|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 14.0, |
|
"eval_steps": 500, |
|
"global_step": 4781, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.29282576866764276, |
|
"grad_norm": 0.2411503791809082, |
|
"learning_rate": 0.0001, |
|
"loss": 2.0009, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.5856515373352855, |
|
"grad_norm": 0.22286373376846313, |
|
"learning_rate": 0.0001, |
|
"loss": 1.9538, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.8784773060029283, |
|
"grad_norm": 0.24079065024852753, |
|
"learning_rate": 0.0001, |
|
"loss": 1.9571, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.9985358711566618, |
|
"eval_accuracy": 0.45379487179487177, |
|
"eval_loss": 3.9512319564819336, |
|
"eval_runtime": 5.0647, |
|
"eval_samples_per_second": 98.722, |
|
"eval_steps_per_second": 12.439, |
|
"step": 341 |
|
}, |
|
{ |
|
"epoch": 0.9985358711566618, |
|
"eval_bleu": 0.28767194245703276, |
|
"eval_exact_match": 0.03, |
|
"eval_prefix_exact_match": 0.042, |
|
"step": 341 |
|
}, |
|
{ |
|
"epoch": 1.171303074670571, |
|
"grad_norm": 0.33546799421310425, |
|
"learning_rate": 0.0001, |
|
"loss": 1.9255, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 1.4641288433382138, |
|
"grad_norm": 0.3515666723251343, |
|
"learning_rate": 0.0001, |
|
"loss": 1.8684, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.7569546120058566, |
|
"grad_norm": 0.35733896493911743, |
|
"learning_rate": 0.0001, |
|
"loss": 1.8819, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.4482564102564103, |
|
"eval_loss": 4.112804412841797, |
|
"eval_runtime": 5.0848, |
|
"eval_samples_per_second": 98.333, |
|
"eval_steps_per_second": 12.39, |
|
"step": 683 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_bleu": 0.24521038067633488, |
|
"eval_exact_match": 0.02, |
|
"eval_prefix_exact_match": 0.044, |
|
"step": 683 |
|
}, |
|
{ |
|
"epoch": 2.049780380673499, |
|
"grad_norm": 0.3784458339214325, |
|
"learning_rate": 0.0001, |
|
"loss": 1.8613, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 2.342606149341142, |
|
"grad_norm": 0.46702849864959717, |
|
"learning_rate": 0.0001, |
|
"loss": 1.743, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 2.6354319180087846, |
|
"grad_norm": 0.5156407356262207, |
|
"learning_rate": 0.0001, |
|
"loss": 1.7653, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 2.9282576866764276, |
|
"grad_norm": 0.4755038619041443, |
|
"learning_rate": 0.0001, |
|
"loss": 1.7702, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.998535871156662, |
|
"eval_accuracy": 0.446051282051282, |
|
"eval_loss": 4.327652454376221, |
|
"eval_runtime": 5.1641, |
|
"eval_samples_per_second": 96.822, |
|
"eval_steps_per_second": 12.2, |
|
"step": 1024 |
|
}, |
|
{ |
|
"epoch": 2.998535871156662, |
|
"eval_bleu": 0.2598440457703149, |
|
"eval_exact_match": 0.018, |
|
"eval_prefix_exact_match": 0.034, |
|
"step": 1024 |
|
}, |
|
{ |
|
"epoch": 3.22108345534407, |
|
"grad_norm": 0.5288150906562805, |
|
"learning_rate": 0.0001, |
|
"loss": 1.6325, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 3.513909224011713, |
|
"grad_norm": 0.5784010887145996, |
|
"learning_rate": 0.0001, |
|
"loss": 1.5896, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 3.8067349926793557, |
|
"grad_norm": 0.6458429098129272, |
|
"learning_rate": 0.0001, |
|
"loss": 1.6163, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_accuracy": 0.44235897435897437, |
|
"eval_loss": 4.584905624389648, |
|
"eval_runtime": 4.9243, |
|
"eval_samples_per_second": 101.538, |
|
"eval_steps_per_second": 12.794, |
|
"step": 1366 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_bleu": 0.27967222841565076, |
|
"eval_exact_match": 0.01, |
|
"eval_prefix_exact_match": 0.038, |
|
"step": 1366 |
|
}, |
|
{ |
|
"epoch": 4.099560761346998, |
|
"grad_norm": 0.732841432094574, |
|
"learning_rate": 0.0001, |
|
"loss": 1.5441, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 4.392386530014641, |
|
"grad_norm": 0.7372816205024719, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4142, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 4.685212298682284, |
|
"grad_norm": 0.8235049247741699, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4198, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 4.978038067349927, |
|
"grad_norm": 0.7765743732452393, |
|
"learning_rate": 0.0001, |
|
"loss": 1.4427, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 4.998535871156662, |
|
"eval_accuracy": 0.43861538461538463, |
|
"eval_loss": 4.850330829620361, |
|
"eval_runtime": 4.9395, |
|
"eval_samples_per_second": 101.224, |
|
"eval_steps_per_second": 12.754, |
|
"step": 1707 |
|
}, |
|
{ |
|
"epoch": 4.998535871156662, |
|
"eval_bleu": 0.2170008640509292, |
|
"eval_exact_match": 0.006, |
|
"eval_prefix_exact_match": 0.032, |
|
"step": 1707 |
|
}, |
|
{ |
|
"epoch": 5.270863836017569, |
|
"grad_norm": 0.9119400978088379, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2469, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 5.563689604685212, |
|
"grad_norm": 0.8966185450553894, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2294, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 5.856515373352855, |
|
"grad_norm": 0.9629669189453125, |
|
"learning_rate": 0.0001, |
|
"loss": 1.2498, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_accuracy": 0.4348717948717949, |
|
"eval_loss": 5.09260892868042, |
|
"eval_runtime": 4.9572, |
|
"eval_samples_per_second": 100.863, |
|
"eval_steps_per_second": 12.709, |
|
"step": 2049 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_bleu": 0.20196479624851718, |
|
"eval_exact_match": 0.002, |
|
"eval_prefix_exact_match": 0.03, |
|
"step": 2049 |
|
}, |
|
{ |
|
"epoch": 6.149341142020498, |
|
"grad_norm": 1.0808535814285278, |
|
"learning_rate": 0.0001, |
|
"loss": 1.1238, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 6.44216691068814, |
|
"grad_norm": 1.1639196872711182, |
|
"learning_rate": 0.0001, |
|
"loss": 1.0301, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 6.734992679355783, |
|
"grad_norm": 1.0662291049957275, |
|
"learning_rate": 0.0001, |
|
"loss": 1.0655, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 6.998535871156662, |
|
"eval_accuracy": 0.4326153846153846, |
|
"eval_loss": 5.270845890045166, |
|
"eval_runtime": 5.089, |
|
"eval_samples_per_second": 98.251, |
|
"eval_steps_per_second": 12.38, |
|
"step": 2390 |
|
}, |
|
{ |
|
"epoch": 6.998535871156662, |
|
"eval_bleu": 0.17780932376070374, |
|
"eval_exact_match": 0.004, |
|
"eval_prefix_exact_match": 0.024, |
|
"step": 2390 |
|
}, |
|
{ |
|
"epoch": 7.027818448023426, |
|
"grad_norm": 1.6462695598602295, |
|
"learning_rate": 0.0001, |
|
"loss": 1.0358, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 7.320644216691068, |
|
"grad_norm": 1.2542779445648193, |
|
"learning_rate": 0.0001, |
|
"loss": 0.8413, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 7.613469985358711, |
|
"grad_norm": 1.334762454032898, |
|
"learning_rate": 0.0001, |
|
"loss": 0.8654, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 7.906295754026354, |
|
"grad_norm": 1.2401787042617798, |
|
"learning_rate": 0.0001, |
|
"loss": 0.8733, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"eval_accuracy": 0.4316923076923077, |
|
"eval_loss": 5.402440547943115, |
|
"eval_runtime": 4.9383, |
|
"eval_samples_per_second": 101.25, |
|
"eval_steps_per_second": 12.758, |
|
"step": 2732 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"eval_bleu": 0.21418356891762502, |
|
"eval_exact_match": 0.002, |
|
"eval_prefix_exact_match": 0.026, |
|
"step": 2732 |
|
}, |
|
{ |
|
"epoch": 8.199121522693996, |
|
"grad_norm": 1.255996584892273, |
|
"learning_rate": 0.0001, |
|
"loss": 0.7302, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 8.49194729136164, |
|
"grad_norm": 1.2836525440216064, |
|
"learning_rate": 0.0001, |
|
"loss": 0.6875, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 8.784773060029282, |
|
"grad_norm": 1.375715732574463, |
|
"learning_rate": 0.0001, |
|
"loss": 0.7219, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 8.998535871156662, |
|
"eval_accuracy": 0.42938461538461536, |
|
"eval_loss": 5.534832000732422, |
|
"eval_runtime": 5.2363, |
|
"eval_samples_per_second": 95.487, |
|
"eval_steps_per_second": 12.031, |
|
"step": 3073 |
|
}, |
|
{ |
|
"epoch": 8.998535871156662, |
|
"eval_bleu": 0.20459866130314297, |
|
"eval_exact_match": 0.0, |
|
"eval_prefix_exact_match": 0.024, |
|
"step": 3073 |
|
}, |
|
{ |
|
"epoch": 9.077598828696924, |
|
"grad_norm": 1.481998324394226, |
|
"learning_rate": 0.0001, |
|
"loss": 0.6722, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 9.370424597364568, |
|
"grad_norm": 1.4086748361587524, |
|
"learning_rate": 0.0001, |
|
"loss": 0.5489, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 9.66325036603221, |
|
"grad_norm": 1.4381046295166016, |
|
"learning_rate": 0.0001, |
|
"loss": 0.565, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 9.956076134699854, |
|
"grad_norm": 1.7364269495010376, |
|
"learning_rate": 0.0001, |
|
"loss": 0.5932, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"eval_accuracy": 0.42605128205128207, |
|
"eval_loss": 5.768980503082275, |
|
"eval_runtime": 5.0821, |
|
"eval_samples_per_second": 98.384, |
|
"eval_steps_per_second": 12.396, |
|
"step": 3415 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"eval_bleu": 0.19515580420261575, |
|
"eval_exact_match": 0.002, |
|
"eval_prefix_exact_match": 0.026, |
|
"step": 3415 |
|
}, |
|
{ |
|
"epoch": 10.248901903367496, |
|
"grad_norm": 1.599753737449646, |
|
"learning_rate": 0.0001, |
|
"loss": 0.4581, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 10.541727672035138, |
|
"grad_norm": 1.5595277547836304, |
|
"learning_rate": 0.0001, |
|
"loss": 0.4519, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 10.834553440702782, |
|
"grad_norm": 1.6201436519622803, |
|
"learning_rate": 0.0001, |
|
"loss": 0.4719, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 10.998535871156662, |
|
"eval_accuracy": 0.42538461538461536, |
|
"eval_loss": 5.89434289932251, |
|
"eval_runtime": 5.0717, |
|
"eval_samples_per_second": 98.586, |
|
"eval_steps_per_second": 12.422, |
|
"step": 3756 |
|
}, |
|
{ |
|
"epoch": 10.998535871156662, |
|
"eval_bleu": 0.2074466521823303, |
|
"eval_exact_match": 0.0, |
|
"eval_prefix_exact_match": 0.02, |
|
"step": 3756 |
|
}, |
|
{ |
|
"epoch": 11.127379209370424, |
|
"grad_norm": 1.6209241151809692, |
|
"learning_rate": 0.0001, |
|
"loss": 0.4263, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 11.420204978038067, |
|
"grad_norm": 1.5148394107818604, |
|
"learning_rate": 0.0001, |
|
"loss": 0.3625, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 11.71303074670571, |
|
"grad_norm": 1.6862399578094482, |
|
"learning_rate": 0.0001, |
|
"loss": 0.3838, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 12.0, |
|
"eval_accuracy": 0.4247179487179487, |
|
"eval_loss": 6.019149303436279, |
|
"eval_runtime": 5.3028, |
|
"eval_samples_per_second": 94.291, |
|
"eval_steps_per_second": 11.881, |
|
"step": 4098 |
|
}, |
|
{ |
|
"epoch": 12.0, |
|
"eval_bleu": 0.16189430124724263, |
|
"eval_exact_match": 0.002, |
|
"eval_prefix_exact_match": 0.028, |
|
"step": 4098 |
|
}, |
|
{ |
|
"epoch": 12.005856515373353, |
|
"grad_norm": 1.3296407461166382, |
|
"learning_rate": 0.0001, |
|
"loss": 0.3952, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 12.298682284040996, |
|
"grad_norm": 1.3439353704452515, |
|
"learning_rate": 0.0001, |
|
"loss": 0.2901, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 12.591508052708638, |
|
"grad_norm": 1.5329118967056274, |
|
"learning_rate": 0.0001, |
|
"loss": 0.3114, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 12.88433382137628, |
|
"grad_norm": 1.649775743484497, |
|
"learning_rate": 0.0001, |
|
"loss": 0.329, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 12.998535871156662, |
|
"eval_accuracy": 0.42456410256410254, |
|
"eval_loss": 6.104373455047607, |
|
"eval_runtime": 5.9944, |
|
"eval_samples_per_second": 83.411, |
|
"eval_steps_per_second": 10.51, |
|
"step": 4439 |
|
}, |
|
{ |
|
"epoch": 12.998535871156662, |
|
"eval_bleu": 0.17170793542335236, |
|
"eval_exact_match": 0.002, |
|
"eval_prefix_exact_match": 0.028, |
|
"step": 4439 |
|
}, |
|
{ |
|
"epoch": 13.177159590043924, |
|
"grad_norm": 1.5262281894683838, |
|
"learning_rate": 0.0001, |
|
"loss": 0.2758, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 13.469985358711567, |
|
"grad_norm": 1.5962131023406982, |
|
"learning_rate": 0.0001, |
|
"loss": 0.2594, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 13.762811127379209, |
|
"grad_norm": 1.4258382320404053, |
|
"learning_rate": 0.0001, |
|
"loss": 0.2742, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 14.0, |
|
"eval_accuracy": 0.4216410256410256, |
|
"eval_loss": 6.146524906158447, |
|
"eval_runtime": 4.9576, |
|
"eval_samples_per_second": 100.855, |
|
"eval_steps_per_second": 12.708, |
|
"step": 4781 |
|
}, |
|
{ |
|
"epoch": 14.0, |
|
"eval_bleu": 0.20004000747911055, |
|
"eval_exact_match": 0.002, |
|
"eval_prefix_exact_match": 0.026, |
|
"step": 4781 |
|
} |
|
], |
|
"logging_steps": 100, |
|
"max_steps": 17050, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 50, |
|
"save_steps": 500, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": false |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 4.1055093812258406e+17, |
|
"train_batch_size": 1, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|