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

fix response processsors (#3826)

Browse Source 
* fix response processsors

* fix ci

* fix ut

---------

Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>
This commit is contained in:
RichardWooSJTU
2025-09-04 16:01:25 +08:00
committed by GitHub
parent 22c165d6dd
commit f36a388ffe
3 changed files with 9 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
fastdeploy/entrypoints/openai/response_processors.py
View File

@@ -17,6 +17,7 @@
from typing import Any, List, Optional
from fastdeploy.input.tokenzier_client import AsyncTokenizerClient, ImageDecodeRequest
from fastdeploy.utils import api_server_logger
class ChatResponseProcessor:
@@ -41,6 +42,8 @@ class ChatResponseProcessor:
self.eos_token_id = eos_token_id
if decoder_base_url is not None:
self.decoder_client = AsyncTokenizerClient(base_url=decoder_base_url)
else:
self.decoder_client = None
self._mm_buffer: List[Any] = [] # Buffer for accumulating image token_ids
self._end_image_code_request_output: Optional[Any] = None
self._multipart_buffer = []
@@ -74,6 +77,7 @@ class ChatResponseProcessor:
include_stop_str_in_output: Whether or not to include stop strings in the output.
"""
for request_output in request_outputs:
api_server_logger.debug(f"request_output {request_output}")
if not self.enable_mm_output:
yield self.data_processor.process_response_dict(
response_dict=request_output,
@@ -112,7 +116,7 @@ class ChatResponseProcessor:
yield request_output
elif decode_type == 1:
self._mm_buffer.extend(token_ids)
self._mm_buffer.append(token_ids)
self._end_image_code_request_output = request_output
else:
self.accumulate_token_ids(request_output)