import gradio as gr
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# Load base model & tokenizer
base_model = "vilsonrodrigues/falcon-7b-instruct-sharded"
tokenizer = AutoTokenizer.from_pretrained(base_model)
model = AutoModelForCausalLM.from_pretrained(base_model, device_map="cpu", torch_dtype=torch.float32)

# Load LoRA adapter
adapter_path = "./model"
model = PeftModel.from_pretrained(model, adapter_path)

def generate_response(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to("cpu")
    with torch.no_grad():
        outputs = model.generate(**inputs, max_length=200)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Gradio Interface
interface = gr.Interface(
    fn=generate_response,
    inputs=gr.Textbox(label="Ask AI"),
    outputs=gr.Textbox(label="Answer"),
    title="Financial AI Chatbot",
    description="Fine-tuned Falcon 7B Model for QnA Finansial."
)

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