ner_column_bert-base-NER

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

• Loss: 0.1855
• Precision: 0.7651
• Recall: 0.7786
• F1: 0.7718
• Accuracy: 0.9026

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: 64
• eval_batch_size: 64
• 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
No log	1.0	702	0.7382	0.2576	0.1887	0.2178	0.7127
0.9356	2.0	1404	0.4405	0.5139	0.4331	0.4700	0.8157
0.5445	3.0	2106	0.3608	0.5712	0.5143	0.5413	0.8404
0.5445	4.0	2808	0.3226	0.6188	0.5840	0.6009	0.8550
0.4316	5.0	3510	0.2757	0.6788	0.6569	0.6676	0.8728
0.3605	6.0	4212	0.2828	0.6584	0.6346	0.6463	0.8697
0.3605	7.0	4914	0.2456	0.7108	0.6926	0.7015	0.8820
0.3153	8.0	5616	0.2385	0.7055	0.6986	0.7021	0.8855
0.282	9.0	6318	0.2345	0.7044	0.6961	0.7002	0.8853
0.2587	10.0	7020	0.2313	0.7081	0.7049	0.7065	0.8862
0.2587	11.0	7722	0.2026	0.7734	0.7537	0.7634	0.8968
0.239	12.0	8424	0.1980	0.7651	0.7687	0.7669	0.8991
0.2241	13.0	9126	0.2091	0.7368	0.7423	0.7395	0.8936
0.2241	14.0	9828	0.1954	0.7693	0.7684	0.7689	0.8987
0.2124	15.0	10530	0.1916	0.7668	0.7749	0.7708	0.9008
0.2025	16.0	11232	0.1841	0.7699	0.7794	0.7746	0.9024
0.2025	17.0	11934	0.1938	0.7527	0.7626	0.7576	0.8992
0.193	18.0	12636	0.1849	0.7705	0.7841	0.7772	0.9040
0.1877	19.0	13338	0.1927	0.7510	0.7649	0.7579	0.9005
0.1821	20.0	14040	0.1855	0.7651	0.7786	0.7718	0.9026

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3