Fine-tuning LLMs

QLoRA fine-tuning of Mistral 7B in Google Colab

Overview

This is my first article on Medium — Fine-tuning Mistral 7B in Google Colab with QLoRA, a complete guide.

The resulting fine-tuned model, trained on synthetic data, outperforms OpenAI's GPT-4 on a domain-specific benchmark.

The project is also on my GitHub and the models are on Hugging Face.

Fine-tuning LLMs — Medium article