SLM_vs_LLM_experiments
/
max_seq_length_128_experiments
/google_bert
/bert_base_uncased_patent
/all_results.json
{ | |
"epoch": 3.0, | |
"eval_accuracy": 0.6778, | |
"eval_f1_macro": 0.6087469050367866, | |
"eval_f1_micro": 0.6778, | |
"eval_loss": 0.9466402530670166, | |
"eval_runtime": 4.7626, | |
"eval_samples": 5000, | |
"eval_samples_per_second": 1049.842, | |
"eval_steps_per_second": 16.588, | |
"test_accuracy": 0.6684, | |
"test_f1_macro": 0.6172965996785956, | |
"test_f1_micro": 0.6684, | |
"test_loss": 0.9713044762611389, | |
"test_runtime": 4.7688, | |
"test_samples_per_second": 1048.472, | |
"test_steps_per_second": 16.566, | |
"train_loss": 0.9952963392447938, | |
"train_runtime": 362.5909, | |
"train_samples": 25000, | |
"train_samples_per_second": 206.845, | |
"train_steps_per_second": 3.235 | |
} |