bert-base-cased-finetuned-conll2003-ner

This model is a fine-tuned version of BERT (bert-base-cased) on the CoNLL-2003 (Conference on Computational Natural Language Learning) dataset.

The model performs named entity recognition (NER). It pertains to section 2 of chapter 7 of the Hugging Face "NLP Course" (https://huggingface.co/learn/nlp-course/chapter7/2).

It was trained using the Trainer API of Hugging Face Transformers.

Code: https://github.com/sambitmukherjee/huggingface-notebooks/blob/main/course/en/chapter7/section2_pt.ipynb

Experiment tracking: https://wandb.ai/sadhaklal/bert-base-cased-finetuned-conll2003-ner

Usage

from transformers import pipeline

model_checkpoint = "sadhaklal/bert-base-cased-finetuned-conll2003-ner"
token_classifier = pipeline("token-classification", model=model_checkpoint, aggregation_strategy="simple")

print(token_classifier("My name is Sylvain and I work at Hugging Face in Brooklyn."))

Dataset

From the dataset page:

The shared task of CoNLL-2003 concerns language-independent named entity recognition. We will concentrate on four types of named entities: persons, locations, organizations and names of miscellaneous entities that do not belong to the previous three groups.

Examples: https://huggingface.co/datasets/conll2003/viewer

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.0125	1.0	1756	0.0729	0.9095	0.9339	0.9215	0.9810
0.0001	2.0	3512	0.0558	0.9265	0.9487	0.9375	0.9862
0.0001	3.0	5268	0.0578	0.9366	0.9515	0.9440	0.9867

Framework versions

• Transformers 4.37.2
• PyTorch 2.1.0+cu121
• Datasets 2.17.1
• Tokenizers 0.15.2