app.py

import gradio as gr
from huggingface_hub import InferenceClient
from transformers import AutoTokenizer  # Import the tokenizer

# Use the appropriate tokenizer for your model.
tokenizer = AutoTokenizer.from_pretrained("HuggingFaceH4/zephyr-7b-beta")
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")

# Define a maximum context length (tokens).  Check your model's documentation!
MAX_CONTEXT_LENGTH = 4096  # Example: Adjust this based on your model!

# Read the default prompt from a file
with open("prompt.txt", "r") as file:
    nvc_prompt_template = file.read()

def count_tokens(text: str) -> int:
    """Counts the number of tokens in a given string."""
    return len(tokenizer.encode(text))

def truncate_history(history: list[tuple[str, str]], system_message: str, max_length: int) -> list[tuple[str, str]]:
    """Truncates the conversation history to fit within the maximum token limit.
    Args:
        history: The conversation history (list of user/assistant tuples).
        system_message: The system message.
        max_length: The maximum number of tokens allowed.
    Returns:
        The truncated history.
    """
    truncated_history = []
    system_message_tokens = count_tokens(system_message)
    current_length = system_message_tokens

    # Iterate backwards through the history (newest to oldest)
    for user_msg, assistant_msg in reversed(history):
        user_tokens = count_tokens(user_msg) if user_msg else 0
        assistant_tokens = count_tokens(assistant_msg) if assistant_msg else 0
        turn_tokens = user_tokens + assistant_tokens

        if current_length + turn_tokens <= max_length:
            truncated_history.insert(0, (user_msg, assistant_msg))  # Add to the beginning
            current_length += turn_tokens
        else:
            break  # Stop adding turns if we exceed the limit

    return truncated_history

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    """Responds to a user message, maintaining conversation history, using special tokens and message list."""
    formatted_system_message = nvc_prompt_template

    truncated_history = truncate_history(history, formatted_system_message, MAX_CONTEXT_LENGTH - max_tokens - 100)  # Reserve space for the new message and some generation

    messages = [{"role": "system", "content": formatted_system_message}]  # Start with system message
    for user_msg, assistant_msg in truncated_history:
        if user_msg:
            messages.append({"role": "user", "content": f"<|user|>\n{user_msg}</s>"})
        if assistant_msg:
            messages.append({"role": "assistant", "content": f"<|assistant|>\n{assistant_msg}</s>"})

    messages.append({"role": "user", "content": f"<|user|>\n{message}</s>"})

    response = ""
    try:
        for chunk in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            token = chunk.choices[0].delta.content
            response += token
            yield response
    except Exception as e:
        print(f"An error occurred: {e}")
        yield "I'm sorry, I encountered an error. Please try again."

# --- Gradio Interface ---
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value=nvc_prompt_template, label="System message", visible=False),
        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()