hello

Dockerfile

@@ -57,4 +57,4 @@ HEALTHCHECK --interval=30s --timeout=30s --start-period=90s --retries=3 \
     CMD curl -f http://localhost:7860/health || exit 1
 
 # Run the application
-CMD ["python", "app.py"]
+CMD ["python", "main.py"]

app.py → main.py

@@ -1,4 +1,5 @@
 import os
+import gradio as gr
 from fastapi import FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.staticfiles import StaticFiles
@@ -258,6 +259,148 @@ if Path("static").exists():
     app.mount("/static", StaticFiles(directory="static"), name="static")
 
 # Gradio interface
+def chat_with_bot(message, history, max_length, temperature, top_p):
+    """Gradio chat function with advanced parameters"""
+    if not message.strip():
+        return "Please enter a message."
+    
+    response_text, _ = generate_response(message.strip(), max_length, temperature, top_p)
+    return response_text
+
+def create_gradio_interface():
+    """Create enhanced Gradio interface"""
+    
+    # Custom CSS
+    css = """
+    .gradio-container {
+        font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
+        max-width: 1200px;
+        margin: 0 auto;
+    }
+    .chat-message {
+        font-size: 14px !important;
+        line-height: 1.4;
+    }
+    .gradio-chatbot {
+        height: 500px;
+    }
+    """
+    
+    # Create interface with advanced controls
+    with gr.Blocks(css=css, title="FastAPI Chatbot", theme=gr.themes.Soft()) as demo:
+        
+        gr.HTML("<h1 style='text-align: center; color: #2563eb;'>🤖 FastAPI Chatbot</h1>")
+        gr.HTML(f"<p style='text-align: center; color: #6b7280;'>Powered by {MODEL_NAME}</p>")
+        
+        with gr.Row():
+            with gr.Column(scale=3):
+                chatbot = gr.Chatbot(
+                    height=500,
+                    show_copy_button=True,
+                    bubble_full_width=False,
+                    avatar_images=("👤", "🤖")
+                )
+                
+                with gr.Row():
+                    msg = gr.Textbox(
+                        placeholder="Type your message here...",
+                        container=False,
+                        scale=4,
+                        max_lines=3
+                    )
+                    submit_btn = gr.Button("Send 📤", scale=1, variant="primary")
+                
+                with gr.Row():
+                    clear_btn = gr.Button("Clear Chat 🗑️", scale=1)
+                    retry_btn = gr.Button("Retry Last ↻", scale=1)
+            
+            with gr.Column(scale=1):
+                gr.HTML("<h3>Settings</h3>")
+                
+                max_length = gr.Slider(
+                    minimum=50,
+                    maximum=200,
+                    value=MAX_LENGTH,
+                    step=10,
+                    label="Max Response Length"
+                )
+                
+                temperature = gr.Slider(
+                    minimum=0.1,
+                    maximum=1.5,
+                    value=DEFAULT_TEMPERATURE,
+                    step=0.1,
+                    label="Temperature (Creativity)"
+                )
+                
+                top_p = gr.Slider(
+                    minimum=0.1,
+                    maximum=1.0,
+                    value=0.9,
+                    step=0.05,
+                    label="Top-p (Focus)"
+                )
+                
+                gr.HTML("<h4>Example Messages:</h4>")
+                examples = gr.Examples(
+                    examples=[
+                        ["Hello! How are you today?"],
+                        ["Tell me a joke"],
+                        ["What's your favorite hobby?"],
+                        ["Can you help me with a creative writing prompt?"],
+                        ["What do you think about technology?"]
+                    ],
+                    inputs=msg,
+                    label="Click to try:"
+                )
+        
+        # Event handlers
+        def respond(message, history, max_len, temp, top_p):
+            if not message.strip():
+                return history, ""
+            
+            # Add user message
+            history.append([message, None])
+            
+            # Generate bot response
+            bot_response = chat_with_bot(message, history, max_len, temp, top_p)
+            history[-1][1] = bot_response
+            
+            return history, ""
+        
+        def clear_chat():
+            return [], ""
+        
+        def retry_last(history, max_len, temp, top_p):
+            if not history:
+                return history
+            
+            last_user_msg = history[-1][0]
+            history[-1][1] = "Thinking..."
+            
+            # Regenerate response
+            bot_response = chat_with_bot(last_user_msg, history, max_len, temp, top_p)
+            history[-1][1] = bot_response
+            
+            return history
+        
+        # Wire up events
+        submit_btn.click(
+            respond,
+            [msg, chatbot, max_length, temperature, top_p],
+            [chatbot, msg]
+        )
+        
+        msg.submit(
+            respond,
+            [msg, chatbot, max_length, temperature, top_p],
+            [chatbot, msg]
+        )
+        
+        clear_btn.click(clear_chat, outputs=[chatbot, msg])
+        retry_btn.click(retry_last, [chatbot, max_length, temperature, top_p], chatbot)
+    
+    return demo
 
 def run_fastapi():
     """Run FastAPI server"""
@@ -283,10 +426,25 @@ def main():
     
     logger.info("✅ Model loaded successfully!")
     
+    # Create Gradio interface
+    logger.info("🎨 Creating Gradio interface...")
+    demo = create_gradio_interface()
+    
     # Start FastAPI server in a separate thread
     logger.info("🌐 Starting FastAPI server...")
     fastapi_thread = threading.Thread(target=run_fastapi, daemon=True)
     fastapi_thread.start()
+    
+    # Launch Gradio interface
+    logger.info("🚀 Launching Gradio interface...")
+    demo.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        share=False,
+        show_error=True,
+        quiet=False,
+        show_api=False
+    )
 
 if __name__ == "__main__":
     main()

requirements.txt

@@ -4,6 +4,7 @@ transformers==4.35.2
 torch==2.1.0
 tokenizers==0.15.0
 accelerate==0.24.1
+gradio==4.7.1
 requests==2.31.0
 numpy==1.24.3
 pydantic==2.4.2