apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-06 09:07:10 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Feature/online/vs think 20250813 (#3440)

Browse Source 
* add stream

* fix ernie_vl_reasoning_parsers

* fix bug
This commit is contained in:
luukunn
2025-08-15 18:33:58 +08:00
committed by GitHub
parent 33abfddd9b
commit edf1ca07af
8 changed files with 206 additions and 194 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
fastdeploy/entrypoints/openai/serving_chat.py
View File

@@ -252,22 +252,20 @@ class OpenAIServingChat:
logprobs_res = self._create_chat_logprobs(
output_top_logprobs, request.logprobs, request.top_logprobs
)
if self.engine_client.data_processor.tool_parser_obj and not res["finished"]:
tool_delta_message = output["tool_delta_message"]
if tool_delta_message is None:
continue
delta_message = tool_delta_message
delta_message.reasoning_content = output.get("reasoning_content")
if delta_message.tool_calls:
tool_called = True
if not res["finished"]:
if "reasoning_delta_message" in output:
delta_message = output["reasoning_delta_message"]
elif "tool_delta_message" in output:
delta_message = output["tool_delta_message"]
if delta_message is not None and delta_message.tool_calls:
tool_called = True
else:
delta_message = DeltaMessage(content=delta_text)
else:
delta_message = DeltaMessage(
content=delta_text,
reasoning_content=output.get("reasoning_content"),
prompt_token_ids=None,
completion_token_ids=None,
tool_calls=None,
)
delta_message = DeltaMessage(content=delta_text)
if delta_message is None:
continue
choice = ChatCompletionResponseStreamChoice(
index=0,