cuLA

Python Medium

inclusionAI/cuLA

View on GitHub
336 stars
31 forks
8 open issues
Active Apr 2026

Beginner-Friendly Issues 8

Issues tagged for new contributors

View all on GitHub
Lightning-attn supports pretransposed states
#32 · Apr 5, 2026
Easy
good first issue help wanted performance
We need a KDA decode kernel
#29 · Apr 5, 2026
Medium
enhancement help wanted inference
More aggressive fusion of small neighboring kernels (cumsum etc.) for inference
#18 · Apr 3, 2026
Medium
enhancement help wanted inference
KDA MTP (Multi-Token Prediction) support
#17 · Apr 3, 2026
Easy
enhancement help wanted inference
Modular GDN forward / backward kernels (compatible with Kimi CP)
#13 · Apr 3, 2026
Medium
enhancement help wanted
Small B/H/S optimizations
#11 · Apr 3, 2026
Medium
enhancement help wanted performance
Larger chunk size and 2-CTA on SM10X for improved throughput
#10 · Apr 3, 2026
Medium
enhancement research performance
Polynomial approximation to mitigate the exponential bottleneck (Flash-Attention-4)
#9 · Apr 3, 2026
Medium
help wanted research performance

Sign in required

Authenticate to use favourites & bookmarks

5