bert-keyword-extractor_oknashar

This model is a fine-tuned version of yanekyuk/bert-keyword-extractor on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2965
• Precision: 0.5862
• Recall: 0.6937
• F1: 0.6354
• Accuracy: 0.9555

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.204	1.0	1357	0.1717	0.3345	0.3419	0.3382	0.9375
0.161	2.0	2714	0.1547	0.3949	0.4845	0.4352	0.9426
0.1333	3.0	4071	0.1506	0.4369	0.5342	0.4807	0.9481
0.1147	4.0	5428	0.1461	0.5121	0.5424	0.5268	0.9522
0.0925	5.0	6785	0.1514	0.4864	0.5868	0.5319	0.9506
0.0788	6.0	8142	0.1477	0.5173	0.6043	0.5574	0.9529
0.0676	7.0	9499	0.1765	0.5126	0.6277	0.5644	0.9543
0.0542	8.0	10856	0.1638	0.5066	0.6704	0.5771	0.9500
0.0453	9.0	12213	0.1746	0.5158	0.6598	0.5790	0.9517
0.0366	10.0	13570	0.1905	0.5876	0.6429	0.6140	0.9569
0.0327	11.0	14927	0.2215	0.5709	0.6593	0.6119	0.9557
0.0296	12.0	16284	0.2109	0.5648	0.6598	0.6086	0.9557
0.0269	13.0	17641	0.2226	0.5632	0.6745	0.6138	0.9543
0.0203	14.0	18998	0.2464	0.5682	0.6861	0.6217	0.9543
0.0163	15.0	20355	0.2738	0.5747	0.6791	0.6226	0.9541
0.016	16.0	21712	0.2805	0.5868	0.6832	0.6314	0.9557
0.0115	17.0	23069	0.2923	0.5798	0.6879	0.6292	0.9551
0.0107	18.0	24426	0.2923	0.5943	0.6926	0.6397	0.9558
0.0105	19.0	25783	0.2992	0.5890	0.6902	0.6356	0.9553
0.0087	20.0	27140	0.2965	0.5862	0.6937	0.6354	0.9555

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1