deepseek-ai

deepseek-ai/flashmla

steady

FlashMLA: Efficient Multi-head Latent Attention Kernels

C++ View on GitHub

Stars

12,745

Forks

1,091

Open issues

68

24h

0

0.0%

7d

+16

+0.1%

Refresh

2h

Star history (7 days)

Last checked

53m ago

Last pushed

30 Apr 2026

Next check

just now