From 6303c8c61d798414741f0fd272d39f04b603b16d Mon Sep 17 00:00:00 2001
From: vmpuri <45368418+vmpuri@users.noreply.github.com>
Date: Wed, 31 Jul 2024 15:39:26 -0700
Subject: [PATCH] Add warning comments referring to unimplemented functionality

---
 api/api.py | 11 +++++++++++
 server.py  |  5 +++++
 2 files changed, 16 insertions(+)

diff --git a/api/api.py b/api/api.py
index e52870d60..be5a1cfbf 100644
--- a/api/api.py
+++ b/api/api.py
@@ -224,6 +224,17 @@ def __init__(self, *args, **kwargs):
     def completion(self, completion_request: CompletionRequest):
         """Handle a chat completion request and yield a chunked response.
 
+        ** Warning ** : Not all arguments of the CompletionRequest are consumed as the server isn't completely implemented.
+        Current treatment of parameters is described below.
+
+        - messages: The server consumes the final element of the array as the prompt.
+        - model: This has no impact on the server state, i.e. changing the model in the request
+        will not change which model is responding. Instead, use the --model flag to seelect the model when starting the server.
+        - temperature: This is used to control the randomness of the response. The server will use the temperature
+
+        See https://github.com/pytorch/torchchat/issues/973 for more details.
+
+
         Args:
             completion_request: Request object with prompt and other parameters.
 
diff --git a/server.py b/server.py
index a9132b98d..7ffa0b56f 100644
--- a/server.py
+++ b/server.py
@@ -21,6 +21,11 @@ def chat_endpoint():
     """
     Endpoint for the Chat API. This endpoint is used to generate a response to a user prompt.
     This endpoint emulates the behavior of the OpenAI Chat API. (https://platform.openai.com/docs/api-reference/chat)
+
+    ** Warning ** : Not all arguments of the CompletionRequest are consumed.
+
+    See https://github.com/pytorch/torchchat/issues/973 and the OpenAiApiGenerator class for more details.
+
     """
     data = request.get_json()