From 3d21ed9ee86cce2e4f4ca2852c4d31d1a366091a Mon Sep 17 00:00:00 2001
From: Peipei <51022443+booker123456@users.noreply.github.com>
Date: Sun, 28 Sep 2025 14:08:24 +0800
Subject: [PATCH] [Bugfix]Fix quant_config input parameter bug in qwenvl series
 (#3220)

### What this PR does / why we need it?
Fix quant_config input parameter bug in qwenvl series. Currently,
non-instantiated variables should be passed.
### Does this PR introduce _any_ user-facing change?
None
### How was this patch tested?

- vLLM version: v0.10.2
- vLLM main:
https://github.com/vllm-project/vllm/commit/releases/v0.11.0

Signed-off-by: booker123456 <945658361@qq.com>
---
 vllm_ascend/models/qwen2_5_vl.py                 | 2 +-
 vllm_ascend/models/qwen2_5_vl_without_padding.py | 6 +++---
 vllm_ascend/models/qwen2_vl.py                   | 2 +-
 3 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/vllm_ascend/models/qwen2_5_vl.py b/vllm_ascend/models/qwen2_5_vl.py
index a3e53c0..f240fd1 100644
--- a/vllm_ascend/models/qwen2_5_vl.py
+++ b/vllm_ascend/models/qwen2_5_vl.py
@@ -509,7 +509,7 @@ class AscendQwen2_5_VLForConditionalGeneration(
             self.visual = AscendQwen2_5_VisionTransformer(
                 vision_config=config.vision_config,
                 norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self.quant_config,
+                quant_config=quant_config,
                 prefix=maybe_prefix(prefix, "visual"),
             )
 
diff --git a/vllm_ascend/models/qwen2_5_vl_without_padding.py b/vllm_ascend/models/qwen2_5_vl_without_padding.py
index 54f5947..f62009b 100644
--- a/vllm_ascend/models/qwen2_5_vl_without_padding.py
+++ b/vllm_ascend/models/qwen2_5_vl_without_padding.py
@@ -495,7 +495,7 @@ class AscendQwen2_5_VLForConditionalGeneration_Without_Padding(
             self.visual = AscendQwen2_5_VisionTransformer_Without_Padding(
                 vision_config=config.vision_config,
                 norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self.quant_config,
+                quant_config=quant_config,
                 prefix=maybe_prefix(prefix, "visual"),
             )
 
@@ -574,7 +574,7 @@ class AscendQwen3VLForConditionalGeneration(Qwen3VLForConditionalGeneration):
             self.visual = AscendQwen3_VisionTransformer(
                 config.vision_config,
                 norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self.quant_config,
+                quant_config=quant_config,
                 prefix=maybe_prefix(prefix, "visual"),
                 use_data_parallel=self.use_data_parallel)
 
@@ -625,7 +625,7 @@ class AscendQwen3VLMoeForConditionalGeneration(
             self.visual = AscendQwen3_VisionTransformer(
                 config.vision_config,
                 norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self.quant_config,
+                quant_config=quant_config,
                 prefix=maybe_prefix(prefix, "visual"),
                 use_data_parallel=self.use_data_parallel,
             )
diff --git a/vllm_ascend/models/qwen2_vl.py b/vllm_ascend/models/qwen2_vl.py
index 0eea6f8..9648e07 100644
--- a/vllm_ascend/models/qwen2_vl.py
+++ b/vllm_ascend/models/qwen2_vl.py
@@ -357,6 +357,6 @@ class AscendQwen2VLForConditionalGeneration(Qwen2VLForConditionalGeneration):
             self.visual = AscendQwen2VisionTransformer(
                 self.config.vision_config,
                 norm_eps=getattr(self.config, "rms_norm_eps", 1e-6),
-                quant_config=self.vllm_config.quant_config,
+                quant_config=vllm_config.quant_config,
                 prefix=maybe_prefix(prefix, "visual"),
             )