frozenleaves/vllm
1
0
Fork 0
mirror of https://github.com/vllm-project/vllm.git synced 2025-10-20 14:53:52 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix][ROCm] Fix AITER MLA V1 (#17880)

Browse Source 
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
This commit is contained in:
vllmellm
2025-05-09 16:38:21 +08:00
committed by GitHub
parent ff8c400502
commit 217db4baa6
1 changed files with 1 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/v1/attention/backends/mla/rocm_aiter_mla.py
View File

@ -95,8 +95,7 @@ class AiterMLAMetadataBuilder(MLACommonMetadataBuilder[AiterMLAMetadata]):
paged_kv_last_page_len,
)
def _build_decode(self, input_positions: torch.Tensor,
block_table: torch.Tensor,
def _build_decode(self, block_table: torch.Tensor,
seq_lens: torch.Tensor) -> AiterMLADecodeMetadata:
(
@ -106,7 +105,6 @@ class AiterMLAMetadataBuilder(MLACommonMetadataBuilder[AiterMLAMetadata]):
) = self._get_paged_kv_tensors(block_table, seq_lens)
attn_metadata = AiterMLADecodeMetadata(
input_positions=input_positions,
block_table=block_table,
seq_lens=seq_lens,
paged_kv_indptr=paged_kv_indptr,