SLM_vs_LLM_experiments
/
max_seq_length_128_experiments
/FacebookAI
/roberta_base_amazon
/trainer_state.json
| { | |
| "best_metric": 0.6847230792045593, | |
| "best_model_checkpoint": "../../experiments_checkpoints/MAdAiLab/FacebookAI/roberta_base_amazon/checkpoint-550", | |
| "epoch": 3.0, | |
| "eval_steps": 50, | |
| "global_step": 570, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 1.8449782133102417, | |
| "learning_rate": 1.9649122807017544e-05, | |
| "loss": 3.1004, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 3.084895133972168, | |
| "learning_rate": 1.929824561403509e-05, | |
| "loss": 3.0595, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 4.537363529205322, | |
| "learning_rate": 1.894736842105263e-05, | |
| "loss": 2.802, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 5.116600036621094, | |
| "learning_rate": 1.8596491228070176e-05, | |
| "loss": 2.3885, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 4.867352485656738, | |
| "learning_rate": 1.824561403508772e-05, | |
| "loss": 2.0556, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_accuracy": 0.616600790513834, | |
| "eval_f1_macro": 0.46316309278720524, | |
| "eval_f1_micro": 0.616600790513834, | |
| "eval_loss": 1.7788082361221313, | |
| "eval_runtime": 1.4418, | |
| "eval_samples_per_second": 1052.82, | |
| "eval_steps_per_second": 16.645, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 6.37776517868042, | |
| "learning_rate": 1.7894736842105264e-05, | |
| "loss": 1.7967, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 8.01982593536377, | |
| "learning_rate": 1.754385964912281e-05, | |
| "loss": 1.5339, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 6.54938268661499, | |
| "learning_rate": 1.719298245614035e-05, | |
| "loss": 1.398, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 5.9745659828186035, | |
| "learning_rate": 1.6842105263157896e-05, | |
| "loss": 1.337, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 7.999369144439697, | |
| "learning_rate": 1.649122807017544e-05, | |
| "loss": 1.2395, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_accuracy": 0.7140974967061924, | |
| "eval_f1_macro": 0.6033304658823342, | |
| "eval_f1_micro": 0.7140974967061924, | |
| "eval_loss": 1.1504093408584595, | |
| "eval_runtime": 1.4443, | |
| "eval_samples_per_second": 1051.003, | |
| "eval_steps_per_second": 16.617, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 5.7676262855529785, | |
| "learning_rate": 1.6140350877192984e-05, | |
| "loss": 1.1179, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 7.363275051116943, | |
| "learning_rate": 1.578947368421053e-05, | |
| "loss": 1.1031, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 9.764667510986328, | |
| "learning_rate": 1.543859649122807e-05, | |
| "loss": 1.1, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 7.389791965484619, | |
| "learning_rate": 1.5087719298245615e-05, | |
| "loss": 1.0124, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 9.022140502929688, | |
| "learning_rate": 1.4736842105263159e-05, | |
| "loss": 0.9724, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "eval_accuracy": 0.738471673254282, | |
| "eval_f1_macro": 0.6449968622620541, | |
| "eval_f1_micro": 0.738471673254282, | |
| "eval_loss": 0.964474618434906, | |
| "eval_runtime": 1.4518, | |
| "eval_samples_per_second": 1045.572, | |
| "eval_steps_per_second": 16.531, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 7.70412540435791, | |
| "learning_rate": 1.4385964912280704e-05, | |
| "loss": 0.9852, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 9.079597473144531, | |
| "learning_rate": 1.4035087719298246e-05, | |
| "loss": 0.9693, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 10.334077835083008, | |
| "learning_rate": 1.3684210526315791e-05, | |
| "loss": 0.9577, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 9.137624740600586, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.9565, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 8.054055213928223, | |
| "learning_rate": 1.2982456140350879e-05, | |
| "loss": 0.8878, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_accuracy": 0.7733860342555995, | |
| "eval_f1_macro": 0.7001366407404722, | |
| "eval_f1_micro": 0.7733860342555995, | |
| "eval_loss": 0.870373010635376, | |
| "eval_runtime": 1.4548, | |
| "eval_samples_per_second": 1043.435, | |
| "eval_steps_per_second": 16.497, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 8.359734535217285, | |
| "learning_rate": 1.263157894736842e-05, | |
| "loss": 0.8157, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 11.497844696044922, | |
| "learning_rate": 1.2280701754385966e-05, | |
| "loss": 0.8643, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "grad_norm": 10.738656997680664, | |
| "learning_rate": 1.192982456140351e-05, | |
| "loss": 0.8478, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 7.716694355010986, | |
| "learning_rate": 1.1578947368421053e-05, | |
| "loss": 0.7587, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 10.482646942138672, | |
| "learning_rate": 1.1228070175438597e-05, | |
| "loss": 0.7261, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "eval_accuracy": 0.7766798418972332, | |
| "eval_f1_macro": 0.7188389846095605, | |
| "eval_f1_micro": 0.7766798418972332, | |
| "eval_loss": 0.815912127494812, | |
| "eval_runtime": 1.4582, | |
| "eval_samples_per_second": 1041.022, | |
| "eval_steps_per_second": 16.459, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 8.698047637939453, | |
| "learning_rate": 1.0877192982456142e-05, | |
| "loss": 0.7495, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 7.04860782623291, | |
| "learning_rate": 1.0526315789473684e-05, | |
| "loss": 0.7562, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 16.800617218017578, | |
| "learning_rate": 1.017543859649123e-05, | |
| "loss": 0.7407, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 11.83535385131836, | |
| "learning_rate": 9.824561403508772e-06, | |
| "loss": 0.7224, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 8.78000545501709, | |
| "learning_rate": 9.473684210526315e-06, | |
| "loss": 0.8083, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_accuracy": 0.7878787878787878, | |
| "eval_f1_macro": 0.7240527586663765, | |
| "eval_f1_micro": 0.7878787878787878, | |
| "eval_loss": 0.7717685699462891, | |
| "eval_runtime": 1.4583, | |
| "eval_samples_per_second": 1040.974, | |
| "eval_steps_per_second": 16.458, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 8.961960792541504, | |
| "learning_rate": 9.12280701754386e-06, | |
| "loss": 0.7576, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "grad_norm": 9.152416229248047, | |
| "learning_rate": 8.771929824561405e-06, | |
| "loss": 0.7021, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "grad_norm": 12.08969783782959, | |
| "learning_rate": 8.421052631578948e-06, | |
| "loss": 0.7218, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "grad_norm": 11.890509605407715, | |
| "learning_rate": 8.070175438596492e-06, | |
| "loss": 0.806, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 13.19563102722168, | |
| "learning_rate": 7.719298245614036e-06, | |
| "loss": 0.7209, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "eval_accuracy": 0.7997364953886693, | |
| "eval_f1_macro": 0.7414355218281033, | |
| "eval_f1_micro": 0.7997364953886693, | |
| "eval_loss": 0.7306675910949707, | |
| "eval_runtime": 1.4597, | |
| "eval_samples_per_second": 1039.954, | |
| "eval_steps_per_second": 16.442, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 10.341647148132324, | |
| "learning_rate": 7.368421052631579e-06, | |
| "loss": 0.6538, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "grad_norm": 10.095640182495117, | |
| "learning_rate": 7.017543859649123e-06, | |
| "loss": 0.7268, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 14.569330215454102, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.7441, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 8.13227653503418, | |
| "learning_rate": 6.31578947368421e-06, | |
| "loss": 0.6634, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "grad_norm": 8.250567436218262, | |
| "learning_rate": 5.964912280701755e-06, | |
| "loss": 0.6535, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "eval_accuracy": 0.8043478260869565, | |
| "eval_f1_macro": 0.7452177798047408, | |
| "eval_f1_micro": 0.8043478260869565, | |
| "eval_loss": 0.7204875349998474, | |
| "eval_runtime": 1.4623, | |
| "eval_samples_per_second": 1038.057, | |
| "eval_steps_per_second": 16.412, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 10.287467002868652, | |
| "learning_rate": 5.6140350877192985e-06, | |
| "loss": 0.569, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "grad_norm": 8.486634254455566, | |
| "learning_rate": 5.263157894736842e-06, | |
| "loss": 0.6313, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 8.879149436950684, | |
| "learning_rate": 4.912280701754386e-06, | |
| "loss": 0.5844, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 10.877876281738281, | |
| "learning_rate": 4.56140350877193e-06, | |
| "loss": 0.64, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "grad_norm": 7.0523834228515625, | |
| "learning_rate": 4.210526315789474e-06, | |
| "loss": 0.6283, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "eval_accuracy": 0.8089591567852438, | |
| "eval_f1_macro": 0.7498355070996475, | |
| "eval_f1_micro": 0.8089591567852438, | |
| "eval_loss": 0.7047255635261536, | |
| "eval_runtime": 1.4632, | |
| "eval_samples_per_second": 1037.454, | |
| "eval_steps_per_second": 16.402, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 8.621397972106934, | |
| "learning_rate": 3.859649122807018e-06, | |
| "loss": 0.5349, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "grad_norm": 7.53057861328125, | |
| "learning_rate": 3.5087719298245615e-06, | |
| "loss": 0.6345, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 12.083460807800293, | |
| "learning_rate": 3.157894736842105e-06, | |
| "loss": 0.6337, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 10.678044319152832, | |
| "learning_rate": 2.8070175438596493e-06, | |
| "loss": 0.5657, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "grad_norm": 8.628580093383789, | |
| "learning_rate": 2.456140350877193e-06, | |
| "loss": 0.5214, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_accuracy": 0.810935441370224, | |
| "eval_f1_macro": 0.7541066085678262, | |
| "eval_f1_micro": 0.810935441370224, | |
| "eval_loss": 0.687921404838562, | |
| "eval_runtime": 1.4612, | |
| "eval_samples_per_second": 1038.859, | |
| "eval_steps_per_second": 16.425, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 11.350371360778809, | |
| "learning_rate": 2.105263157894737e-06, | |
| "loss": 0.5632, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 10.923992156982422, | |
| "learning_rate": 1.7543859649122807e-06, | |
| "loss": 0.602, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "grad_norm": 11.192131996154785, | |
| "learning_rate": 1.4035087719298246e-06, | |
| "loss": 0.6038, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 9.871031761169434, | |
| "learning_rate": 1.0526315789473685e-06, | |
| "loss": 0.5776, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "grad_norm": 9.84094524383545, | |
| "learning_rate": 7.017543859649123e-07, | |
| "loss": 0.5808, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "eval_accuracy": 0.80566534914361, | |
| "eval_f1_macro": 0.7451626163175262, | |
| "eval_f1_micro": 0.80566534914361, | |
| "eval_loss": 0.6847230792045593, | |
| "eval_runtime": 1.4607, | |
| "eval_samples_per_second": 1039.224, | |
| "eval_steps_per_second": 16.43, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "grad_norm": 8.289957046508789, | |
| "learning_rate": 3.5087719298245616e-07, | |
| "loss": 0.6075, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 13.524730682373047, | |
| "learning_rate": 0.0, | |
| "loss": 0.596, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 570, | |
| "total_flos": 2400025338445824.0, | |
| "train_loss": 0.9787228534096166, | |
| "train_runtime": 143.678, | |
| "train_samples_per_second": 253.567, | |
| "train_steps_per_second": 3.967 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 570, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 50, | |
| "total_flos": 2400025338445824.0, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |