BetaRare
HKUSTDial/flash-sparse-attention
Trainable fast and memory-efficient sparse attention
- Owner
- HKUSTDial (organization)
- Stars
- Stars: 762
- Forks
- Forks: 54
- Language
- Python
- License
- BSD-3-Clause
- Created
- 2025
- Topics
- flash-attentionsparse-attentionkernelflash-sparse-attentiontriton
- Tier
- Beta (700 to 1,999 ★) What this tier means
More repos by HKUSTDial
All 14 repos by HKUSTDial- HKUSTDial/Supervisor-SkillsStars: 11,471PythonStable
- HKUSTDial/NL2SQL_HandbookStars: 1,628PythonBeta
- HKUSTDial/awesome-data-agentsStars: 798PythonBeta
- HKUSTDial/DataMagicStars: 359TypeScriptAlpha
- HKUSTDial/DeepFundStars: 303PythonAlpha
- HKUSTDial/DeepEarStars: 288PythonAlpha
Nearby in Python
All Python repos- RPG-fan/Cline-Recursive-Chain-of-Thought-System-CRCT-Stars: 762PythonBeta
- dsta022/Loop-Engineering-for-VLAStars: 762PythonBeta
- zj-unicom-ai/UniEmployeeStars: 762PythonBeta
- revsys/django-friendshipStars: 762PythonBeta
- joeyespo/pytest-watchStars: 762PythonBeta
- cbyn/bitpredictStars: 762PythonBeta