deepseek-ai / deepseek-ai/FlashMLA
flash mla kernal是否支持同一batch下不同query的动态token数？
Open
- Dominant language
- C++
- Stars
- 12.9k
- Forks
- 1.2k
- Avg merge
- 4h 20m
- Merged PRs (30d)
- 2
Description
q: (batch_size, seq_len_q, num_heads_q, head_dim)
flash mla算子的输入这个seq_len_q是定死的，那么是否不支持不同的seq_len_q进入kernal计算？
Contributor guide
No contributing guide indexed for this repository
Research direction
Start at the FlashMLA operator entry point and trace how (batch_size, seq_len_q, num_heads_q, head_dim) is passed into the kernel. Check whether each query can carry a different seq_len_q within one batch; done means a documented answer backed by the relevant behavior, or a scoped implementation plan if unsupported.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- machine-learning, performance
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100