A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
fixing mm placeholder replacement issue with gemma3 (#27538)
Signed-off-by: tingtingtang1992 <streamttt@gmail.com>
tingtinggithub committed
Oct 27, 2025 at 07:34 UTC
23ad820553845c546ecbe9914abd799e92e45ce6
1 file changed
+1
-1
vllm/model_executor/models/gemma3_mm.py
+1
-1
@@ -403,7 +403,7 @@ class Gemma3MultiModalProcessor(BaseMultiModalProcessor[Gemma3ProcessingInfo]):
403
404
def get_repl_toks(tok: int) -> list[int]:
405
if tok == newline_3:
406
- return [newline_1, newline_2]
406
+ return [newline_2, newline_1]
407
if tok == newline_4:
408
return [newline_2, newline_2]
409