LLM-Finetuning

Why we included this project

These notebooks show how to take an open large language model and bend it to your own data. The main lessons work through LoRA and QLoRA on top of Hugging Face's transformers and PEFT libraries, which is enough to fine-tune a model like Llama 2, Falcon, or Mistral on a modest GPU, and every notebook opens in Colab so you can run it without setting up a local environment. It reads as a tutorial set more than a production framework: good for developers and students who want working templates to copy and modify, with the path from basic LoRA through 8-bit and 4-bit quantization to RLHF laid out in one place. The later notebooks branch into RAG with LangChain and knowledge-graph question answering, so there is value beyond fine-tuning itself.