Skip to content

GitLab

Unverified Commit d75dc20f authored by Matthias Gerstgrasser's avatar Matthias Gerstgrasser Committed by GitHub
Browse files

Add finish_reason to OpenAI API (#446)

parent 690d162d
Hide whitespace changes
Inline Side-by-side
Showing with 7 additions and 5 deletions
python/sglang/srt/openai_api_adapter.py
View file @ d75dc20f
...@@ -144,7 +144,7 @@ async def v1_completions(tokenizer_manager, raw_request: Request): ...@@ -144,7 +144,7 @@ async def v1_completions(tokenizer_manager, raw_request: Request):
index=0, index=0,
text=delta, text=delta,
logprobs=logprobs, logprobs=logprobs,
finish_reason=None, finish_reason=content["meta_info"]["finish_reason"],
) )
chunk = CompletionStreamResponse( chunk = CompletionStreamResponse(
id=content["meta_info"]["id"], id=content["meta_info"]["id"],
...@@ -193,7 +193,7 @@ async def v1_completions(tokenizer_manager, raw_request: Request): ...@@ -193,7 +193,7 @@ async def v1_completions(tokenizer_manager, raw_request: Request):
index=0, index=0,
text=text, text=text,
logprobs=logprobs, logprobs=logprobs,
finish_reason=None, # TODO(comaniac): Add finish reason. finish_reason=ret["meta_info"]["finish_reason"],
) )
response = CompletionResponse( response = CompletionResponse(
id=ret["meta_info"]["id"], id=ret["meta_info"]["id"],
...@@ -273,7 +273,7 @@ async def v1_chat_completions(tokenizer_manager, raw_request: Request): ...@@ -273,7 +273,7 @@ async def v1_chat_completions(tokenizer_manager, raw_request: Request):
choice_data = ChatCompletionResponseStreamChoice( choice_data = ChatCompletionResponseStreamChoice(
index=0, index=0,
delta=DeltaMessage(role="assistant"), delta=DeltaMessage(role="assistant"),
finish_reason=None, finish_reason=content["meta_info"]["finish_reason"],
) )
chunk = ChatCompletionStreamResponse( chunk = ChatCompletionStreamResponse(
id=content["meta_info"]["id"], id=content["meta_info"]["id"],
...@@ -286,7 +286,9 @@ async def v1_chat_completions(tokenizer_manager, raw_request: Request): ...@@ -286,7 +286,9 @@ async def v1_chat_completions(tokenizer_manager, raw_request: Request):
delta = text[len(stream_buffer) :] delta = text[len(stream_buffer) :]
stream_buffer = text stream_buffer = text
choice_data = ChatCompletionResponseStreamChoice( choice_data = ChatCompletionResponseStreamChoice(
index=0, delta=DeltaMessage(content=delta), finish_reason=None index=0,
delta=DeltaMessage(content=delta),
finish_reason=content["meta_info"]["finish_reason"],
) )
chunk = ChatCompletionStreamResponse( chunk = ChatCompletionStreamResponse(
id=content["meta_info"]["id"], id=content["meta_info"]["id"],
...@@ -305,7 +307,7 @@ async def v1_chat_completions(tokenizer_manager, raw_request: Request): ...@@ -305,7 +307,7 @@ async def v1_chat_completions(tokenizer_manager, raw_request: Request):
choice_data = ChatCompletionResponseChoice( choice_data = ChatCompletionResponseChoice(
index=0, index=0,
message=ChatMessage(role="assistant", content=ret["text"]), message=ChatMessage(role="assistant", content=ret["text"]),
finish_reason=None, # TODO(comaniac): Add finish reason. finish_reason=ret["meta_info"]["finish_reason"],
) )
response = ChatCompletionResponse( response = ChatCompletionResponse(
id=ret["meta_info"]["id"], id=ret["meta_info"]["id"],
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment