GitHub · Python
HKUSTDial/flash-sparse-attention
Trainable fast and memory-efficient sparse attention
- Stars
- 750
- Forks
- 54
- Stars gained in 24 hours
- −9
- Forks gained in 24 hours
- +1
Topics: flash-attention, sparse-attention, kernel, flash-sparse-attention, triton
Latest data point: Nov 22, 2026 · 750 stars · 54 forks