RoBERTa-Base Quantized Model for Named Entity Recognition (NER)

This repository contains a quantized version of the RoBERTa model fine-tuned for Named Entity Recognition (NER) on the WikiANN (English) dataset. The model is particularly suitable for tagging named entities in news articles, such as persons, organizations, and locations. It has been optimized for efficient deployment using quantization techniques.

Model Details

Model Architecture: RoBERTa Base
Task: Named Entity Recognition
Dataset: WikiANN (English)
Use Case: Tagging news articles with named entities
Quantization: Float16
Fine-tuning Framework: Hugging Face Transformers

Usage

Installation

pip install transformers torch

Loading the Model

 
from transformers import RobertaTokenizerFast, RobertaForSequenceClassification, Trainer, TrainingArguments
import torch


 
# Load tokenizer

tokenizer = RobertaTokenizerFast.from_pretrained("roberta-base")
 
# Create NER pipeline
ner_pipeline = pipeline(
    "ner",
    model=model,
    tokenizer=tokenizer,
    aggregation_strategy="simple"
)

# Sample news headline
text = "Apple Inc. is planning to open a new campus in London by the end of 2025."

# Inference
entities = ner_pipeline(text)

# Display results
for ent in entities:
    print(f"{ent['word']}: {ent['entity_group']} ({ent['score']:.2f})")

Performance Metrics

Accuracy: 0.923422
Precision: 0.923052
Recall: 0.923422
F1: 0.923150

Fine-Tuning Details

Dataset

The dataset is taken from Hugging Face WikiANN (English).

Training

Number of epochs: 5
Batch size: 16
Evaluation strategy: epoch
Learning rate: 3e-5

Quantization

Post-training quantization was applied using PyTorch's built-in quantization framework to reduce the model size and improve inference efficiency.

Repository Structure


.
├── config.json
├── tokenizer_config.json    
├── sepcial_tokens_map.json 
├── tokenizer.json        
├── model.safetensors    # Fine Tuned Model
├── README.md            # Model documentation

Limitations

The model may not generalize well to domains outside the fine-tuning dataset.
Quantization may result in minor accuracy degradation compared to full-precision models.

Contributing

Contributions are welcome! Feel free to open an issue or submit a pull request if you have suggestions or improvements.