Update app.py

app.py

@@ -3,70 +3,57 @@ import gradio as gr
 from openai import OpenAI
 import os
 import time
-from typing import List, Tuple
 
 # ==================== Configuration ====================
 DEFAULT_SYSTEM_PROMPT = "You are DeepSeek-V4, an advanced AI assistant with strong reasoning capabilities. Provide accurate, helpful, and well-reasoned responses."
 
-# Reasoning effort mapping
 REASONING_EFFORT_MAP = {
     "Non-think": "minimal",
     "Think High": "high", 
     "Think Max": "maximum"
 }
 
-# Thinking type mapping
 THINKING_TYPE_MAP = {
     "Non-think": "disabled",
     "Think High": "enabled",
     "Think Max": "enabled"
 }
 
-# ==================== API Client Setup ====================
+# ==================== API Client ====================
 def get_client():
-    """Initialize DeepSeek API client"""
     api_key = os.environ.get('DEEPSEEK_API_KEY')
     if not api_key:
         raise ValueError(
             "⚠️ DEEPSEEK_API_KEY not found!\n\n"
-            "Please set your API key:\n"
-            "1. Get your key from: https://platform.deepseek.com/api_keys\n"
-            "2. Set environment variable:\n"
-            "   export DEEPSEEK_API_KEY='your-api-key-here'"
+            "Get your key: https://platform.deepseek.com/api_keys\n"
+            "Then set: DEEPSEEK_API_KEY=your-key-here"
         )
-    
-    return OpenAI(
-        api_key=api_key,
-        base_url="https://api.deepseek.com"
-    )
+    return OpenAI(api_key=api_key, base_url="https://api.deepseek.com")
 
 # ==================== Response Generation ====================
 def generate_response(
     message: str,
-    history: List[Tuple[str, str]],
-    thinking_mode: str = "Think High",
-    max_tokens: int = 4096,
-    temperature: float = 0.7,
-    top_p: float = 1.0,
-    system_prompt: str = DEFAULT_SYSTEM_PROMPT,
-    show_thinking: bool = True
+    history: list,
+    thinking_mode: str,
+    max_tokens: int,
+    temperature: float,
+    top_p: float,
+    system_prompt: str,
+    show_thinking: bool
 ):
-    """Generate response using DeepSeek API"""
     if not message.strip():
-        yield "", history, "", "Please enter a message."
+        yield history, "", "Please enter a message."
         return
     
     client = get_client()
     
-    # Build messages array
-    messages = [{"role": "system", "content": system_prompt}]
+    # Build messages from history (dictionaries)
+    api_messages = [{"role": "system", "content": system_prompt}]
     
-    for user_msg, assistant_msg in history:
-        messages.append({"role": "user", "content": user_msg})
-        if assistant_msg:
-            messages.append({"role": "assistant", "content": assistant_msg})
+    for msg in history:
+        api_messages.append({"role": msg["role"], "content": msg["content"]})
     
-    messages.append({"role": "user", "content": message})
+    api_messages.append({"role": "user", "content": message})
     
     reasoning_effort = REASONING_EFFORT_MAP.get(thinking_mode, "high")
     thinking_type = THINKING_TYPE_MAP.get(thinking_mode, "enabled")
@@ -74,10 +61,9 @@ def generate_response(
     try:
         start_time = time.time()
         
-        # Streaming call
         stream = client.chat.completions.create(
             model="deepseek-v4-pro",
-            messages=messages,
+            messages=api_messages,
             stream=True,
             max_tokens=max_tokens,
             temperature=temperature,
@@ -90,53 +76,66 @@ def generate_response(
         thinking_chunks = []
         
         for chunk in stream:
-            if chunk.choices[0].delta.content:
-                content_chunks.append(chunk.choices[0].delta.content)
+            delta = chunk.choices[0].delta
+            
+            if delta.content:
+                content_chunks.append(delta.content)
             
-            if hasattr(chunk.choices[0].delta, 'reasoning_content'):
-                if chunk.choices[0].delta.reasoning_content:
-                    thinking_chunks.append(chunk.choices[0].delta.reasoning_content)
+            if hasattr(delta, 'reasoning_content') and delta.reasoning_content:
+                thinking_chunks.append(delta.reasoning_content)
             
             current_content = ''.join(content_chunks)
             current_thinking = ''.join(thinking_chunks)
             
-            full_response = current_content
+            # Build response text
             if show_thinking and current_thinking:
-                full_response = f"[Thinking]\n{current_thinking}\n\n[Response]\n{current_content}"
+                response_text = f"**[Thinking]**\n{current_thinking}\n\n**[Response]**\n{current_content}"
+            else:
+                response_text = current_content
             
             elapsed = time.time() - start_time
-            status = f"🔄 Streaming... ({elapsed:.1f}s)"
             
-            yield "", history + [(message, full_response)], current_thinking, status
+            # Update history with current state
+            new_history = history.copy()
+            new_history.append({"role": "user", "content": message})
+            new_history.append({"role": "assistant", "content": response_text})
+            
+            yield new_history, current_thinking, f"🔄 Streaming... ({elapsed:.1f}s)"
         
-        # Final
-        end_time = time.time()
+        # Final response
         final_content = ''.join(content_chunks)
         final_thinking = ''.join(thinking_chunks)
         
-        full_response = final_content
         if show_thinking and final_thinking:
-            full_response = f"[Thinking]\n{final_thinking}\n\n[Response]\n{final_content}"
+            final_response = f"**[Thinking]**\n{final_thinking}\n\n**[Response]**\n{final_content}"
+        else:
+            final_response = final_content
         
-        status = f"✅ Done in {end_time - start_time:.2f}s"
-        yield "", history + [(message, full_response)], final_thinking, status
+        end_time = time.time()
+        
+        final_history = history.copy()
+        final_history.append({"role": "user", "content": message})
+        final_history.append({"role": "assistant", "content": final_response})
+        
+        yield final_history, final_thinking, f"✅ Done in {end_time - start_time:.2f}s"
         
     except Exception as e:
         error_msg = f"❌ Error: {str(e)}"
-        yield "", history + [(message, error_msg)], "", error_msg
+        new_history = history.copy()
+        new_history.append({"role": "user", "content": message})
+        new_history.append({"role": "assistant", "content": error_msg})
+        yield new_history, "", error_msg
 
 # ==================== Gradio Interface ====================
 def create_demo():
-    """Create the Gradio interface"""
-    
     with gr.Blocks(title="DeepSeek-V4 Pro Demo") as demo:
         
         # Header
         gr.Markdown("""
         # 🚀 DeepSeek-V4 Pro
-        **Towards Highly Efficient Million-Token Context Intelligence**
         
-        Powered by DeepSeek API • 1.6T Parameters • 49B Activated • 1M Context
+        **Million-Token Context Intelligence**
+        1.6T Parameters | 49B Activated | 1M Context Length
         """)
         
         with gr.Row():
@@ -147,12 +146,13 @@ def create_demo():
                 thinking_mode = gr.Radio(
                     choices=["Non-think", "Think High", "Think Max"],
                     value="Think High",
-                    label="🧠 Reasoning Mode"
+                    label="🧠 Reasoning Mode",
+                    info="Non-think: Fast | Think High: Analysis | Think Max: Deep reasoning"
                 )
                 
                 show_thinking = gr.Checkbox(
                     value=True,
-                    label="📝 Show Thinking Process"
+                    label="📝 Show thinking process"
                 )
                 
                 system_prompt = gr.Textbox(
@@ -161,42 +161,38 @@ def create_demo():
                     lines=3
                 )
                 
-                with gr.Accordion("🔧 Advanced", open=False):
+                with gr.Accordion("🔧 Advanced Parameters", open=False):
                     max_tokens = gr.Slider(64, 32768, value=4096, step=64, label="Max Tokens")
                     temperature = gr.Slider(0.0, 2.0, value=0.7, step=0.05, label="Temperature")
                     top_p = gr.Slider(0.0, 1.0, value=1.0, step=0.05, label="Top P")
-                
-                gr.Markdown("### 💡 Examples")
-                gr.Examples(
-                    examples=[
-                        "Explain quantum computing simply",
-                        "Write a Python Fibonacci function",
-                        "What's new in DeepSeek-V4?",
-                    ],
-                    inputs=gr.Textbox(label="Try an example", visible=False),
-                )
             
-            # Right - Chat
+            # Right - Chat interface
             with gr.Column(scale=2):
                 chatbot = gr.Chatbot(
                     label="💬 Chat with DeepSeek-V4 Pro",
-                    height=500
+                    height=500,
+                    type="messages"
                 )
                 
                 with gr.Accordion("🧠 Thinking Process", open=True):
-                    thinking_display = gr.Markdown("*Reasoning will appear here...*")
+                    thinking_display = gr.Textbox(
+                        label="Reasoning",
+                        value="*Waiting for input...*",
+                        lines=5,
+                        interactive=False
+                    )
                 
                 with gr.Row():
                     message_input = gr.Textbox(
-                        label="Message",
-                        placeholder="Type your message...",
+                        label="Your Message",
+                        placeholder="Type your message here...",
                         lines=2,
                         scale=9
                     )
                     send_btn = gr.Button("🚀 Send", variant="primary", scale=1)
                 
                 with gr.Row():
-                    clear_btn = gr.Button("🗑️ Clear", size="sm")
+                    clear_btn = gr.Button("🗑️ Clear Chat", size="sm")
                 
                 status_display = gr.Textbox(
                     label="Status",
@@ -207,53 +203,62 @@ def create_demo():
         # Footer
         gr.Markdown("""
         ---
-        [📦 Model Card](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro) | 
-        [🔑 Get API Key](https://platform.deepseek.com/api_keys) | 
-        [📚 Docs](https://platform.deepseek.com/docs)
+        🔑 [Get API Key](https://platform.deepseek.com/api_keys) | 
+        📦 [Model Card](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro) | 
+        📚 [API Docs](https://platform.deepseek.com/docs)
         """)
         
         # ==================== Event Handlers ====================
         
-        def process_message(
+        def on_send(
             message, history, thinking_mode, show_thinking,
             system_prompt, max_tokens, temperature, top_p
         ):
             if not message.strip():
-                yield message, history, "", "Please enter a message."
+                yield history, "*Waiting for input...*", "Please enter a message."
                 return
             
             if not os.environ.get('DEEPSEEK_API_KEY'):
-                error = "⚠️ Please set DEEPSEEK_API_KEY environment variable"
+                error_msg = {
+                    "role": "assistant",
+                    "content": "⚠️ **API Key Missing**\n\nPlease set `DEEPSEEK_API_KEY` environment variable.\nGet one: https://platform.deepseek.com/api_keys"
+                }
                 history = history or []
-                history.append((message, error))
-                yield "", history, "", "❌ API Key missing"
+                history.append({"role": "user", "content": message})
+                history.append(error_msg)
+                yield history, "*API Key not configured*", "❌ Error"
                 return
             
             history = history or []
             
-            for msg, hist, thinking, status in generate_response(
+            for hist, thinking, status in generate_response(
                 message, history, thinking_mode, max_tokens,
                 temperature, top_p, system_prompt, show_thinking
             ):
-                yield msg, hist, thinking, status
+                yield hist, thinking if thinking else "*No reasoning to display*", status
         
-        # Events
         send_btn.click(
-            fn=process_message,
+            fn=on_send,
             inputs=[message_input, chatbot, thinking_mode, show_thinking,
                     system_prompt, max_tokens, temperature, top_p],
-            outputs=[message_input, chatbot, thinking_display, status_display]
+            outputs=[chatbot, thinking_display, status_display]
+        ).then(
+            fn=lambda: "",
+            outputs=[message_input]
         )
         
         message_input.submit(
-            fn=process_message,
+            fn=on_send,
             inputs=[message_input, chatbot, thinking_mode, show_thinking,
                     system_prompt, max_tokens, temperature, top_p],
-            outputs=[message_input, chatbot, thinking_display, status_display]
+            outputs=[chatbot, thinking_display, status_display]
+        ).then(
+            fn=lambda: "",
+            outputs=[message_input]
         )
         
         clear_btn.click(
-            fn=lambda: ([], "*Reasoning will appear here...*", "✅ Cleared"),
+            fn=lambda: ([], "*Chat cleared*", "✅ Cleared"),
             outputs=[chatbot, thinking_display, status_display]
         )
     
@@ -272,18 +277,12 @@ if __name__ == "__main__":
         print("\n" + "=" * 50)
         print("⚠️  DEEPSEEK_API_KEY not found!")
         print("=" * 50)
-        print("Get one: https://platform.deepseek.com/api_keys")
-        print("Set it:  export DEEPSEEK_API_KEY='your-key'\n")
+        print("1. Get key: https://platform.deepseek.com/api_keys")
+        print("2. Set: export DEEPSEEK_API_KEY='your-key'\n")
     
     demo = create_demo()
-    
-    # Try launch without theme/css first (most compatible)
-    try:
-        demo.queue(max_size=50).launch(
-            server_name="0.0.0.0",
-            server_port=7860,
-            share=False
-        )
-    except TypeError:
-        # Fallback: simplest launch
-        demo.queue().launch()
+    demo.queue(max_size=50).launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        share=False
+    )