apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 16:48:03 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Merge branch 'feature/online/vs_think_20250813' into release/2.1

Browse Source
This commit is contained in:
Jiang-Jia-Jun
2025-08-14 17:13:36 +08:00
committed by GitHub
parent 03347626a6 81092c0fe3
commit 02596fc537
23 changed files with 1056 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
fastdeploy/entrypoints/openai/serving_chat.py
View File

@@ -141,6 +141,7 @@ class OpenAIServingChat:
previous_num_tokens = 0
num_prompt_tokens = 0
num_choices = 1
tool_called = False
max_streaming_response_tokens = (
request.max_streaming_response_tokens
if request.max_streaming_response_tokens is not None
@@ -245,20 +246,28 @@ class OpenAIServingChat:
output = res["outputs"]
delta_text = output["text"]
output_top_logprobs = output["top_logprobs"]
previous_num_tokens += len(output["token_ids"])
logprobs_res: Optional[LogProbs] = None
if request.logprobs and output_top_logprobs is not None:
logprobs_res = self._create_chat_logprobs(
output_top_logprobs, request.logprobs, request.top_logprobs
)
previous_num_tokens += len(output["token_ids"])
delta_message = DeltaMessage(
content=delta_text,
reasoning_content=output.get("reasoning_content"),
prompt_token_ids=None,
completion_token_ids=None,
tool_calls=output.get("tool_call_content", []),
)
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
else:
delta_message = DeltaMessage(
content=delta_text,
reasoning_content=output.get("reasoning_content"),
prompt_token_ids=None,
completion_token_ids=None,
tool_calls=None,
)
choice = ChatCompletionResponseStreamChoice(
index=0,
@@ -276,10 +285,7 @@ class OpenAIServingChat:
max_tokens = request.max_completion_tokens or request.max_tokens
if has_no_token_limit or previous_num_tokens != max_tokens:
choice.finish_reason = "stop"
if (
self.engine_client.reasoning_parser == "ernie_x1"
and output.get("finish_reason", "") == "tool_calls"
):
if tool_called:
choice.finish_reason = "tool_calls"
else:
choice.finish_reason = "length"
@@ -419,7 +425,7 @@ class OpenAIServingChat:
role="assistant",
content=output["text"],
reasoning_content=output.get("reasoning_content"),
tool_calls=output.get("tool_call_content"),
tool_calls=output.get("tool_call"),
prompt_token_ids=prompt_token_ids if request.return_token_ids else None,
completion_token_ids=completion_token_ids if request.return_token_ids else None,
text_after_process=text_after_process if request.return_token_ids else None,
@@ -439,7 +445,7 @@ class OpenAIServingChat:
max_tokens = request.max_completion_tokens or request.max_tokens
if has_no_token_limit or previous_num_tokens != max_tokens:
choice.finish_reason = "stop"
if self.engine_client.reasoning_parser == "ernie_x1" and output.get("finish_reason", "") == "tool_calls":
if output.get("tool_call"):
choice.finish_reason = "tool_calls"
else:
choice.finish_reason = "length"