QLoRA: Efficient Finetuning of Quantized LLMs

Authors: Dettmers et al. Year: 2023 ArXiv/Link: https://arxiv.org/abs/2305.14314

Summary

Combined 4-bit quantization with LoRA to enable fine-tuning of large models on consumer GPUs.

Key Concepts

• Quantization
• LoRA
• 4-bit precision
• Memory efficiency
• Consumer-grade GPU training

Impact

Democratized fine-tuning of large language models

Category

Fine-tuning & Adaptation

---