Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-22 16:17:57 +01:00
Code Issues Packages Projects Releases Wiki Activity

Prevent deadlocks in OpenAI API with simultaneous requests

Browse Source
This commit is contained in:
oobabooga 2023-11-08 20:55:39 -08:00
parent 21ed9a260e
commit effb3aef42
1 changed files with 12 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
extensions/openai/script.py
View File

@ -1,3 +1,4 @@
import asyncio
import json import json
import os import os
import traceback import traceback
@ -46,6 +47,9 @@ params = {
} }
streaming_semaphore = asyncio.Semaphore(1)
def verify_api_key(authorization: str = Header(None)) -> None: def verify_api_key(authorization: str = Header(None)) -> None:
expected_api_key = shared.args.api_key expected_api_key = shared.args.api_key
if expected_api_key and (authorization is None or authorization != f"Bearer {expected_api_key}"): if expected_api_key and (authorization is None or authorization != f"Bearer {expected_api_key}"):
@ -84,6 +88,7 @@ async def openai_completions(request: Request, request_data: CompletionRequest):
if request_data.stream: if request_data.stream:
async def generator(): async def generator():
async with streaming_semaphore:
response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy) response = OAIcompletions.stream_completions(to_dict(request_data), is_legacy=is_legacy)
for resp in response: for resp in response:
yield {"data": json.dumps(resp)} yield {"data": json.dumps(resp)}
@ -102,6 +107,7 @@ async def openai_chat_completions(request: Request, request_data: ChatCompletion
if request_data.stream: if request_data.stream:
async def generator(): async def generator():
async with streaming_semaphore:
response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy) response = OAIcompletions.stream_chat_completions(to_dict(request_data), is_legacy=is_legacy)
for resp in response: for resp in response:
yield {"data": json.dumps(resp)} yield {"data": json.dumps(resp)}