app.py

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

# Load the LLaMA model and tokenizer
model_name = "meta-llama/Llama-3.2-3B-Instruct"  # Replace this with the actual model name

# Authenticate if needed
# from huggingface_hub import login
# login(token="your_huggingface_token")

# Load the tokenizer and model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,  # Use float16 for better performance on GPUs
    device_map="auto",  # Automatically allocate the model to available devices
)

# Function to generate text
def generate_text(prompt, max_length=150, temperature=0.7, top_p=0.95):
    inputs = tokenizer(prompt, return_tensors="pt")
    outputs = model.generate(
        inputs["input_ids"],
        max_length=max_length,
        temperature=temperature,
        top_p=top_p,
        no_repeat_ngram_size=2,
        num_return_sequences=1,
    )
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Gradio interface
def gradio_interface(prompt, max_length, temperature, top_p):
    return generate_text(prompt, max_length, temperature, top_p)

iface = gr.Interface(
    fn=gradio_interface,
    inputs=[
        gr.Textbox(lines=5, label="Prompt"),
        gr.Slider(50, 500, value=150, step=10, label="Max Length"),
        gr.Slider(0.1, 1.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Top-p"),
    ],
    outputs="text",
    title="LLaMA 3.2 Text Generator",
    description="Generate text using the LLaMA 3.2 model.",
)

# Launch Gradio interface
iface.launch(share=True)