Qwen3-VL-MoE EAGLE support for vLLM-Ascend (#6327)

### What this PR does / why we need it?
Qwen3-VL-MoE EAGLE support for vLLM-Ascend

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?
The patch tested with Qwen3-VL-30B-A3B-Instruct model

- vLLM version: v0.14.1
- vLLM main:
https://github.com/vllm-project/vllm/commit/dc917cceb877dfd13f98c538c4c96158047d98bd

Signed-off-by: Sergey_Zlobin <sirg_zlobin@mail.ru>
This commit is contained in:
Sergey-Zlobin authored and GitHub committed 2026-01-29 16:44:30 +08:00
1 parent 41a52beb26
commit 6a7b3bc29c
1 file changed
+1
@@ -198,6 +198,7 @@ class EagleProposer(VllmEagleProposer):
if self.get_model_name(model) in [
"Qwen2_5_VLForConditionalGeneration",
"Qwen3VLForConditionalGeneration",
"Qwen3VLMoeForConditionalGeneration",
]:
self.model.config.image_token_index = model.config.image_token_id
elif self.get_model_name(