app.py

import os
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "hugging-quants/Meta-Llama-3.1-8B-Instruct-GPTQ-INT4"
hf_token = os.environ.get("HF_TOKEN")

tokenizer = AutoTokenizer.from_pretrained(model_id, token=hf_token)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
    torch_dtype=torch.float16,
    low_cpu_mem_usage=True,
    token=hf_token
)

def generate(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(**inputs, max_new_tokens=128)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)