bobox's picture
Training in progress, step 2, checkpoint
89bc585 verified
raw
history blame
7.07 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.05813953488372093,
"eval_steps": 2,
"global_step": 2,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.029069767441860465,
"grad_norm": 13.73514461517334,
"learning_rate": 3.3333333333333333e-06,
"loss": 6.7536,
"step": 1
},
{
"epoch": 0.05813953488372093,
"grad_norm": 16.892961502075195,
"learning_rate": 6.666666666666667e-06,
"loss": 6.6203,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_NLI-v2_cosine_accuracy": 1.0,
"eval_NLI-v2_dot_accuracy": 0.109375,
"eval_NLI-v2_euclidean_accuracy": 1.0,
"eval_NLI-v2_manhattan_accuracy": 1.0,
"eval_NLI-v2_max_accuracy": 1.0,
"eval_VitaminC_cosine_accuracy": 0.55078125,
"eval_VitaminC_cosine_accuracy_threshold": 0.9469717741012573,
"eval_VitaminC_cosine_ap": 0.5165514227338435,
"eval_VitaminC_cosine_f1": 0.6525198938992042,
"eval_VitaminC_cosine_f1_threshold": 0.4987494945526123,
"eval_VitaminC_cosine_precision": 0.484251968503937,
"eval_VitaminC_cosine_recall": 1.0,
"eval_VitaminC_dot_accuracy": 0.55078125,
"eval_VitaminC_dot_accuracy_threshold": 417.45166015625,
"eval_VitaminC_dot_ap": 0.5125282416460941,
"eval_VitaminC_dot_f1": 0.6525198938992042,
"eval_VitaminC_dot_f1_threshold": 200.84530639648438,
"eval_VitaminC_dot_precision": 0.484251968503937,
"eval_VitaminC_dot_recall": 1.0,
"eval_VitaminC_euclidean_accuracy": 0.546875,
"eval_VitaminC_euclidean_accuracy_threshold": 6.841136932373047,
"eval_VitaminC_euclidean_ap": 0.512994913068431,
"eval_VitaminC_euclidean_f1": 0.6525198938992042,
"eval_VitaminC_euclidean_f1_threshold": 20.22179412841797,
"eval_VitaminC_euclidean_precision": 0.484251968503937,
"eval_VitaminC_euclidean_recall": 1.0,
"eval_VitaminC_manhattan_accuracy": 0.546875,
"eval_VitaminC_manhattan_accuracy_threshold": 117.1143569946289,
"eval_VitaminC_manhattan_ap": 0.5177961641566705,
"eval_VitaminC_manhattan_f1": 0.6542553191489362,
"eval_VitaminC_manhattan_f1_threshold": 292.3347473144531,
"eval_VitaminC_manhattan_precision": 0.48616600790513836,
"eval_VitaminC_manhattan_recall": 1.0,
"eval_VitaminC_max_accuracy": 0.55078125,
"eval_VitaminC_max_accuracy_threshold": 417.45166015625,
"eval_VitaminC_max_ap": 0.5177961641566705,
"eval_VitaminC_max_f1": 0.6542553191489362,
"eval_VitaminC_max_f1_threshold": 292.3347473144531,
"eval_VitaminC_max_precision": 0.48616600790513836,
"eval_VitaminC_max_recall": 1.0,
"eval_sequential_score": 0.5177961641566705,
"eval_sts-test_pearson_cosine": 0.017178505918243114,
"eval_sts-test_pearson_dot": 0.16228524441844774,
"eval_sts-test_pearson_euclidean": 0.024391561236282323,
"eval_sts-test_pearson_manhattan": 0.059272519732015624,
"eval_sts-test_pearson_max": 0.16228524441844774,
"eval_sts-test_spearman_cosine": 0.07115563415775981,
"eval_sts-test_spearman_dot": 0.1967074668301984,
"eval_sts-test_spearman_euclidean": 0.05117306486959643,
"eval_sts-test_spearman_manhattan": 0.07641354546391785,
"eval_sts-test_spearman_max": 0.1967074668301984,
"eval_vitaminc-pairs_loss": 2.72658371925354,
"eval_vitaminc-pairs_runtime": 1.4719,
"eval_vitaminc-pairs_samples_per_second": 73.373,
"eval_vitaminc-pairs_steps_per_second": 1.359,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_negation-triplets_loss": 5.094177722930908,
"eval_negation-triplets_runtime": 0.3027,
"eval_negation-triplets_samples_per_second": 211.46,
"eval_negation-triplets_steps_per_second": 3.304,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_scitail-pairs-pos_loss": 1.9183871746063232,
"eval_scitail-pairs-pos_runtime": 0.3785,
"eval_scitail-pairs-pos_samples_per_second": 142.654,
"eval_scitail-pairs-pos_steps_per_second": 2.642,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_xsum-pairs_loss": 6.3277668952941895,
"eval_xsum-pairs_runtime": 3.4205,
"eval_xsum-pairs_samples_per_second": 37.422,
"eval_xsum-pairs_steps_per_second": 0.585,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_sciq_pairs_loss": 0.36539140343666077,
"eval_sciq_pairs_runtime": 3.4462,
"eval_sciq_pairs_samples_per_second": 37.143,
"eval_sciq_pairs_steps_per_second": 0.58,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_qasc_pairs_loss": 3.9689104557037354,
"eval_qasc_pairs_runtime": 0.6477,
"eval_qasc_pairs_samples_per_second": 197.636,
"eval_qasc_pairs_steps_per_second": 3.088,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_openbookqa_pairs_loss": 4.7439374923706055,
"eval_openbookqa_pairs_runtime": 0.5759,
"eval_openbookqa_pairs_samples_per_second": 222.248,
"eval_openbookqa_pairs_steps_per_second": 3.473,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_msmarco_pairs_loss": 10.51363754272461,
"eval_msmarco_pairs_runtime": 1.3036,
"eval_msmarco_pairs_samples_per_second": 98.189,
"eval_msmarco_pairs_steps_per_second": 1.534,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_nq_pairs_loss": 4.969011306762695,
"eval_nq_pairs_runtime": 2.5922,
"eval_nq_pairs_samples_per_second": 49.378,
"eval_nq_pairs_steps_per_second": 0.772,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_trivia_pairs_loss": 3.8609507083892822,
"eval_trivia_pairs_runtime": 4.3666,
"eval_trivia_pairs_samples_per_second": 29.314,
"eval_trivia_pairs_steps_per_second": 0.458,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_gooaq_pairs_loss": 8.041110038757324,
"eval_gooaq_pairs_runtime": 0.9119,
"eval_gooaq_pairs_samples_per_second": 140.361,
"eval_gooaq_pairs_steps_per_second": 2.193,
"step": 2
},
{
"epoch": 0.05813953488372093,
"eval_paws-pos_loss": 2.21897029876709,
"eval_paws-pos_runtime": 0.6905,
"eval_paws-pos_samples_per_second": 185.364,
"eval_paws-pos_steps_per_second": 2.896,
"step": 2
}
],
"logging_steps": 1,
"max_steps": 34,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 2,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 224,
"trial_name": null,
"trial_params": null
}