ner-biomedical-maccrobat2018

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: 0.6603
• Accuracy: 0.7785

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• 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: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.6521	1.0	10	1.7474	0.2525
1.3889	2.0	20	1.0400	0.5568
0.8632	3.0	30	0.7819	0.6695
0.6142	4.0	40	0.6757	0.7196
0.4609	5.0	50	0.6340	0.7401
0.3611	6.0	60	0.6265	0.7547
0.2924	7.0	70	0.6077	0.7491
0.2378	8.0	80	0.6112	0.7546
0.1974	9.0	90	0.6043	0.7605
0.1725	10.0	100	0.6336	0.7565
0.1493	11.0	110	0.6230	0.7685
0.1311	12.0	120	0.6339	0.7693
0.1168	13.0	130	0.6295	0.7748
0.1061	14.0	140	0.6472	0.7700
0.0975	15.0	150	0.6553	0.7731
0.0914	16.0	160	0.6526	0.7773
0.0864	17.0	170	0.6558	0.7755
0.0823	18.0	180	0.6624	0.7740
0.0813	19.0	190	0.6627	0.7778
0.0786	20.0	200	0.6603	0.7785

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0