starlight-apk/Qwen
1
0
Fork 0
mirror of https://github.com/QwenLM/Qwen.git synced 2026-05-20 08:25:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #1155 from JianxinMa/main

Browse Source 
bugfix streaming mode of openai_api.py
This commit is contained in:
Jianhong Tu
2024-03-15 20:52:29 +08:00
committed by GitHub
parent 11e00874a9 a6085c2a91
commit 3ad0c83bb9
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
openai_api.py
View File

@@ -484,7 +484,7 @@ async def predict(
stop_words_ids = [tokenizer.encode(s) stop_words_ids = [tokenizer.encode(s)
for s in stop_words] if stop_words else None for s in stop_words] if stop_words else None
delay_token_num = max([len(x) for x in stop_words]) delay_token_num = max([len(x) for x in stop_words]) if stop_words_ids else 0
response_generator = model.chat_stream(tokenizer, response_generator = model.chat_stream(tokenizer,
query, query,
history=history, history=history,
@@ -493,8 +493,8 @@ async def predict(
**gen_kwargs) **gen_kwargs)
for _new_response in response_generator: for _new_response in response_generator:
if len(_new_response) <= delay_token_num: if len(_new_response) <= delay_token_num:
continue continue
new_response = _new_response[:-delay_token_num] new_response = _new_response[:-delay_token_num] if delay_token_num else _new_response
if len(new_response) == current_length: if len(new_response) == current_length:
continue continue