A complete solution for Kaggle's LLM Prompt Recovery Competition, featuring an in-depth guide to QLoRA Fine-Tuning.
Notebooks are hosted on Github Pages with Jupyter Book. They are also available for viewing, copying, and editing on Kaggle (Part 1, Part 2, Part 3), complete with training data, pre-trained LLMs and Adapters, and runtime environment with GPU/TPU access.