app.py

import gradio as gr
from huggingface_hub import InferenceClient
from collections import defaultdict

# Initialize the model client
client = InferenceClient("Futuresony/future_ai_12_10_2024.gguf")

# Store user preferences & history
user_preferences = defaultdict(int)  # Tracks keywords & topics
session_histories = defaultdict(list)  # Stores conversation history per session

def extract_keywords(text):
    """Extracts simple keywords from user input."""
    words = text.lower().split()
    common_words = {"the", "is", "a", "and", "to", "of", "in", "it", "you", "for"}  # Ignore common words
    return [word for word in words if word not in common_words]

def respond(message, history, system_message, max_tokens, temperature, top_p):
    session_id = id(history)  # Unique ID for each session
    session_history = session_histories[session_id]  # Retrieve session history
    
    # Extract keywords & update preferences
    keywords = extract_keywords(message)
    for kw in keywords:
        user_preferences[kw] += 1

    # Add past conversation to message history
    messages = [{"role": "system", "content": system_message}]
    for user_msg, bot_response in session_history:
        messages.append({"role": "user", "content": user_msg})
        messages.append({"role": "assistant", "content": bot_response})

    # Append current user message
    messages.append({"role": "user", "content": message})

    # Generate response from model
    response = ""
    for message in client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = message.choices[0].delta.content
        response += token
        yield response  # Stream response to user

    # Save to session history
    session_history.append((message, response))

    # Optionally, adapt responses based on learned preferences
    most_asked = max(user_preferences, key=user_preferences.get, default=None)
    if most_asked and most_asked in message.lower():
        response += f"\n\nI see you're interested in {most_asked} a lot! Want to explore more details?"
        yield response  # Update response with learning behavior

# Create Chat Interface
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a friendly chatbot that learns user interests.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
    ],
)

if __name__ == "__main__":
    demo.launch()