apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-12-24 13:28:13 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[BugFix] qwen2.5vl enable_thinking=true bug fix (#3920)

Browse Source
This commit is contained in:
CSWYF3634076
2025-09-07 21:06:36 +08:00
committed by GitHub
parent 8d77c1cb51
commit c6e2a37a95
1 changed files with 10 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
fastdeploy/input/qwen_vl_processor/qwen_vl_processor.py
View File

@@ -249,6 +249,16 @@ class QwenVLProcessor(TextProcessor):
# Handle continuation of previous generation by appending existing tokens
if metadata and metadata.get("generated_token_ids"):
self.append_generated_tokens(outputs, metadata["generated_token_ids"])
enable_thinking = False
if metadata:
enable_thinking = metadata.get("enable_thinking", False)
if request.get("chat_template_kwargs"):
chat_template_kwargs = request.get("chat_template_kwargs")
enable_thinking = chat_template_kwargs.get("enable_thinking", False)
request["enable_thinking"] = enable_thinking
outputs = self.pack_outputs(outputs)
request["prompt_token_ids"] = outputs["input_ids"].tolist()