Update app.py

app.py

@@ -2,13 +2,30 @@ import gradio as gr
 from huggingface_hub import InferenceClient
 import os
 
+# Get HF token from environment variable
 HF_TOKEN = os.getenv("HF_TOKEN")
-if HF_TOKEN:
-    client = InferenceClient("HuggingFaceH4/zephyr-7b-beta", token=HF_TOKEN)
-else:
-    client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.1")
 
-#Dynamic prompt builder based on CEFR level
+# Initialize client with proper error handling
+def get_client():
+    if HF_TOKEN:
+        try:
+            # Try with the preferred model first
+            return InferenceClient("HuggingFaceH4/zephyr-7b-beta", token=HF_TOKEN)
+        except Exception as e:
+            print(f"Failed to initialize zephyr model: {e}")
+            # Fallback to mistral with token
+            try:
+                return InferenceClient("mistralai/Mistral-7B-Instruct-v0.1", token=HF_TOKEN)
+            except Exception as e2:
+                print(f"Failed to initialize mistral model: {e2}")
+                return None
+    else:
+        print("No HF_TOKEN found. Please set your Hugging Face token.")
+        return None
+
+client = get_client()
+
+# Dynamic prompt builder based on CEFR level
 def level_to_prompt(level):
     return {
         "A1": "You are a friendly French tutor. Speak mostly in English, use simple French, and explain everything.",
@@ -19,7 +36,7 @@ def level_to_prompt(level):
         "C2": "You are a native French professor. Speak in rich, complex French. Avoid English."
     }.get(level, "You are a helpful French tutor.")
 
-#Custom background CSS
+# Custom background CSS
 css = """
 @import url('https://fonts.googleapis.com/css2?family=Noto+Sans+JP&family=Playfair+Display&display=swap');
 body {
@@ -57,9 +74,13 @@ body {
 }
 """
 
-
-#Chat logic
+# Chat logic with proper error handling
 def respond(message, history, level, max_tokens, temperature, top_p):
+    # Check if client is available
+    if client is None:
+        yield "❌ Désolé! The AI service is not available. Please check your Hugging Face token configuration."
+        return
+    
     system_message = level_to_prompt(level)
     messages = [{"role": "system", "content": system_message}]
     
@@ -81,36 +102,98 @@ def respond(message, history, level, max_tokens, temperature, top_p):
     # Generate response
     response = ""
     try:
-        prompt = "\n".join([f"{m['role'].capitalize()}: {m['content']}" for m in messages])
-        for token in client.text_generation(prompt, max_new_tokens=max_tokens, stream=True, temperature=temperature, top_p=top_p):
-            response += token
-            yield response
-            token = msg.choices[0].delta.content
-            if token is not None:  # Handle None tokens
+        # Create a proper prompt format
+        prompt = ""
+        for msg in messages:
+            if msg["role"] == "system":
+                prompt += f"System: {msg['content']}\n"
+            elif msg["role"] == "user":
+                prompt += f"User: {msg['content']}\n"
+            elif msg["role"] == "assistant":
+                prompt += f"Assistant: {msg['content']}\n"
+        prompt += "Assistant: "
+        
+        # Generate response with streaming
+        for token in client.text_generation(
+            prompt, 
+            max_new_tokens=max_tokens, 
+            stream=True, 
+            temperature=temperature, 
+            top_p=top_p,
+            do_sample=True,
+            return_full_text=False
+        ):
+            if token:  # Handle None tokens
                 response += token
-            yield response
+                yield response
+                
     except Exception as e:
+        error_msg = str(e)
         print(f"Error in chat completion: {e}")
-        yield f"Désolé! There was an error: {str(e)}"
+        
+        if "401" in error_msg or "Unauthorized" in error_msg:
+            yield "🔑 Authentication Error: Please check your Hugging Face token. Make sure it's valid and has the correct permissions."
+        elif "429" in error_msg or "rate limit" in error_msg.lower():
+            yield "⏰ Rate limit exceeded. Please wait a moment before trying again."
+        elif "503" in error_msg or "Service Unavailable" in error_msg:
+            yield "🔧 The AI service is temporarily unavailable. Please try again later."
+        else:
+            yield f"❌ Désolé! There was an error: {error_msg}"
 
-#UI layout
-with gr.Blocks(css=css) as demo:
-    gr.Markdown("French Tutor", elem_id="custom-title")
+# UI layout
+with gr.Blocks(css=css, title="French Tutor") as demo:
+    gr.Markdown("# 🇫🇷 French Tutor", elem_id="custom-title")
+    
+    # Add status indicator
+    if client is None:
+        gr.Markdown("⚠️ **Warning**: No Hugging Face token found. Please set your HF_TOKEN environment variable.")
+    else:
+        gr.Markdown("✅ **Status**: Connected to AI service")
+    
     with gr.Column(elem_id="chat-panel"):
         with gr.Accordion("⚙️ Advanced Settings", open=False):
             level = gr.Dropdown(
                 choices=["A1", "A2", "B1", "B2", "C1", "C2"],
                 value="A1",
-                label="Your French Level (CEFR)"
+                label="Your French Level (CEFR)",
+                info="Choose your current French proficiency level"
+            )
+            max_tokens = gr.Slider(
+                1, 2048, 
+                value=512, 
+                step=1, 
+                label="Response Length",
+                info="Maximum number of tokens in the response"
+            )
+            temperature = gr.Slider(
+                0.1, 2.0, 
+                value=0.7, 
+                step=0.1, 
+                label="Creativity",
+                info="Higher values make responses more creative"
+            )
+            top_p = gr.Slider(
+                0.1, 1.0, 
+                value=0.95, 
+                step=0.05, 
+                label="Dynamic Text",
+                info="Controls text diversity"
             )
-            max_tokens = gr.Slider(1, 2048, value=512, step=1, label="Response Length")
-            temperature = gr.Slider(0.1, 4.0, value=0.7, step=0.1, label="Creativity")
-            top_p = gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Dynamic Text")
         
         gr.ChatInterface(
             fn=respond,
             additional_inputs=[level, max_tokens, temperature, top_p],
-            type="messages"  # ✅ prevents deprecation warning
+            type="messages",
+            title="Chat with your French Tutor",
+            description="Ask questions, practice conversation, or get help with French grammar!",
+            examples=[
+                "Bonjour! Comment allez-vous?",
+                "Can you help me conjugate the verb 'être'?",
+                "What's the difference between 'tu' and 'vous'?",
+                "Tell me about French culture",
+                "How do I order food in a French restaurant?"
+            ],
+            cache_examples=False
         )
 
 if __name__ == "__main__":