Modelpublic

MM2157/AraBERT_token_classification__AraEval24_truncated

sourceHugging Faceupdated 3y agoView on Hugging Face

0likes21downloads

Model Card

AraBERTtokenclassification_AraEval24truncated

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

—Loss: 1.7905
—Precision: 0.2295
—Recall: 0.1820
—F1: 0.2031
—Accuracy: 0.6274

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
—trainbatchsize: 8
—evalbatchsize: 8
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
1.563	1.0	798	1.5140	0.1409	0.0447	0.0679	0.6364
1.1924	2.0	1596	1.4530	0.1759	0.0791	0.1091	0.6424
1.0371	3.0	2394	1.4860	0.2043	0.1060	0.1396	0.6392
0.8527	4.0	3192	1.5512	0.2062	0.1285	0.1584	0.6292
0.772	5.0	3990	1.5781	0.2123	0.1545	0.1789	0.6175
0.6296	6.0	4788	1.6486	0.2398	0.1559	0.1890	0.6288
0.5614	7.0	5586	1.6764	0.2226	0.1797	0.1989	0.6246
0.5119	8.0	6384	1.7380	0.2247	0.1875	0.2044	0.6210
0.4581	9.0	7182	1.7726	0.2389	0.1691	0.1980	0.6347
0.4594	10.0	7980	1.7905	0.2295	0.1820	0.2031	0.6274

Framework versions

—Transformers 4.30.2
—Pytorch 1.12.1
—Datasets 2.13.2
—Tokenizers 0.13.3