BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext-finetuned-ner-chemdis-30-v2

This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3340
• Precision: 0.6703
• Recall: 0.8855
• F1: 0.7631
• Accuracy: 0.9382

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: 32
• 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: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	43	0.2075	0.4889	0.7413	0.5892	0.9164
No log	2.0	86	0.1750	0.5982	0.8305	0.6955	0.9263
No log	3.0	129	0.1479	0.6585	0.8503	0.7422	0.9386
No log	4.0	172	0.1578	0.6537	0.8567	0.7416	0.9344
No log	5.0	215	0.1603	0.6628	0.8835	0.7574	0.9376
No log	6.0	258	0.1724	0.6591	0.8430	0.7398	0.9341
No log	7.0	301	0.1768	0.6681	0.8238	0.7378	0.9352
No log	8.0	344	0.2593	0.6556	0.8832	0.7525	0.9358
No log	9.0	387	0.2100	0.6644	0.8080	0.7292	0.9348
No log	10.0	430	0.2624	0.6694	0.8747	0.7584	0.9378
No log	11.0	473	0.2793	0.6740	0.8712	0.7600	0.9377
0.1228	12.0	516	0.3408	0.6567	0.8937	0.7571	0.9357
0.1228	13.0	559	0.3075	0.6611	0.8593	0.7473	0.9358
0.1228	14.0	602	0.3340	0.6703	0.8855	0.7631	0.9382
0.1228	15.0	645	0.2862	0.6832	0.7556	0.7176	0.9338
0.1228	16.0	688	0.3648	0.6646	0.8896	0.7608	0.9368
0.1228	17.0	731	0.3296	0.6728	0.8733	0.7600	0.9382
0.1228	18.0	774	0.3460	0.6709	0.8663	0.7562	0.9372
0.1228	19.0	817	0.3535	0.6667	0.8640	0.7526	0.9367

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.0+cu126
• Datasets 3.6.0
• Tokenizers 0.22.1