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
[Model] Fix Qwen3VL mm mapping (#27027)
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Jee Jee Li committed
Oct 17, 2025 at 00:45 UTC
9f4e30904bbe6055a5561106d5465d7e870db364
1 file changed
+2
-2
vllm/model_executor/models/qwen3_vl.py
+2
-2
@@ -1776,6 +1776,6 @@ class Qwen3VLForConditionalGeneration(
1776
"""
1777
return MultiModelKeys.from_string_field(
1778
language_model="language_model",
1779
- connector="model.visual.merger",
1780
- tower_model="model.visual.",
1779
+ connector="visual.merger",
1780
+ tower_model="visual.",
1781
)