Project – Text Classification Using LoRA

This project fine-tunes a transformer model on the AG News dataset using Low-Rank Adaptation (LoRA). It demonstrates how to adapt large language models efficiently using fewer trainable parameters.

Key Features

• Applied LoRA for parameter-efficient finetuning
• Used Hugging Face Transformers and PEFT
• Classified AG News into 4 categories with high accuracy

Dataset

• AG News (via Hugging Face Datasets)

Requirements

• Python
• transformers
• peft
• accelerate

How to Run

pip install transformers datasets peft accelerate
jupyter notebook LoRA_Text_Classification.ipynb