bert-finetuned-pos

Model is a fine-tuned version of bert-base-uncased on the conll2003 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2140
• Precision: 0.9258
• Recall: 0.9257
• F1: 0.9257
• Accuracy: 0.9483

Model description

This model fine-tunes bert-base-uncased on the CoNLL-2003 dataset for part-of-speech (POS) tagging. It is trained to label each token in a sentence with its corresponding POS tag, achieving high precision and recall on the evaluation set. Suitable for NLP tasks that require accurate grammatical structure identification in English text.

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: 8
• eval_batch_size: 8
• 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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.2395	1.0	1756	0.2447	0.9170	0.9157	0.9164	0.9417
0.1589	2.0	3512	0.2177	0.9245	0.9209	0.9227	0.9463
0.1191	3.0	5268	0.2140	0.9258	0.9257	0.9257	0.9483

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1