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

[Feature] add tool parser (#3518)

Browse Source 
* [Feature] Pass through the `chat_template_kwargs` to the data processing module (#3421)

* fix chat_template_args

* fix args

* add offline

* add offline

* fix

* fix

* fix default enable_thinking value

* fix default enable_thinking value

* modify condition

* Revert "modify condition"

This reverts commit 26430bdeb1.

* fix unit test

* add Tool Parser (#3272)

* add tool-parser

* add tool-parser

* add tool parser

* add tool parser

* fix

* add offline

* add offline

* fix

* parsers:tool&reasoning

* 修改tool parser名称·

* update

* fix reasoning-parser

* add requirements

* fix finish reason

* fix

* fix reasoning-parser

* fix

* fix

* fix

* fix

* fix

---------

Co-authored-by: zhuzixuan <zhuzixuan@baidu.com>

* [Feature] add tool parser (#3483)

* add tool parser

* add x1 enable_thinking

* restart ci

* fix vl reasoning parser

* modify call style

* modify call style

* add offline enablethinking

* fix completion

* fix

* fix unit test

* fix unit test

* fix unit test

* fix vl reasoning parser

* fix vl reasoning parser

* fix unit test

---------

Co-authored-by: zhuzixuan <zhuzixuan@baidu.com>
This commit is contained in:
luukunn
2025-08-22 11:14:35 +08:00
committed by GitHub
parent d97aab25bc
commit 4a9c04a746
31 changed files with 1289 additions and 70 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
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
@@ -246,20 +247,29 @@ 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"),
reasoning_content="",
prompt_token_ids=None,
completion_token_ids=None,
tool_calls=output.get("tool_call_content", []),
tool_calls=None,
)
if not res["finished"] and "delta_message" in output:
delta_message_output = output["delta_message"]
if delta_message_output is None:
continue
delta_message.content = delta_message_output.content or ""
delta_message.reasoning_content = delta_message_output.reasoning_content or ""
if delta_message_output.tool_calls:
delta_message.tool_calls = delta_message_output.tool_calls
tool_called = True
choice = ChatCompletionResponseStreamChoice(
index=0,
@@ -277,10 +287,7 @@ 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"
@@ -421,7 +428,7 @@ 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,
@@ -443,7 +450,7 @@ 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"