xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-24 03:37:54 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] Fix false assertion with spec-decode=[2,4,..] and TP>2 (#29036)

Browse Source 
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
This commit is contained in:
Lucas Wilkinson 2025-11-19 16:43:54 -05:00 committed by GitHub
parent 22e44ad589
commit 8f4f77a727
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/config/compilation.py
View File

@ -921,7 +921,7 @@ class CompilationConfig:
self, uniform_decode_query_len: int, tensor_parallel_size: int self, uniform_decode_query_len: int, tensor_parallel_size: int
): ):
multiple_of = uniform_decode_query_len multiple_of = uniform_decode_query_len
if tensor_parallel_size > 1: if tensor_parallel_size > 1 and self.pass_config.enable_sequence_parallelism:
multiple_of = max(uniform_decode_query_len, tensor_parallel_size) multiple_of = max(uniform_decode_query_len, tensor_parallel_size)
if ( if (
multiple_of % uniform_decode_query_len != 0 multiple_of % uniform_decode_query_len != 0