minor updates for hf multiple max_tokens fix

.chainlit/config.toml

@@ -7,7 +7,7 @@ enable_telemetry = true
 user_env = []
 
 # Duration (in seconds) during which the session is saved when the connection is lost
-session_timeout = 3600
+session_timeout = 7200  # 2 hours
 
 # Duration (in seconds) of the user session expiry
 user_session_timeout = 1296000  # 15 days
@@ -18,6 +18,11 @@ cache = false
 # Authorized origins
 allow_origins = ["*"]
 
+# Enable WebSocket heartbeat to maintain stable connections
+[websocket]
+ping_interval = 25  # Send ping every 25 seconds
+ping_timeout = 120  # Longer timeout for better stability
+
 [features]
 # Process and display HTML in messages. This can be a security risk (see https://stackoverflow.com/questions/19603097/why-is-it-dangerous-to-render-user-generated-html-or-javascript)
 unsafe_allow_html = false

.gitignore

@@ -1,4 +1,8 @@
 .env
 .venv
-.chainlit
-__pycache__
+__pycache__
+# Keep chainlit config
+!.chainlit/
+!.chainlit/config.toml
+# Ignore translations
+/app/.chainlit/translations/

app.py

@@ -2,6 +2,7 @@ import os
 from dotenv import load_dotenv
 import chainlit as cl
 from openai import AsyncOpenAI
+import time
 
 # Load environment variables from .env file
 load_dotenv()
@@ -10,7 +11,7 @@ load_dotenv()
 DEFAULT_SETTINGS = {
     "model": "gpt-3.5-turbo",
     "temperature": 0.7,
-    "max_tokens": 500,
+    "max_tokens": 1000,
     "top_p": 1,
     "frequency_penalty": 0,
     "presence_penalty": 0,
@@ -28,15 +29,8 @@ async def start():
         if not api_key:
             raise ValueError("OPENAI_API_KEY environment variable is not set")
         
-        # Initialize OpenAI client
+        # Initialize OpenAI client (without test request)
         client = AsyncOpenAI(api_key=api_key)
-        # Test the API key with a simple request
-        await client.chat.completions.create(
-            model="gpt-3.5-turbo",
-            messages=[{"role": "system", "content": "Test"}],
-            max_tokens=5
-        )
-        
         cl.user_session.set("client", client)
         
         # Initialize message history with system prompt
@@ -55,36 +49,46 @@ async def start():
             content=f"⚠️ Configuration Error: {str(e)}\nPlease make sure OPENAI_API_KEY is set in the environment variables."
         ).send()
     except Exception as e:
-        await cl.Message(
-            content=f"⚠️ Error initializing chat: {str(e)}\nPlease check your API key and try again."
-        ).send()
+        error_msg = f"⚠️ Error: {str(e)}"
+        if "session" in str(e).lower():
+            error_msg = "⚠️ Session error. Please refresh the page and try again."
+        await cl.Message(content=error_msg).send()
+
+@cl.on_stop
+async def on_stop():
+    """Cleanup when the chat session ends"""
+    try:
+        cl.user_session.clear()
+    except Exception:
+        pass
 
+async def handle_error(error: Exception) -> str:
+    """Helper function to format error messages"""
+    if "session" in str(error).lower():
+        return "⚠️ Session error occurred. Please refresh the page and try again."
+    return f"⚠️ An error occurred: {str(error)}"
 
 @cl.on_message
 async def main(user_message: cl.Message):
     """
-    Process user messages and generate AI responses:
-    - Update message history with user input
-    - Call OpenAI API with current conversation context
-    - Stream the response back to the user
-    - Update message history with AI response
-    
-    Args:
-        user_message: The message sent by the user
+    Process user messages and generate AI responses
     """
-    # Retrieve session data
-    client = cl.user_session.get("client")
-    message_history = cl.user_session.get("message_history")
-    settings = cl.user_session.get("settings")
-    
-    # Add user message to history
-    message_history.append({"role": "user", "content": user_message.content})
-    
-    # Prepare response message with loading state
-    response_message = cl.Message(content="")
-    await response_message.send()
-    
     try:
+        # Retrieve session data
+        client = cl.user_session.get("client")
+        message_history = cl.user_session.get("message_history")
+        settings = cl.user_session.get("settings")
+        
+        if not client or not message_history or not settings:
+            raise ValueError("Session data not found. Please refresh the page.")
+        
+        # Add user message to history
+        message_history.append({"role": "user", "content": user_message.content})
+        
+        # Prepare response message with loading state
+        response_message = cl.Message(content="")
+        await response_message.send()
+        
         # Call OpenAI API to get response
         stream = await client.chat.completions.create(
             messages=message_history,
@@ -92,22 +96,36 @@ async def main(user_message: cl.Message):
             **settings
         )
         
-        # Stream the response
+        # Stream the response with buffering
         full_response = ""
+        buffer = ""
+        update_interval = 0.1  # Update every 100ms
+        last_update_time = 0
+        
         async for chunk in stream:
             if chunk.choices[0].delta.content:
-                content_chunk = chunk.choices[0].delta.content
-                full_response += content_chunk
+                buffer += chunk.choices[0].delta.content
+                current_time = time.time()
                 
-                # Update message in real-time
-                response_message.content = full_response
-                await response_message.update()
+                # Update UI when buffer reaches certain size or time has passed
+                if len(buffer) >= 50 or (current_time - last_update_time) >= update_interval:
+                    full_response += buffer
+                    response_message.content = full_response
+                    await response_message.update()
+                    buffer = ""  # Clear buffer
+                    last_update_time = current_time
+        
+        # Send any remaining buffer content
+        if buffer:
+            full_response += buffer
+            response_message.content = full_response
+            await response_message.update()
         
         # Add AI response to message history
         message_history.append({"role": "assistant", "content": full_response})
         cl.user_session.set("message_history", message_history)
         
     except Exception as e:
-        # Handle errors
-        response_message.content = f"Error: {str(e)}"
+        error_message = await handle_error(e)
+        response_message.content = error_message
         await response_message.update()