app.py

from transformers import pipeline

# Load the pre-trained or fine-tuned model
model_pipeline = pipeline("text-generation", model="your_model_name")

# Function to generate responses
def respond(message, history=[], max_tokens=100, temperature=0.8, top_p=0.9):
    """
    Generates a meaningful response while avoiding repetition.
    
    Parameters:
    - message (str): User's input message.
    - history (list): List of previous (user, model) interactions.
    - max_tokens (int): Maximum number of tokens to generate.
    - temperature (float): Controls randomness (higher = more creative).
    - top_p (float): Controls nucleus sampling.

    Returns:
    - response (str): Generated response.
    """

    message = message.strip()
    
    # Prevent repetition by checking history
    if history and message.lower() == history[-1][0].lower():
        return "Tafadhali uliza swali tofauti."

    # Generate response
    response = model_pipeline(
        message, 
        max_length=max_tokens, 
        temperature=temperature, 
        top_p=top_p, 
        pad_token_id=50256  # Ensures clean output for some models
    )[0]["generated_text"].strip()

    # Ensure response is meaningful and not just repeating input
    if response.lower() == message.lower():
        response = "Samahani, siwezi kujibu hilo kwa sasa."

    # Store history for context (optional)
    history.append((message, response))

    return response

# Example conversation loop
if __name__ == "__main__":
    print("🤖 Chatbot Ready! Type 'quit' to exit.")
    chat_history = []
    
    while True:
        user_input = input("You: ")
        if user_input.lower() == "quit":
            print("Chatbot: Karibu tena!")
            break

        bot_response = respond(user_input, chat_history)
        print(f"Chatbot: {bot_response}")