diff --git a/llama_cpp/server/app.py b/llama_cpp/server/app.py index c54e4eb5c..fed0a6deb 100644 --- a/llama_cpp/server/app.py +++ b/llama_cpp/server/app.py @@ -197,7 +197,36 @@ async def authenticate( @router.post( - "/v1/completions", summary="Completion", dependencies=[Depends(authenticate)] + "/v1/completions", + summary="Completion", + dependencies=[Depends(authenticate)], + response_model= Union[ + llama_cpp.CreateCompletionResponse, + str, + ], + responses={ + "200": { + "description": "Successful Response", + "content": { + "application/json": { + "schema": { + "anyOf": [ + {"$ref": "#/components/schemas/CreateCompletionResponse"} + ], + "title": "Completion response, when stream=False", + } + }, + "text/event-stream":{ + "schema": { + "type": "string", + "title": "Server Side Streaming response, when stream=True. " + + "See SSE format: https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events/Using_server-sent_events#Event_stream_format", # noqa: E501 + "example": """data: {... see CreateCompletionResponse ...} \\n\\n data: ... \\n\\n ... data: [DONE]""" + } + } + }, + } + }, ) @router.post( "/v1/engines/copilot-codex/completions", @@ -280,7 +309,33 @@ async def create_embedding( @router.post( - "/v1/chat/completions", summary="Chat", dependencies=[Depends(authenticate)] + "/v1/chat/completions", summary="Chat", dependencies=[Depends(authenticate)], + response_model= Union[ + llama_cpp.ChatCompletion, str + ], + responses={ + "200": { + "description": "Successful Response", + "content": { + "application/json": { + "schema": { + "anyOf": [ + {"$ref": "#/components/schemas/CreateChatCompletionResponse"} + ], + "title": "Completion response, when stream=False", + } + }, + "text/event-stream":{ + "schema": { + "type": "string", + "title": "Server Side Streaming response, when stream=True" + + "See SSE format: https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events/Using_server-sent_events#Event_stream_format", # noqa: E501 + "example": """data: {... see CreateChatCompletionResponse ...} \\n\\n data: ... \\n\\n ... data: [DONE]""" + } + } + }, + } + }, ) async def create_chat_completion( request: Request, pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy