adding support for completions endpoint in proxy

litellm/main.py

@@ -1431,7 +1431,25 @@ def text_completion(*args, **kwargs):
         messages = [{"role": "system", "content": kwargs["prompt"]}]
         kwargs["messages"] = messages
         kwargs.pop("prompt")
-        return completion(*args, **kwargs)
+        response = completion(*args, **kwargs) # assume the response is the openai response object 
+        response_2 = {
+            "id": response["id"],
+            "object": "text_completion",
+            "created": response["created"],
+            "model": response["model"],
+            "choices": [
+            {
+                "text": response["choices"][0]["message"]["content"],
+                "index": response["choices"][0]["index"],
+                "logprobs": None,
+                "finish_reason": response["choices"][0]["finish_reason"]
+            }
+            ],
+            "usage": response["usage"]
+        }
+        return response_2
+    else:
+        raise ValueError("please pass prompt into the `text_completion` endpoint - `text_completion(model, prompt='hello world')`")
 
 ##### Moderation #######################
 def moderation(input: str, api_key: Optional[str]=None):

litellm/proxy/proxy_cli.py

@@ -7,7 +7,8 @@ load_dotenv()
 @click.option('--api_base', default=None, help='API base URL.')
 @click.option('--model', required=True, help='The model name to pass to litellm expects') 
 def run_server(port, api_base, model):
-    from .proxy_server import app, initialize
+    # from .proxy_server import app, initialize
+    from proxy_server import app, initialize
     initialize(model, api_base)
     try:
         import uvicorn

litellm/proxy/proxy_server.py

@@ -1,4 +1,10 @@
+import sys, os
+sys.path.insert(
+    0, os.path.abspath("../..")
+)  # Adds the parent directory to the system path
+
 import litellm
+print(litellm.__file__)
 from fastapi import FastAPI, Request
 from fastapi.responses import StreamingResponse
 import json
@@ -25,8 +31,21 @@ def model_list():
         object="list",
     )
 
-@app.post("/chat/completions")
+@app.post("/completions")
 async def completion(request: Request):
+    data = await request.json()
+    if (user_model is None):
+        raise ValueError("Proxy model needs to be set")
+    data["model"] = user_model
+    if user_api_base:
+        data["api_base"] = user_api_base
+    response = litellm.text_completion(**data)
+    if 'stream' in data and data['stream'] == True: # use generate_responses to stream responses
+            return StreamingResponse(data_generator(response), media_type='text/event-stream')
+    return response
+
+@app.post("/chat/completions")
+async def chat_completion(request: Request):
     data = await request.json()
     if (user_model is None):
         raise ValueError("Proxy model needs to be set")

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "litellm"
-version = "0.1.789"
+version = "0.1.790"
 description = "Library to easily interface with LLM API providers"
 authors = ["BerriAI"]
 license = "MIT License"