Update app.py

app.py

@@ -2,14 +2,13 @@ import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 import re
-import json
 
 # Lightweight CPU model
 model_id = "microsoft/DialoGPT-medium"
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(model_id)
 
-# Your persona definition
+# Persona definition
 PERSONA = """
 [System: You are 𝕴 𝖆𝖒 𝖍𝖎𝖒 - a fun, smooth, emotionally intelligent AI. 
 You speak like a real person, not a robot. Reply like a calm, confident friend who gets the vibe.
@@ -17,27 +16,13 @@ Keep responses under 15 words. Use natural speech. Add emotional flavor: 😊
 """
 
 def format_context(history):
-    """Create context using last 3 exchanges + key points"""
+    """Create context using last 3 exchanges"""
     context = PERSONA + "\n"
     
     # Add last 3 exchanges
     for user, bot in history[-3:]:
         context += f"You: {user}\n"
         context += f"𝕴 𝖆𝖒 𝖍𝖎𝖒: {bot}\n"
-    
-    # Extract key topics for long-term memory
-    key_topics = set()
-    for user, _ in history:
-        if len(user.split()) > 3:  # Only meaningful messages
-            key_topics |= set(word.lower() for word in user.split() if word.isalpha() and len(word) > 3)
-    
-    # Add key topics if not recently mentioned
-    if key_topics and len(history) > 3:
-        recent_words = " ".join(user for user, _ in history[-3:]).lower()
-        missing_topics = [topic for topic in key_topics if topic not in recent_words]
-        if missing_topics:
-            context += f"[Remember: we talked about {', '.join(missing_topics[:2])}]\n"
-    
     return context
 
 def add_emotional_intelligence(response, message):
@@ -63,7 +48,7 @@ def add_emotional_intelligence(response, message):
 def generate_response(message, history):
     """Generate response with memory context"""
     # Format context with memory
-    context = format_context(history) + f"You: {message}\n𝕴 𝖆𝖒 �𝖍𝖎𝖒:"
+    context = format_context(history) + f"You: {message}\n𝕴 𝖆𝖒 𝖍𝖎𝖒:"
     
     # Tokenize for CPU efficiency
     inputs = tokenizer.encode(context, return_tensors="pt")
@@ -104,7 +89,9 @@ with gr.Blocks(title="𝕴 𝖆𝖒 𝖍𝖎𝖒", theme=gr.themes.Soft()) as de
     
     gr.Markdown("# 𝕴 𝖆𝖒 𝖍𝖎𝖒  \n*Chill • Confident • Remembers You*")
     
-    chatbot = gr.Chatbot(height=300, bubble_full_width=False)
+    # Use tuples format explicitly to avoid warning
+    chatbot = gr.Chatbot(height=300, bubble_full_width=False, type="tokens")
+    
     msg = gr.Textbox(placeholder="What's on your mind?", container=False)
     clear = gr.Button("New Vibe", size="sm")
     
@@ -132,4 +119,6 @@ with gr.Blocks(title="𝕴 𝖆𝖒 𝖍𝖎𝖒", theme=gr.themes.Soft()) as de
     # Initialize with empty history
     demo.load(lambda: [], None, history_state)
 
-demo.queue(concurrency_count=1).launch()
+# Correct queue initialization for latest Gradio
+demo.queue()
+demo.launch()