learning_rate: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 4
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.6262	0.0718	100	0.5156	0.8468	0.8471	0.8468	0.8464
0.2347	0.1436	200	0.1725	0.9446	0.9451	0.9446	0.9445
0.1062	0.2154	300	0.0719	0.9752	0.9755	0.9752	0.9752
0.0731	0.2873	400	0.0485	0.9839	0.9839	0.9839	0.9839
0.049	0.3591	500	0.0479	0.9850	0.9852	0.9850	0.9850
0.0561	0.4309	600	0.0423	0.9847	0.9851	0.9847	0.9847
0.0276	0.5027	700	0.0324	0.9903	0.9903	0.9903	0.9903
0.0593	0.5745	800	0.0329	0.9888	0.9889	0.9888	0.9888
0.0302	0.6463	900	0.0343	0.9888	0.9889	0.9888	0.9888
0.035	0.7181	1000	0.0309	0.9905	0.9906	0.9905	0.9905
0.0326	0.7899	1100	0.0266	0.9924	0.9924	0.9924	0.9924
0.028	0.8618	1200	0.0236	0.9922	0.9922	0.9922	0.9922
0.0209	0.9336	1300	0.0285	0.9915	0.9915	0.9915	0.9915
0.0166	1.0050	1400	0.0248	0.9924	0.9924	0.9924	0.9924

tyaroshko/distilbert-base-uncased-fake-news-detection

fake_news_model