IRyS-NER-Paper

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

• Loss: 0.1197
• Precision: 0.7812
• Recall: 0.7548
• F1: 0.7677
• Accuracy: 0.9686

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: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	81	0.1969	0.6799	0.5433	0.6040	0.9448
No log	2.0	162	0.1423	0.7634	0.6617	0.7089	0.9623
No log	3.0	243	0.1197	0.7812	0.7548	0.7677	0.9686
No log	4.0	324	0.1335	0.7819	0.7505	0.7659	0.9678
No log	5.0	405	0.1326	0.7345	0.8013	0.7664	0.9650
No log	6.0	486	0.1427	0.7471	0.8182	0.7810	0.9657
0.1446	7.0	567	0.1439	0.7447	0.8203	0.7807	0.9666
0.1446	8.0	648	0.1586	0.7368	0.8288	0.7801	0.9650
0.1446	9.0	729	0.1707	0.7273	0.8288	0.7747	0.9629
0.1446	10.0	810	0.1650	0.7438	0.8288	0.784	0.9649

Framework versions

• Transformers 4.27.2
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2