This directory provides demonstrations of various quantization techniques for LLMs. Currently, it contains the "Quantization Demo" notebook.
- Filename:
quantization.ipynb - Objective: To debias and detoxify text using advanced machine learning techniques.
- Key Features:
- Fine-tuning the Llama2-7b model using LoRA and Quantization techniques.
- Utilizes the dataset located at
../../data/debiased_profanity_check_with_keywords.csv.