app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import os

# Check if the token is being accessed
hf_token = os.environ.get("HF_HOME", None)

# Load the model and tokenizer
model_name = "meta-llama/CodeLlama-7b-Python-hf"
model = AutoModelForCausalLM.from_pretrained(model_name, token=hf_token, torch_dtype="float16", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_name, token=hf_token)


def generate_code(prompt):
    batch_size = 10
    inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True, max_length=512)
    outputs = model.generate(inputs['input_ids'], max_length=512, num_return_sequences=batch_size)
    code = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return code

# Set up the Gradio interface
demo = gr.Interface(fn=generate_code, 
             inputs="text", 
             outputs="text", 
             title="CodeLlama 7B Model",
             description="Generate code with CodeLlama-7b-hf.").launch()