Skip to content

Issues: deepseek-ai/FlashMLA

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

31 Open 3 Closed
31 Open 3 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

why head_dim and vhead_dim only support 576,512
#43 opened Feb 25, 2025 by liuchanchen
FlashMLA是不是要求一条序列的一层的kv cache是连续的，block idx的作用是什么
#42 opened Feb 25, 2025 by HPC4AI
Question about kernel synchronization in a warp group
#41 opened Feb 25, 2025 by sleepwalker2017
Could the model trained in A100 use FlashMLA in H100 for inference?
#40 opened Feb 25, 2025 by hdjsjyl
【执行测试错误】：no kernel image is available for execution on the device
#39 opened Feb 25, 2025 by RoadTLife
3
step by step study FlashMLA, what does fixed_overhead_num_blocks mean?
#38 opened Feb 25, 2025 by defei-coder
How can I compile on GPU with Compute Capability 8.6 like A6000
#34 opened Feb 24, 2025 by Song-WenPo
1
啥时候支持昇腾910B
#31 opened Feb 24, 2025 by JerryZ01
7
on H20， Could you help me improve the TFLOPS and GB/s?
#30 opened Feb 24, 2025 by robotzheng
2
[Feature]Support flashMLA decoding via flashAttn2
#29 opened Feb 24, 2025 by zhangjinnan
3
q的sequence length也是任意的吗？还是只是支持小数字的？
#28 opened Feb 24, 2025 by liuqi123123
3
我来解释下具体思路
#27 opened Feb 24, 2025 by alishangtian
5
来自deepseek r1分析的进一步优化思路，大家看对不对 O(∩_∩)O
#26 opened Feb 24, 2025 by arnewc
6
nice one
#25 opened Feb 24, 2025 by d1hr2uv
Just come to show my respect, salute!
#24 opened Feb 24, 2025 by PeterPeng-CN
4
能适配40系吗？
#23 opened Feb 24, 2025 by shesung
run error
#22 opened Feb 24, 2025 by Harrison-cc
纯外行来凑热闹，谁能简单一句话说说这个库是用来干啥的🤔
#21 opened Feb 24, 2025 by yoyogis
8
[benchmark] run flash MLA bench on H100
#19 opened Feb 24, 2025 by weedge
2
Some random scripts some may find useful
#18 opened Feb 24, 2025 by jinleic
编译错误，怎么解决？有人遇到么
#17 opened Feb 24, 2025 by DoctorBoshi
2
up matrix is not involved?
#16 opened Feb 24, 2025 by yejunguo
6
Looking forward to additional configs in Flash Attenion 4
#15 opened Feb 24, 2025 by ddiddi
compiling error
#13 opened Feb 24, 2025 by z18256199275
[Bug] Sequence length = 0
#11 opened Feb 24, 2025 by YLGH
ProTip! Updated in the last three days: updated:>2025-02-22.