sursani/Mistral-7B-v0.1-sft-ultrachat1000

General Information

For this supervised fine-tuning, I am using:

• Mistral-7B-v0.1 LLM
• Datasets for loading a SFT dataset from the 🤗 hub, and preparing it for the model
• BitsandBytes and PEFT for fine-tuning the model on consumer hardware, leveraging Q-LoRa, a technique which drastically reduces the compute requirements for fine-tuning
• TRL, a library which includes useful Trainer classes for LLM fine-tuning.