| { | |
| "best_metric": 0.9522393282015396, | |
| "best_model_checkpoint": "../../checkpoint/ddi/pubmedbert-abstract/checkpoint-13447", | |
| "epoch": 20.0, | |
| "eval_steps": 500, | |
| "global_step": 15820, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.938593421973408, | |
| "eval_loss": 0.23401664197444916, | |
| "eval_runtime": 9.3525, | |
| "eval_samples_per_second": 611.174, | |
| "eval_steps_per_second": 2.459, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 22.48932647705078, | |
| "learning_rate": 4.683944374209861e-05, | |
| "loss": 0.1776, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9419174247725682, | |
| "eval_loss": 0.27159374952316284, | |
| "eval_runtime": 9.4015, | |
| "eval_samples_per_second": 607.985, | |
| "eval_steps_per_second": 2.446, | |
| "step": 1582 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 0.30297374725341797, | |
| "learning_rate": 4.367888748419722e-05, | |
| "loss": 0.0855, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9431420573827851, | |
| "eval_loss": 0.27298399806022644, | |
| "eval_runtime": 9.4208, | |
| "eval_samples_per_second": 606.741, | |
| "eval_steps_per_second": 2.441, | |
| "step": 2373 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "grad_norm": 0.08143208175897598, | |
| "learning_rate": 4.051833122629583e-05, | |
| "loss": 0.0627, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9382435269419174, | |
| "eval_loss": 0.33226683735847473, | |
| "eval_runtime": 9.4056, | |
| "eval_samples_per_second": 607.725, | |
| "eval_steps_per_second": 2.445, | |
| "step": 3164 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9450664800559833, | |
| "eval_loss": 0.3307989239692688, | |
| "eval_runtime": 9.3945, | |
| "eval_samples_per_second": 608.444, | |
| "eval_steps_per_second": 2.448, | |
| "step": 3955 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "grad_norm": 0.012913365848362446, | |
| "learning_rate": 3.735777496839444e-05, | |
| "loss": 0.0463, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9412176347095871, | |
| "eval_loss": 0.39864641427993774, | |
| "eval_runtime": 9.3742, | |
| "eval_samples_per_second": 609.757, | |
| "eval_steps_per_second": 2.454, | |
| "step": 4746 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "grad_norm": 2.0644302368164062, | |
| "learning_rate": 3.419721871049305e-05, | |
| "loss": 0.0308, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.9419174247725682, | |
| "eval_loss": 0.42111822962760925, | |
| "eval_runtime": 9.3868, | |
| "eval_samples_per_second": 608.939, | |
| "eval_steps_per_second": 2.45, | |
| "step": 5537 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "grad_norm": 30.287784576416016, | |
| "learning_rate": 3.1036662452591655e-05, | |
| "loss": 0.0312, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.943666899930021, | |
| "eval_loss": 0.36164581775665283, | |
| "eval_runtime": 9.3894, | |
| "eval_samples_per_second": 608.769, | |
| "eval_steps_per_second": 2.45, | |
| "step": 6328 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "grad_norm": 0.03705460578203201, | |
| "learning_rate": 2.7876106194690264e-05, | |
| "loss": 0.0221, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.9396431070678797, | |
| "eval_loss": 0.4309725761413574, | |
| "eval_runtime": 9.4203, | |
| "eval_samples_per_second": 606.777, | |
| "eval_steps_per_second": 2.442, | |
| "step": 7119 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.9438418474457663, | |
| "eval_loss": 0.4221705198287964, | |
| "eval_runtime": 9.4325, | |
| "eval_samples_per_second": 605.989, | |
| "eval_steps_per_second": 2.438, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "grad_norm": 0.11050642281770706, | |
| "learning_rate": 2.4715549936788876e-05, | |
| "loss": 0.0181, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.9445416375087474, | |
| "eval_loss": 0.4184616804122925, | |
| "eval_runtime": 9.4176, | |
| "eval_samples_per_second": 606.948, | |
| "eval_steps_per_second": 2.442, | |
| "step": 8701 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "grad_norm": 0.03253033012151718, | |
| "learning_rate": 2.1554993678887485e-05, | |
| "loss": 0.0141, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.945591322603219, | |
| "eval_loss": 0.46782681345939636, | |
| "eval_runtime": 9.4133, | |
| "eval_samples_per_second": 607.226, | |
| "eval_steps_per_second": 2.443, | |
| "step": 9492 | |
| }, | |
| { | |
| "epoch": 12.64, | |
| "grad_norm": 0.005740019958466291, | |
| "learning_rate": 1.8394437420986094e-05, | |
| "loss": 0.0133, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.9503149055283415, | |
| "eval_loss": 0.40269356966018677, | |
| "eval_runtime": 9.4053, | |
| "eval_samples_per_second": 607.74, | |
| "eval_steps_per_second": 2.445, | |
| "step": 10283 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "grad_norm": 0.0024239453487098217, | |
| "learning_rate": 1.5233881163084704e-05, | |
| "loss": 0.0082, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.9473407977606718, | |
| "eval_loss": 0.45041143894195557, | |
| "eval_runtime": 9.639, | |
| "eval_samples_per_second": 593.006, | |
| "eval_steps_per_second": 2.386, | |
| "step": 11074 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.9504898530440867, | |
| "eval_loss": 0.47598323225975037, | |
| "eval_runtime": 9.4326, | |
| "eval_samples_per_second": 605.982, | |
| "eval_steps_per_second": 2.438, | |
| "step": 11865 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "grad_norm": 0.0034840325824916363, | |
| "learning_rate": 1.2073324905183313e-05, | |
| "loss": 0.0052, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.9448915325402379, | |
| "eval_loss": 0.45725104212760925, | |
| "eval_runtime": 9.4366, | |
| "eval_samples_per_second": 605.728, | |
| "eval_steps_per_second": 2.437, | |
| "step": 12656 | |
| }, | |
| { | |
| "epoch": 16.43, | |
| "grad_norm": 0.4831530451774597, | |
| "learning_rate": 8.912768647281922e-06, | |
| "loss": 0.0042, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.9522393282015396, | |
| "eval_loss": 0.43564239144325256, | |
| "eval_runtime": 9.4029, | |
| "eval_samples_per_second": 607.9, | |
| "eval_steps_per_second": 2.446, | |
| "step": 13447 | |
| }, | |
| { | |
| "epoch": 17.7, | |
| "grad_norm": 0.0005753316800110042, | |
| "learning_rate": 5.752212389380531e-06, | |
| "loss": 0.0037, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.948740377886634, | |
| "eval_loss": 0.45767056941986084, | |
| "eval_runtime": 9.3872, | |
| "eval_samples_per_second": 608.913, | |
| "eval_steps_per_second": 2.45, | |
| "step": 14238 | |
| }, | |
| { | |
| "epoch": 18.96, | |
| "grad_norm": 0.0019946701359003782, | |
| "learning_rate": 2.59165613147914e-06, | |
| "loss": 0.0024, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.9492652204338698, | |
| "eval_loss": 0.4641525149345398, | |
| "eval_runtime": 9.4075, | |
| "eval_samples_per_second": 607.599, | |
| "eval_steps_per_second": 2.445, | |
| "step": 15029 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.9501399580125962, | |
| "eval_loss": 0.46178367733955383, | |
| "eval_runtime": 9.3958, | |
| "eval_samples_per_second": 608.356, | |
| "eval_steps_per_second": 2.448, | |
| "step": 15820 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 15820, | |
| "total_flos": 3.327918265884672e+16, | |
| "train_loss": 0.03330213655864846, | |
| "train_runtime": 3337.005, | |
| "train_samples_per_second": 151.609, | |
| "train_steps_per_second": 4.741 | |
| } | |
| ], | |
| "logging_steps": 1000, | |
| "max_steps": 15820, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 3.327918265884672e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |