| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.3888888888888888, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06944444444444445, | |
| "grad_norm": 36.698929739881684, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": -2.7539002895355225, | |
| "logits/rejected": -2.7327029705047607, | |
| "logps/chosen": -163.69387817382812, | |
| "logps/rejected": -163.82852172851562, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.0032427930273115635, | |
| "rewards/margins": 0.003264186205342412, | |
| "rewards/rejected": -2.1393225324572995e-05, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.1388888888888889, | |
| "grad_norm": 33.19716118176963, | |
| "learning_rate": 1e-06, | |
| "logits/chosen": -2.725349187850952, | |
| "logits/rejected": -2.7233974933624268, | |
| "logps/chosen": -158.5753631591797, | |
| "logps/rejected": -163.6913299560547, | |
| "loss": 0.6852, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": 0.1477239578962326, | |
| "rewards/margins": 0.009601245634257793, | |
| "rewards/rejected": 0.13812272250652313, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.20833333333333334, | |
| "grad_norm": 32.17261180089332, | |
| "learning_rate": 9.985471028179154e-07, | |
| "logits/chosen": -2.6893043518066406, | |
| "logits/rejected": -2.6972908973693848, | |
| "logps/chosen": -163.91458129882812, | |
| "logps/rejected": -160.10305786132812, | |
| "loss": 0.6754, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.5012314319610596, | |
| "rewards/margins": 0.07390480488538742, | |
| "rewards/rejected": 0.42732667922973633, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 34.14748336984378, | |
| "learning_rate": 9.94196854912548e-07, | |
| "logits/chosen": -2.6901488304138184, | |
| "logits/rejected": -2.706470251083374, | |
| "logps/chosen": -160.76934814453125, | |
| "logps/rejected": -159.16995239257812, | |
| "loss": 0.6546, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.3948652148246765, | |
| "rewards/margins": 0.21753108501434326, | |
| "rewards/rejected": 0.17733411490917206, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3472222222222222, | |
| "grad_norm": 32.71971381381323, | |
| "learning_rate": 9.869745381355905e-07, | |
| "logits/chosen": -2.6964094638824463, | |
| "logits/rejected": -2.682394504547119, | |
| "logps/chosen": -157.96328735351562, | |
| "logps/rejected": -162.65863037109375, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.2566204071044922, | |
| "rewards/margins": 0.2564144432544708, | |
| "rewards/rejected": 0.00020598471746779978, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 35.675812969071345, | |
| "learning_rate": 9.769221256218162e-07, | |
| "logits/chosen": -2.6471545696258545, | |
| "logits/rejected": -2.620529890060425, | |
| "logps/chosen": -160.15093994140625, | |
| "logps/rejected": -163.9252166748047, | |
| "loss": 0.6267, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.07839958369731903, | |
| "rewards/margins": 0.13437125086784363, | |
| "rewards/rejected": -0.0559716634452343, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4861111111111111, | |
| "grad_norm": 35.65139996099045, | |
| "learning_rate": 9.64098037858483e-07, | |
| "logits/chosen": -2.5851612091064453, | |
| "logits/rejected": -2.5945792198181152, | |
| "logps/chosen": -153.91452026367188, | |
| "logps/rejected": -158.6124267578125, | |
| "loss": 0.6275, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.3417479395866394, | |
| "rewards/margins": 0.34512001276016235, | |
| "rewards/rejected": -0.0033720836509019136, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 32.23578863597255, | |
| "learning_rate": 9.485768031694871e-07, | |
| "logits/chosen": -2.5328378677368164, | |
| "logits/rejected": -2.5566062927246094, | |
| "logps/chosen": -149.2768096923828, | |
| "logps/rejected": -150.4925537109375, | |
| "loss": 0.593, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.4136400818824768, | |
| "rewards/margins": 0.36537498235702515, | |
| "rewards/rejected": 0.048265062272548676, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 32.86956863981884, | |
| "learning_rate": 9.304486245873971e-07, | |
| "logits/chosen": -2.6404240131378174, | |
| "logits/rejected": -2.6186232566833496, | |
| "logps/chosen": -154.857421875, | |
| "logps/rejected": -160.54421997070312, | |
| "loss": 0.5804, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": 0.21952423453330994, | |
| "rewards/margins": 0.6369880437850952, | |
| "rewards/rejected": -0.41746383905410767, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.6944444444444444, | |
| "grad_norm": 32.98463521922708, | |
| "learning_rate": 9.098188556305262e-07, | |
| "logits/chosen": -2.7281055450439453, | |
| "logits/rejected": -2.7109835147857666, | |
| "logps/chosen": -164.81529235839844, | |
| "logps/rejected": -172.00509643554688, | |
| "loss": 0.5904, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.22729039192199707, | |
| "rewards/margins": 0.6152051687240601, | |
| "rewards/rejected": -0.3879148066043854, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6944444444444444, | |
| "eval_logits/chosen": -2.7504920959472656, | |
| "eval_logits/rejected": -2.7399511337280273, | |
| "eval_logps/chosen": -158.1156463623047, | |
| "eval_logps/rejected": -165.36752319335938, | |
| "eval_loss": 0.5584720969200134, | |
| "eval_rewards/accuracies": 0.6953125, | |
| "eval_rewards/chosen": 0.1767115592956543, | |
| "eval_rewards/margins": 0.707548201084137, | |
| "eval_rewards/rejected": -0.5308365821838379, | |
| "eval_runtime": 126.2022, | |
| "eval_samples_per_second": 16.188, | |
| "eval_steps_per_second": 0.254, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7638888888888888, | |
| "grad_norm": 32.243042350659394, | |
| "learning_rate": 8.868073880316123e-07, | |
| "logits/chosen": -2.716021776199341, | |
| "logits/rejected": -2.7205090522766113, | |
| "logps/chosen": -160.37879943847656, | |
| "logps/rejected": -165.23492431640625, | |
| "loss": 0.5532, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.1015612855553627, | |
| "rewards/margins": 0.6121625304222107, | |
| "rewards/rejected": -0.7137238383293152, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 35.25482149048248, | |
| "learning_rate": 8.615479549763755e-07, | |
| "logits/chosen": -2.6721854209899902, | |
| "logits/rejected": -2.6540756225585938, | |
| "logps/chosen": -154.70159912109375, | |
| "logps/rejected": -160.74624633789062, | |
| "loss": 0.5415, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.3344821035861969, | |
| "rewards/margins": 0.9801710247993469, | |
| "rewards/rejected": -1.3146532773971558, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.9027777777777778, | |
| "grad_norm": 30.012069786549304, | |
| "learning_rate": 8.341873539012443e-07, | |
| "logits/chosen": -2.6360578536987305, | |
| "logits/rejected": -2.6559650897979736, | |
| "logps/chosen": -155.54788208007812, | |
| "logps/rejected": -159.24798583984375, | |
| "loss": 0.534, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.024257740005850792, | |
| "rewards/margins": 0.7708773016929626, | |
| "rewards/rejected": -0.7466195821762085, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.9722222222222222, | |
| "grad_norm": 32.593167513928634, | |
| "learning_rate": 8.048845933670271e-07, | |
| "logits/chosen": -2.6497814655303955, | |
| "logits/rejected": -2.643947124481201, | |
| "logps/chosen": -175.2610626220703, | |
| "logps/rejected": -181.9915771484375, | |
| "loss": 0.5238, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.22413644194602966, | |
| "rewards/margins": 0.8313905596733093, | |
| "rewards/rejected": -1.0555269718170166, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0416666666666667, | |
| "grad_norm": 18.23226121550072, | |
| "learning_rate": 7.738099689665539e-07, | |
| "logits/chosen": -2.7001595497131348, | |
| "logits/rejected": -2.7125773429870605, | |
| "logps/chosen": -158.98019409179688, | |
| "logps/rejected": -174.9591522216797, | |
| "loss": 0.3717, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.3584290146827698, | |
| "rewards/margins": 1.3700459003448486, | |
| "rewards/rejected": -1.7284749746322632, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 17.63451132872378, | |
| "learning_rate": 7.41144073636728e-07, | |
| "logits/chosen": -2.7544779777526855, | |
| "logits/rejected": -2.7487056255340576, | |
| "logps/chosen": -172.69039916992188, | |
| "logps/rejected": -199.56082153320312, | |
| "loss": 0.253, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": 0.5693989396095276, | |
| "rewards/margins": 2.6911113262176514, | |
| "rewards/rejected": -2.1217124462127686, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.1805555555555556, | |
| "grad_norm": 19.435839013678418, | |
| "learning_rate": 7.070767481266492e-07, | |
| "logits/chosen": -2.7354109287261963, | |
| "logits/rejected": -2.7295777797698975, | |
| "logps/chosen": -166.5186767578125, | |
| "logps/rejected": -190.08285522460938, | |
| "loss": 0.2397, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 0.3293834328651428, | |
| "rewards/margins": 2.566129207611084, | |
| "rewards/rejected": -2.236745595932007, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 20.070654461149143, | |
| "learning_rate": 6.718059777212565e-07, | |
| "logits/chosen": -2.7108664512634277, | |
| "logits/rejected": -2.699066638946533, | |
| "logps/chosen": -165.696533203125, | |
| "logps/rejected": -187.604736328125, | |
| "loss": 0.2257, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.10144776105880737, | |
| "rewards/margins": 2.6659486293792725, | |
| "rewards/rejected": -2.7673964500427246, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.3194444444444444, | |
| "grad_norm": 21.350071509710315, | |
| "learning_rate": 6.355367416322778e-07, | |
| "logits/chosen": -2.6924808025360107, | |
| "logits/rejected": -2.6694552898406982, | |
| "logps/chosen": -162.50173950195312, | |
| "logps/rejected": -189.10617065429688, | |
| "loss": 0.2252, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -0.26245421171188354, | |
| "rewards/margins": 2.784482955932617, | |
| "rewards/rejected": -3.0469374656677246, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 19.5783540001623, | |
| "learning_rate": 5.984798217433531e-07, | |
| "logits/chosen": -2.669064998626709, | |
| "logits/rejected": -2.683467388153076, | |
| "logps/chosen": -166.97213745117188, | |
| "logps/rejected": -197.8631591796875, | |
| "loss": 0.2124, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.2085113823413849, | |
| "rewards/margins": 2.9906728267669678, | |
| "rewards/rejected": -3.1991844177246094, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "eval_logits/chosen": -2.702752113342285, | |
| "eval_logits/rejected": -2.684614419937134, | |
| "eval_logps/chosen": -169.66738891601562, | |
| "eval_logps/rejected": -182.4733123779297, | |
| "eval_loss": 0.5329886078834534, | |
| "eval_rewards/accuracies": 0.734375, | |
| "eval_rewards/chosen": -0.9784606099128723, | |
| "eval_rewards/margins": 1.2629573345184326, | |
| "eval_rewards/rejected": -2.2414181232452393, | |
| "eval_runtime": 125.9231, | |
| "eval_samples_per_second": 16.224, | |
| "eval_steps_per_second": 0.254, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 216, | |
| "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 | |
| } | |