Show HN: Fast NF4 dequantization Triton kernel (1.41x faster than bitsandbytes)
By Griffith-7 · 2026-07-15 · 3 points · 0 comments
https://github.com/Griffith-7/nf4-triton-kernel
"I wanted to see if I could optimize the dequantization bottleneck during 4-bit LLM inference. By writing a custom kernel in Triton to optimize memory access patterns, I managed to get up to a 1.41x speedup over the standard bitsandbytes implementation. Check out the source code…
Open the full discussion on BetterNews