bert-base-uncased-ag-news-finetuned-dwnews-categories

This model is a fine-tuned version of nateraw/bert-base-uncased-ag-news on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8659
• Accuracy: 0.74
• Precision Weighted: 0.7435
• Precision Macro: 0.7557
• Recall Weighted: 0.74
• Recall Macro: 0.7304
• F1 Weighted: 0.7294
• F1 Macro: 0.7250

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Weighted	Precision Macro	Recall Weighted	Recall Macro	F1 Weighted	F1 Macro
1.8428	0.25	50	1.5099	0.53	0.5008	0.4655	0.53	0.3544	0.4363	0.3246
1.3134	0.5	100	1.2200	0.605	0.4765	0.4247	0.605	0.4681	0.5207	0.4261
1.1562	0.75	150	1.0473	0.68	0.7007	0.6603	0.68	0.5534	0.6483	0.5507
1.0008	1.0	200	0.9491	0.67	0.6281	0.5935	0.67	0.5881	0.6390	0.5778
0.8173	1.25	250	0.9218	0.7	0.7028	0.6599	0.7	0.6694	0.6874	0.6417
0.8385	1.5	300	0.8900	0.715	0.7250	0.7131	0.715	0.6600	0.7059	0.6637
0.6988	1.75	350	0.9198	0.7	0.6941	0.6875	0.7	0.6825	0.6866	0.6704
0.6851	2.0	400	0.8607	0.72	0.7248	0.7310	0.72	0.6978	0.7067	0.6965
0.548	2.25	450	0.8659	0.74	0.7435	0.7557	0.74	0.7304	0.7294	0.7250
0.4898	2.5	500	0.9184	0.73	0.7379	0.7079	0.73	0.7599	0.7229	0.7203
0.5683	2.75	550	0.9207	0.72	0.7188	0.7089	0.72	0.7429	0.7150	0.7195
0.4971	3.0	600	0.9256	0.72	0.7257	0.7104	0.72	0.7384	0.7141	0.7126

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.0
• Tokenizers 0.13.2