A high-performance NF4 dequantization Triton kernel that is 1.41x faster than bitsandbytes.
Why it made the shelf
It is a developer tool and open-source library/kernel designed to optimize machine learning model dequantization.
Seen on
Behind the tool
People, backing, growth, and ownership. Every claim links to its source.
Source review queued
Public trail
1 reference from 1 publisher.