roberta-base-ontonotes

This model is a fine-tuned version of roberta-base on the tner/ontonotes5 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0695
• Recall: 0.9227
• Precision: 0.9013
• F1: 0.9118
• Accuracy: 0.9820

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: 8e-05
• train_batch_size: 32
• eval_batch_size: 160
• seed: 75241309
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• training_steps: 6000

Training results

Training Loss	Epoch	Step	Validation Loss	Recall	Precision	F1	Accuracy
0.1305	0.31	600	0.1169	0.8550	0.8139	0.8340	0.9681
0.118	0.63	1200	0.0925	0.8769	0.8592	0.8680	0.9750
0.0937	0.94	1800	0.0874	0.8939	0.8609	0.8771	0.9764
0.0698	1.25	2400	0.0821	0.9066	0.8775	0.8918	0.9784
0.0663	1.56	3000	0.0827	0.9124	0.8764	0.8940	0.9789
0.0624	1.88	3600	0.0732	0.9179	0.8868	0.9021	0.9804
0.0364	2.19	4200	0.0750	0.9204	0.8968	0.9085	0.9816
0.0429	2.5	4800	0.0699	0.9198	0.9031	0.9114	0.9818
0.0323	2.82	5400	0.0697	0.9227	0.9008	0.9116	0.9819
0.0334	3.13	6000	0.0695	0.9227	0.9013	0.9118	0.9820

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0