Update app.py

app.py

@@ -8,19 +8,20 @@ def respond(
     max_tokens,
     temperature,
     top_p,
-    hf_token: gr.OAuthToken, # Gradio injects this if logged in
+    hf_token: gr.OAuthToken, 
 ):
-    # Ensure token exists (User must click Login)
+    # Security Check: Ensure user is logged in
     if not hf_token or not hf_token.token:
-        yield "⚠️ Please login using the button in the sidebar to access the @frusto360 AI."
+        yield "⚠️ Please **Login** using the button in the sidebar to talk to @frusto360 AI."
         return
 
-    # Use direct InferenceClient (more stable for custom models)
+    # Direct model access to avoid the 400/404 Router errors
     client = InferenceClient(model="Frusto/llama-3.2-1b-frusto360-final", token=hf_token.token)
 
     # 1. Manually build the Llama 3.2 Chat Template
     prompt = f"<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\n{system_message}<|eot_id|>"
     
+    # In Gradio 6, history is ALWAYS a list of dicts
     for msg in history:
         role = msg['role']
         content = msg['content']
@@ -31,14 +32,14 @@ def respond(
 
     response = ""
     try:
-        # 2. Use text_generation instead of chat_completion
+        # 2. Use text_generation for maximum stability
         for token in client.text_generation(
             prompt,
             max_new_tokens=max_tokens,
             stream=True,
             temperature=temperature,
             top_p=top_p,
-            stop=["<|eot_id|>"] # Stop generating at the end-of-turn token
+            stop=["<|eot_id|>"] 
         ):
             response += token
             yield response
@@ -46,14 +47,14 @@ def respond(
     except Exception as e:
         error_msg = str(e)
         if "503" in error_msg:
-            yield "⏳ Model is waking up... please wait 60 seconds and try again."
+            yield "⏳ Model is warming up on Hugging Face servers... please wait 60 seconds."
         else:
             yield f"❌ Error: {error_msg}"
 
 # --- UI Setup ---
+# REMOVED type="messages" to fix the TypeError
 chatbot = gr.ChatInterface(
     respond,
-    type="messages", # Ensures history is a list of dictionaries
     additional_inputs=[
         gr.Textbox(value="You are the @frusto360 AI assistant. Created by @frusto360.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
@@ -64,9 +65,9 @@ chatbot = gr.ChatInterface(
 
 with gr.Blocks(theme="glass") as demo:
     with gr.Sidebar():
-        gr.Markdown("### 🔐 Authentication")
+        gr.Markdown("### 🔐 @frusto360 Auth")
         gr.LoginButton()
-        gr.Markdown("Login to use your Hugging Face account permissions.")
+        gr.Markdown("Click login to authenticate with your HF account.")
     
     chatbot.render()