app.py

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

# --------------------
# Load Base Model and LoRA Adapter
# --------------------
def load_model_and_adapter():
    base_model_name = "unsloth/Llama-3.2-3B-Instruct"  # Replace with your base model name
    adapter_repo = "Futuresony/future_ai_12_10_2024"   # Your Hugging Face LoRA repo
    
    # Load tokenizer and base model
    tokenizer = AutoTokenizer.from_pretrained(base_model_name)
    base_model = AutoModelForCausalLM.from_pretrained(
        base_model_name,
        torch_dtype=torch.float16,  # Use float16 for efficiency if GPU is available
        device_map="auto"           # Automatically map to GPU or CPU
    )
    
    # Load LoRA adapter
    model = PeftModel.from_pretrained(base_model, adapter_repo)
    model.eval()  # Set to evaluation mode
    return tokenizer, model

# Load the model and tokenizer once
tokenizer, model = load_model_and_adapter()

# --------------------
# Generate Response Function
# --------------------
def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    messages = [{"role": "system", "content": system_message}]
    
    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})
    
    messages.append({"role": "user", "content": message})

    # Prepare input prompt for generation
    prompt = "\n".join([f"{m['role']}: {m['content']}" for m in messages])
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

    # Generate response
    outputs = model.generate(
        **inputs,
        max_length=max_tokens,
        temperature=temperature,
        top_p=top_p,
        pad_token_id=tokenizer.eos_token_id,
        eos_token_id=tokenizer.eos_token_id
    )
    
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    response = response.split("assistant:")[-1].strip()  # Clean response
    return response

# --------------------
# Gradio Interface
# --------------------
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a helpful assistant.", label="System message"),
        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)"),
    ],
)

# --------------------
# Launch the Interface
# --------------------
if __name__ == "__main__":
    demo.launch()