Update app.py

app.py

@@ -28,22 +28,24 @@ tokenizer, model = load_model()
 
 def remove_think_tags(text):
     """
-    Remove <think>...</think> tags from text - METHOD 1
+    Remove <think>...</think> tags from text
     """
     cleaned_text = re.sub(r'<think>.*?</think>', '', text, flags=re.DOTALL)
     return cleaned_text.strip()
 
 def generate_response(message, history, temperature=0.7, max_length=256):
     """
-    Generate a response using Qwen3-0.6B with your specified method
+    Generate a response using Qwen3-0.6B
     """
     if tokenizer is None or model is None:
         return "⚠️ Model is not loaded properly. Please check the console logs."
     
     try:
-        # Convert history to messages format
+        # Convert history to messages format - LIMIT HISTORY to last 6 exchanges
         messages = []
-        for human_msg, assistant_msg in history:
+        recent_history = history[-6:] if history else []  # Keep only last 6 exchanges
+        
+        for human_msg, assistant_msg in recent_history:
             messages.extend([
                 {"role": "user", "content": human_msg},
                 {"role": "assistant", "content": assistant_msg}
@@ -52,7 +54,7 @@ def generate_response(message, history, temperature=0.7, max_length=256):
         # Add current message
         messages.append({"role": "user", "content": message})
         
-        # Apply chat template exactly as in your example
+        # Apply chat template
         inputs = tokenizer.apply_chat_template(
             messages,
             add_generation_prompt=True,
@@ -91,22 +93,22 @@ def generate_response(message, history, temperature=0.7, max_length=256):
 
 def chat_interface(message, history, temperature, max_length):
     """
-    Main chat interface function
+    Main chat interface function - FIXED to prevent infinite loops
     """
     if not message or not message.strip():
         return "", history or []
     
-    # Generate response
+    # Generate response with limited history context
     bot_response = generate_response(message, history or [], temperature, max_length)
     
-    # Update history
+    # Update history - this will naturally grow but we limit context in generation
     new_history = (history or []) + [[message, bot_response]]
     
     return "", new_history
 
 def clear_chat():
     """
-    Clear the chat history
+    Clear the chat history - COMPLETELY RESET
     """
     return []
 
@@ -129,7 +131,7 @@ def retry_last_response(history, temperature, max_length):
     
     return new_history
 
-# Enhanced CSS for beautiful UI with settings below chat
+# Enhanced CSS for beautiful UI
 custom_css = """
 .gradio-container {
     font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, sans-serif;
@@ -219,6 +221,7 @@ custom_css = """
     padding: 30px !important;
     margin: 0 !important;
     border-radius: 0 !important;
+    overflow-y: auto !important;
 }
 
 #chatbot .message {
@@ -447,6 +450,16 @@ custom_css = """
     background: #ed8936;
     box-shadow: 0 0 10px #ed8936;
 }
+
+.chat-history-info {
+    text-align: center;
+    color: #718096;
+    font-size: 14px;
+    padding: 10px;
+    background: #f7fafc;
+    border-radius: 10px;
+    margin: 10px 0;
+}
 """
 
 # Create the Gradio interface
@@ -464,7 +477,7 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
                 # Chatbot
                 with gr.Column(elem_classes="chatbot-wrapper"):
                     chatbot = gr.Chatbot(
-                        value=[["Hello! How can I assist you today? 😊", ""]],
+                        value=[],  # Start with empty chat
                         label="",
                         elem_id="chatbot",
                         show_copy_button=True,
@@ -493,7 +506,7 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
                             clear_btn = gr.Button("🗑️ Clear Chat", elem_classes="clear-btn")
                             retry_btn = gr.Button("🔄 Retry Last", elem_classes="retry-btn")
                             gr.HTML("""<div style="flex: 1; text-align: center; color: #666; font-size: 14px; padding: 12px;">
-                                ✨ Powered by Qwen3-0.6B • Think tags automatically removed
+                                ✨ Powered by Qwen3-0.6B • Context limited to 6 recent messages
                             </div>""")
         
         # Controls Section (Below Chat)
@@ -539,6 +552,7 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
                             
                             **Model:** Qwen3-0.6B ✅  
                             **Think Tags:** Auto-Removed ✅
+                            **Context:** Last 6 messages
                             
                             <div class="model-features">
                             **Key Features:**
@@ -581,25 +595,31 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
                     )
 
     # Event handlers
+    def handle_submit(message, history, temp, max_len):
+        """Wrapper function to handle chat submission"""
+        if not message or not message.strip():
+            return "", history or []
+        return chat_interface(message, history, temp, max_len)
+    
     submit_event = msg.submit(
-        chat_interface,
+        fn=handle_submit,
         inputs=[msg, chatbot, temperature, max_length],
         outputs=[msg, chatbot]
     )
     
     submit_btn.click(
-        chat_interface,
+        fn=handle_submit,
         inputs=[msg, chatbot, temperature, max_length],
         outputs=[msg, chatbot]
     )
     
     clear_btn.click(
-        clear_chat,
+        fn=clear_chat,
         outputs=[chatbot]
     )
     
     retry_btn.click(
-        retry_last_response,
+        fn=retry_last_response,
         inputs=[chatbot, temperature, max_length],
         outputs=[chatbot]
     )