NF4

Concept

A 4-bit Normal Float quantization format used by Q-LoRA, significantly reducing memory usage for model weights.

Mentioned in 1 video

Videos Mentioning NF4

Building with Instruction-Tuned LLMs: A Step-by-Step Guide

Building with Instruction-Tuned LLMs: A Step-by-Step Guide

DeepLearningAI

A 4-bit Normal Float quantization format used by Q-LoRA, significantly reducing memory usage for model weights.