xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-31 06:36:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

migrate pydantic from v1 to v2 (#2531)

Browse Source
This commit is contained in:
Jannis Schönleber 2024-01-22 01:05:56 +01:00 committed by GitHub
parent d75c40734a
commit 71d63ed72e
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
7 changed files with 26 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
requirements-neuron.txt
View File

@ -5,5 +5,5 @@ torch-neuronx >= 2.1.0
neuronx-cc
fastapi
uvicorn[standard]
pydantic == 1.10.13 # Required for OpenAI server.
pydantic >= 2.0 # Required for OpenAI server.
aioprometheus[starlette]

2
requirements-rocm.txt
View File

@ -9,5 +9,5 @@ tokenizers>=0.15.0
transformers >= 4.36.0 # Required for Mixtral.
fastapi
uvicorn[standard]
pydantic == 1.10.13 # Required for OpenAI server.
pydantic >= 2.0 # Required for OpenAI server.
aioprometheus[starlette]

2
requirements.txt
View File

@ -8,5 +8,5 @@ transformers >= 4.36.0 # Required for Mixtral.
xformers == 0.0.23.post1 # Required for CUDA 12.1.
fastapi
uvicorn[standard]
pydantic == 1.10.13 # Required for OpenAI server.
pydantic >= 2.0 # Required for OpenAI server.
aioprometheus[starlette]

18
vllm/entrypoints/openai/api_server.py
View File

@ -106,7 +106,7 @@ app.add_route("/metrics", metrics) # Exposes HTTP metrics
@app.exception_handler(RequestValidationError)
async def validation_exception_handler(_, exc):
err = openai_serving_chat.create_error_response(message=str(exc))
return JSONResponse(err.dict(), status_code=HTTPStatus.BAD_REQUEST)
return JSONResponse(err.model_dump(), status_code=HTTPStatus.BAD_REQUEST)
@app.get("/health")
@ -118,7 +118,7 @@ async def health() -> Response:
@app.get("/v1/models")
async def show_available_models():
models = await openai_serving_chat.show_available_models()
return JSONResponse(content=models.dict())
return JSONResponse(content=models.model_dump())
@app.post("/v1/chat/completions")
@ -126,22 +126,28 @@ async def create_chat_completion(request: ChatCompletionRequest,
raw_request: Request):
generator = await openai_serving_chat.create_chat_completion(
request, raw_request)
if request.stream and not isinstance(generator, ErrorResponse):
if isinstance(generator, ErrorResponse):
return JSONResponse(content=generator.model_dump(),
status_code=generator.code)
if request.stream:
return StreamingResponse(content=generator,
media_type="text/event-stream")
else:
return JSONResponse(content=generator.dict())
return JSONResponse(content=generator.model_dump())
@app.post("/v1/completions")
async def create_completion(request: CompletionRequest, raw_request: Request):
generator = await openai_serving_completion.create_completion(
request, raw_request)
if request.stream and not isinstance(generator, ErrorResponse):
if isinstance(generator, ErrorResponse):
return JSONResponse(content=generator.model_dump(),
status_code=generator.code)
if request.stream:
return StreamingResponse(content=generator,
media_type="text/event-stream")
else:
return JSONResponse(content=generator.dict())
return JSONResponse(content=generator.model_dump())
if __name__ == "__main__":

7
vllm/entrypoints/openai/protocol.py
View File

@ -14,7 +14,7 @@ class ErrorResponse(BaseModel):
message: str
type: str
param: Optional[str] = None
code: Optional[str] = None
code: int
class ModelPermission(BaseModel):
@ -189,7 +189,7 @@ class CompletionStreamResponse(BaseModel):
created: int = Field(default_factory=lambda: int(time.time()))
model: str
choices: List[CompletionResponseStreamChoice]
usage: Optional[UsageInfo]
usage: Optional[UsageInfo] = Field(default=None)
class ChatMessage(BaseModel):
@ -229,5 +229,4 @@ class ChatCompletionStreamResponse(BaseModel):
created: int = Field(default_factory=lambda: int(time.time()))
model: str
choices: List[ChatCompletionResponseStreamChoice]
usage: Optional[UsageInfo] = Field(
default=None, description="data about request and response")
usage: Optional[UsageInfo] = Field(default=None)

11
vllm/entrypoints/openai/serving_chat.py
View File

@ -102,7 +102,7 @@ class OpenAIServingChat(OpenAIServing):
created=created_time,
choices=[choice_data],
model=model_name)
data = chunk.json(exclude_unset=True, ensure_ascii=False)
data = chunk.model_dump_json(exclude_unset=True)
yield f"data: {data}\n\n"
# Send response to echo the input portion of the last message
@ -125,7 +125,7 @@ class OpenAIServingChat(OpenAIServing):
created=created_time,
choices=[choice_data],
model=model_name)
data = chunk.json(exclude_unset=True, ensure_ascii=False)
data = chunk.model_dump_json(exclude_unset=True)
yield f"data: {data}\n\n"
# Send response for each token for each request.n (index)
@ -156,7 +156,7 @@ class OpenAIServingChat(OpenAIServing):
created=created_time,
choices=[choice_data],
model=model_name)
data = chunk.json(exclude_unset=True, ensure_ascii=False)
data = chunk.model_dump_json(exclude_unset=True)
yield f"data: {data}\n\n"
else:
# Send the finish response for each request.n only once
@ -178,9 +178,8 @@ class OpenAIServingChat(OpenAIServing):
model=model_name)
if final_usage is not None:
chunk.usage = final_usage
data = chunk.json(exclude_unset=True,
exclude_none=True,
ensure_ascii=False)
data = chunk.model_dump_json(exclude_unset=True,
exclude_none=True)
yield f"data: {data}\n\n"
finish_reason_sent[i] = True
# Send the final done message after all response.n are finished

6
vllm/entrypoints/openai/serving_completion.py
View File

@ -74,7 +74,7 @@ async def completion_stream_generator(
logprobs=logprobs,
finish_reason=finish_reason,
)
]).json(exclude_unset=True, ensure_ascii=False)
]).model_dump_json(exclude_unset=True)
yield f"data: {response_json}\n\n"
if output.finish_reason is not None:
@ -99,7 +99,7 @@ async def completion_stream_generator(
)
],
usage=final_usage,
).json(exclude_unset=True, ensure_ascii=False)
).model_dump_json(exclude_unset=True)
yield f"data: {response_json}\n\n"
yield "data: [DONE]\n\n"
@ -279,7 +279,7 @@ class OpenAIServingCompletion(OpenAIServing):
# When user requests streaming but we don't stream, we still need to
# return a streaming response with a single event.
if request.stream:
response_json = response.json(ensure_ascii=False)
response_json = response.model_dump_json()
async def fake_stream_generator() -> AsyncGenerator[str, None]:
yield f"data: {response_json}\n\n"