Learn how to finetune large language models on your own consumer hardware using LoRA and tools from the PyTorch and Hugging Face ecosystem. Parameter Efficient Fine-Tuning (PEFT) methods can help reduce the number of trainable parameters while maintaining performance. LoRA is a popular PEFT method that decomposes weight
Table of contents
IntroductionWhat makes our Llama fine-tuning expensive?Parameter Efficient Fine-Tuning (PEFT) methodsLow-Rank Adaptation for Large Language Models (LoRA) using 🤗 PEFTThe base model can be in any dtype : leveraging SOTA LLM quantization and loading the base model in 4-bit precisionQLoRA: One of the core contributions of bitsandbytes towards the democratization of AIUsing QLoRA in practiceUsing TRL for LLM trainingPutting all the pieces togetherSort: