_logps/chosen": -270.93853759765625, | |
| "eval_logps/rejected": -233.65652465820312, | |
| "eval_loss": 0.5330603718757629, | |
| "eval_rewards/accuracies": 0.7881944179534912, | |
| "eval_rewards/chosen": 1.0137810707092285, | |
| "eval_rewards/margins": 2.17501163482666, | |
| "eval_rewards/rejected": -1.1612308025360107, | |
| "eval_runtime": 151.6106, | |
| "eval_samples_per_second": 15.039, | |
| "eval_steps_per_second": 0.237, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 240, | |
| "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 | |
| } | |