Skip to content
Closed
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion vllm/v1/attention/backends/mla/rocm_aiter_mla.py
Original file line number Diff line number Diff line change
Expand Up @@ -129,7 +129,7 @@ class AiterMLADecodeMetadata(MLACommonDecodeMetadata):
# The max query output length: int
max_qo_len: int | None = None
# Minimum KV length used by Gluon to choose a safe split count.
min_kv_seq_len: int = 1
min_kv_seq_len: int = 16384
# Small-head decode uses Gluon (avoids padding to 16).
use_gluon_decode: bool = False
# Whether persistent MLA metadata was computed
Expand Down
Loading