nci-ner-v2-stage-a

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: 0.0730
• Precision: 0.8738
• Recall: 0.9052
• F1: 0.8892

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: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1
0.1492	0.5339	500	0.1215	0.7791	0.7827	0.7809
0.0893	1.0678	1000	0.0935	0.8005	0.8559	0.8273
0.0763	1.6017	1500	0.0861	0.8284	0.8707	0.8490
0.0593	2.1356	2000	0.0821	0.8435	0.8783	0.8605
0.0492	2.6695	2500	0.0893	0.8343	0.8818	0.8574
0.0423	3.2034	3000	0.0819	0.8570	0.8857	0.8712
0.0379	3.7373	3500	0.0832	0.8562	0.8900	0.8728
0.0359	4.2712	4000	0.0851	0.8535	0.8913	0.8720
0.0386	4.8051	4500	0.0838	0.8555	0.8917	0.8732

Framework versions

• Transformers 4.46.3
• Pytorch 2.10.0+cu128
• Datasets 2.21.0
• Tokenizers 0.20.3