biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

fixing mm placeholder replacement issue with gemma3 (#27538)

Browse Source 
Signed-off-by: tingtingtang1992 <streamttt@gmail.com>
This commit is contained in:
tingtinggithub
2025-10-27 07:34:01 -07:00
committed by GitHub
parent 5d3be3ba4c
commit 23ad820553
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/models/gemma3_mm.py
View File

@@ -403,7 +403,7 @@ class Gemma3MultiModalProcessor(BaseMultiModalProcessor[Gemma3ProcessingInfo]):
def get_repl_toks(tok: int) -> list[int]:
if tok == newline_3:
return [newline_1, newline_2]
return [newline_2, newline_1]
if tok == newline_4:
return [newline_2, newline_2]