bert-finetuned-ner-17-2epochs_even_more_data

This model is a fine-tuned version of FacebookAI/roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2271
• Precision: 0.8671
• Recall: 0.8860
• F1: 0.8764
• Accuracy: 0.9717

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: 32
• eval_batch_size: 128
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.1861	0.1013	400	0.1766	0.4918	0.3811	0.4294	0.9014
0.1197	0.2025	800	0.1510	0.6989	0.6392	0.6677	0.9399
0.0630	0.3038	1200	0.1601	0.7535	0.8117	0.7815	0.9534
0.0395	0.4051	1600	0.1506	0.8215	0.8770	0.8483	0.9652
0.0288	0.5063	2000	0.1764	0.8406	0.8582	0.8493	0.9668
0.0210	0.6076	2400	0.1632	0.8554	0.8785	0.8668	0.9702
0.0177	0.7089	2800	0.1504	0.8677	0.8755	0.8715	0.9717
0.0133	0.8101	3200	0.1946	0.8719	0.8785	0.8752	0.9730
0.0132	0.9114	3600	0.1760	0.8484	0.8777	0.8628	0.9703
0.0112	1.0127	4000	0.2094	0.8468	0.8792	0.8627	0.9687
0.0087	1.1139	4400	0.2003	0.8700	0.8837	0.8768	0.9730
0.0085	1.2152	4800	0.1958	0.8688	0.8942	0.8813	0.9730
0.0070	1.3165	5200	0.2149	0.8495	0.8852	0.8670	0.9706
0.0064	1.4177	5600	0.2139	0.8743	0.8822	0.8783	0.9712
0.0055	1.5190	6000	0.2070	0.8662	0.8890	0.8775	0.9726
0.0053	1.6203	6400	0.2201	0.8603	0.8822	0.8711	0.9710
0.0045	1.7215	6800	0.2325	0.8611	0.8882	0.8744	0.9717
0.0052	1.8228	7200	0.2173	0.8691	0.8867	0.8778	0.9728
0.0043	1.9241	7600	0.2287	0.8662	0.8837	0.8749	0.9715
0.0043	2.0	7900	0.2271	0.8671	0.8860	0.8764	0.9717

Framework versions

• Transformers 5.7.0
• Pytorch 2.10.0+cu128
• Datasets 4.8.5
• Tokenizers 0.22.2