from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
import gradio as gr
import torch

model_id = "AmiyendraOP/llama3-legal-finetuned"


tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, low_cpu_mem_usage=True, torch_dtype=torch.float32)

# Set device
device = 0 if torch.cuda.is_available() else -1

# Use the text-generation pipeline
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, device=device)

# Define a chat function
def chat(prompt):
    response = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.7)[0]["generated_text"]
    return response

# Launch Gradio app
gr.Interface(
    fn=chat,
    inputs=gr.Textbox(lines=4, placeholder="Enter legal question...", label="Your Question"),
    outputs=gr.Textbox(label="Response"),
    title="LLaMA 3 Legal Chatbot (Fine-tuned)"
).launch()