Deepseek v3.2 no supported?
Open
- Dominant language
- Python
- Stars
- 100
- Forks
- 30
- Avg merge
- 3d 14h
- Merged PRs (30d)
- 1
Description
I've noticed that there's no implementation of the new architecture feature (DSA, sparse attn for Deepseek v3.2) in the main simulation procedure.
Though there is "index_topk" in model_config (but without index_head_dim, index_n_heads, etc...). Also a kernel_sim program for DSA (sparse_mla_fp8.py) and some csv result files. But it seems unrelated to the main and model code.
Is the DSA feature in roadmap?
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.