DeepSeek/FlashMLA
1
0
Fork 0
mirror of https://github.com/deepseek-ai/FlashMLA synced 2025-05-06 13:04:26 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

replace c10 optional with std optional

Browse Source
This commit is contained in:
Sijia Chen 2025-02-24 00:25:25 -08:00 committed by Sijia Chen
parent bcb90f2afd
commit 15a82b81b8
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
csrc/flash_api.cpp
View File

@ -61,7 +61,7 @@ std::vector<at::Tensor>
mha_fwd_kvcache_mla(
at::Tensor &q, // batch_size x seqlen_q x num_heads x head_size
const at::Tensor &kcache, // num_blocks x page_block_size x num_heads_k x head_size
c10::optional<const at::Tensor> &vcache_, // num_blocks x page_block_size x num_heads_k x head_size_v
std::optional<const at::Tensor> &vcache_, // num_blocks x page_block_size x num_heads_k x head_size_v
const int head_size_v,
const at::Tensor &seqlens_k, // batch_size
const at::Tensor &block_table, // batch_size x max_num_blocks_per_seq