Update app.py

app.py

@@ -21,24 +21,24 @@ class XylariaChat:
         self.hf_token = os.getenv("HF_TOKEN")
         if not self.hf_token:
             raise ValueError("HuggingFace token not found in environment variables")
-        
+
         # Initialize the inference client with the Qwen model
         self.client = InferenceClient(
-            model="Qwen/QwQ-32B-Preview", 
+            model="Qwen/QwQ-32B-Preview",  # Using the specified model
             api_key=self.hf_token
         )
-        
+
         # Image captioning API setup
-        self.image_api_url = "https://api-inference.huggingface.co/models/microsoft/git-large-coco"
+        self.image_api_url = "https://api-inference.huggingface.co/models/Salesforce/blip-image-captioning-large"
         self.image_api_headers = {"Authorization": f"Bearer {self.hf_token}"}
-        
+
         # Initialize conversation history and persistent memory
         self.conversation_history = []
         self.persistent_memory = {}
-        
+
         # System prompt with more detailed instructions
         self.system_prompt = """You are a helpful and harmless assistant. You are Xylaria developed by Sk Md Saad Amin(india). You should think step-by-step."""
-    
+
     def store_information(self, key, value):
         """Store important information in persistent memory"""
         self.persistent_memory[key] = value
@@ -50,31 +50,31 @@ class XylariaChat:
 
     def reset_conversation(self):
         """
-        Completely reset the conversation history, persistent memory, 
+        Completely reset the conversation history, persistent memory,
         and clear API-side memory
         """
         # Clear local memory
         self.conversation_history = []
         self.persistent_memory.clear()
-        
-        # Reinitialize the client
+
+        # Reinitialize the client (not strictly necessary for the API, but can help with local state)
         try:
             self.client = InferenceClient(
-                model="Qwen/QwQ-32B-Preview", 
+                model="Qwen/QwQ-32B-Preview",
                 api_key=self.hf_token
             )
         except Exception as e:
             print(f"Error resetting API client: {e}")
-        
+
         return None  # To clear the chatbot interface
 
     def caption_image(self, image):
         """
         Caption an uploaded image using Hugging Face API
-        
+
         Args:
             image (str): Base64 encoded image or file path
-        
+
         Returns:
             str: Image caption or error message
         """
@@ -89,83 +89,83 @@ class XylariaChat:
                 if image.startswith('data:image'):
                     image = image.split(',')[1]
                 data = base64.b64decode(image)
-            # If image is a file-like object
+            # If image is a file-like object (unlikely with Gradio, but good to have)
             else:
                 data = image.read()
-            
+
             # Send request to Hugging Face API
             response = requests.post(
-                self.image_api_url, 
-                headers=self.image_api_headers, 
+                self.image_api_url,
+                headers=self.image_api_headers,
                 data=data
             )
-            
+
             # Check response
             if response.status_code == 200:
                 caption = response.json()[0].get('generated_text', 'No caption generated')
                 return caption
             else:
-                return f"Error captioning image: {response.text}"
-        
+                return f"Error captioning image: {response.status_code} - {response.text}"
+
         except Exception as e:
             return f"Error processing image: {str(e)}"
 
     def get_response(self, user_input, image=None):
         """
         Generate a response using chat completions with improved error handling
-        
+
         Args:
             user_input (str): User's message
             image (optional): Uploaded image
-        
+
         Returns:
             Stream of chat completions or error message
         """
         try:
             # Prepare messages with conversation context and persistent memory
             messages = []
-            
+
             # Add system prompt as first message
             messages.append(ChatMessage(
-                role="system", 
+                role="system",
                 content=self.system_prompt
-            ).to_dict()) # Convert to dictionary
-            
+            ).to_dict())
+
             # Add persistent memory context if available
             if self.persistent_memory:
                 memory_context = "Remembered Information:\n" + "\n".join(
                     [f"{k}: {v}" for k, v in self.persistent_memory.items()]
                 )
                 messages.append(ChatMessage(
-                    role="system", 
+                    role="system",
                     content=memory_context
-                ).to_dict()) # Convert to dictionary
-            
+                ).to_dict())
+
             # Convert existing conversation history to ChatMessage objects and then to dictionaries
             for msg in self.conversation_history:
                 messages.append(ChatMessage(
-                    role=msg['role'], 
+                    role=msg['role'],
                     content=msg['content']
-                ).to_dict()) # Convert to dictionary
-            
+                ).to_dict())
+
             # Process image if uploaded
             if image:
                 image_caption = self.caption_image(image)
                 user_input = f"Image description: {image_caption}\n\nUser's message: {user_input}"
-            
+
             # Add user input
             messages.append(ChatMessage(
-                role="user", 
+                role="user",
                 content=user_input
-            ).to_dict()) # Convert to dictionary
-            
+            ).to_dict())
+
             # Generate response with streaming
-            stream = self.client.chat.completions.create(
+            stream = self.client.chat_completion(
+                messages=messages,
                 model="Qwen/QwQ-32B-Preview",
-                messages=messages, # Send dictionaries
-                temperature=0.5,
-                max_tokens=10240,
-                top_p=0.7,
+                temperature=0.7,
+                max_tokens=16384,
+                top_p=0.9,
                 stream=True
             )
             
@@ -175,6 +175,25 @@ class XylariaChat:
             print(f"Detailed error in get_response: {e}")
             return f"Error generating response: {str(e)}"
 
+    def messages_to_prompt(self, messages):
+        """
+        Convert a list of ChatMessage dictionaries to a single prompt string.
+        
+        This is a simple implementation and you might need to adjust it 
+        based on the specific requirements of the model you are using.
+        """
+        prompt = ""
+        for msg in messages:
+            if msg["role"] == "system":
+                prompt += f"<|system|>\n{msg['content']}<|end|>\n"
+            elif msg["role"] == "user":
+                prompt += f"<|user|>\n{msg['content']}<|end|>\n"
+            elif msg["role"] == "assistant":
+                prompt += f"<|assistant|>\n{msg['content']}<|end|>\n"
+        prompt += "<|assistant|>\n"  # Start of assistant's turn
+        return prompt
+        
+    
     def create_interface(self):
         def streaming_response(message, chat_history, image_filepath):
             # Check if an image was actually uploaded
@@ -200,7 +219,7 @@ class XylariaChat:
                     if chunk.choices and chunk.choices[0].delta and chunk.choices[0].delta.content:
                         chunk_content = chunk.choices[0].delta.content
                         full_response += chunk_content
-
+                        
                         # Update the last message in chat history with partial response
                         updated_history[-1][1] = full_response
                         yield "", updated_history, None
@@ -232,8 +251,8 @@ class XylariaChat:
         .chatbot-container .message {
             font-family: 'Inter', sans-serif !important;
         }
-        .gradio-container input, 
-        .gradio-container textarea, 
+        .gradio-container input,
+        .gradio-container textarea,
         .gradio-container button {
             font-family: 'Inter', sans-serif !important;
         }