Glossary: Fine-tuning LLMs
The library QLoRA needs to load and compute with 4-bit weights. It requires an NVIDIA GPU.
QLoRANF4