biomedical-ner-all-finetuned-ner-macrobat-v1

This model is a fine-tuned version of d4data/biomedical-ner-all on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3977
• Precision: 0.7596
• Recall: 0.8095
• F1: 0.7837
• Accuracy: 0.8995

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: 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: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	12	0.9921	0.4039	0.2615	0.3174	0.7220
No log	2.0	24	0.6671	0.4591	0.5475	0.4994	0.8118
No log	3.0	36	0.5148	0.5729	0.6826	0.6229	0.8485
No log	4.0	48	0.4427	0.6332	0.7487	0.6861	0.8718
No log	5.0	60	0.4042	0.6693	0.7646	0.7138	0.8783
No log	6.0	72	0.3779	0.6896	0.7771	0.7308	0.8858
No log	7.0	84	0.3691	0.7077	0.7979	0.7501	0.8902
No log	8.0	96	0.3669	0.7193	0.8037	0.7592	0.8914
No log	9.0	108	0.3684	0.7254	0.8027	0.7621	0.8941
No log	10.0	120	0.3670	0.7398	0.8119	0.7741	0.8974
No log	11.0	132	0.3655	0.7466	0.8017	0.7732	0.8980
No log	12.0	144	0.3732	0.7515	0.8123	0.7807	0.8971
No log	13.0	156	0.3824	0.7487	0.8075	0.7770	0.8989
No log	14.0	168	0.3774	0.7590	0.8051	0.7814	0.8990
No log	15.0	180	0.3837	0.7531	0.8032	0.7773	0.8984
No log	16.0	192	0.3937	0.7554	0.8119	0.7826	0.8987
No log	17.0	204	0.3977	0.7596	0.8095	0.7837	0.8995
No log	18.0	216	0.4063	0.7520	0.8075	0.7788	0.8980
No log	19.0	228	0.4077	0.7537	0.8017	0.7770	0.8969
No log	20.0	240	0.4140	0.7489	0.8099	0.7782	0.8975
No log	21.0	252	0.4186	0.7515	0.8123	0.7807	0.8976
No log	22.0	264	0.4266	0.7383	0.8109	0.7729	0.8947

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 3.6.0
• Tokenizers 0.22.1