fine_tuned_BERT_cs_wikann

This model is a fine-tuned version of bert-base-multilingual-cased on the wikiann dataset. It achieves the following results on the evaluation set:

• Loss: 0.1428
• Overall Precision: 0.9090
• Overall Recall: 0.9274
• Overall F1: 0.9181
• Overall Accuracy: 0.9673

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.3011	0.4	500	0.1781	0.8588	0.8721	0.8654	0.9501
0.1717	0.8	1000	0.1524	0.8733	0.9033	0.8880	0.9565
0.1307	1.2	1500	0.1443	0.9058	0.9051	0.9054	0.9639
0.0968	1.6	2000	0.1392	0.9075	0.9107	0.9091	0.9651
0.0974	2.0	2500	0.1352	0.9030	0.9201	0.9115	0.9647
0.0603	2.4	3000	0.1410	0.9091	0.9217	0.9154	0.9667
0.054	2.8	3500	0.1428	0.9090	0.9274	0.9181	0.9673

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.15.0