import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import os

# Get the value of the HF_TOKEN environment variable
token = os.environ.get('HF_TOKEN')

# Load model and tokenizer from Hugging Face
model_name = "iqrabatool/finetuned_LLaMA"

# Define a smaller subset of the model or load a smaller version if available
model = AutoModelForCausalLM.from_pretrained(model_name, token=token)
tokenizer = AutoTokenizer.from_pretrained(model_name, token=token)

def respond(message, system_message, max_tokens, temperature, top_p):
    # Generate response
    inputs = tokenizer(message, return_tensors="pt", max_length=max_tokens, truncation=True, padding=True)
    outputs = model.generate(**inputs, temperature=temperature, top_p=top_p)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

# Define simplified interface components
additional_inputs = [
    gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
    gr.Slider(minimum=1, maximum=512, value=256, step=1, label="Max new tokens"),  # Limit max tokens
    gr.Slider(minimum=0.1, maximum=2.0, value=0.7, step=0.1, label="Temperature"),  # Reduce temperature range
    gr.Slider(minimum=0.1, maximum=0.9, value=0.5, step=0.05, label="Top-p (nucleus sampling)"),  # Reduce top-p range
]

# Create the simplified ChatInterface
demo = gr.Interface(
    fn=respond,
    inputs=["text", "text", "number", "number", "number"],
    outputs="text",
    title="Health Bot",
    description="A simplified chatbot for health-related inquiries.",
    article="The Health Bot assists users with health-related questions and provides information based on a pre-trained language model.",
    examples=[["What are the symptoms of COVID-19?", "Health Bot: COVID-19 symptoms include..."]],
    additional_inputs=additional_inputs
)

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