model/model_utils.py

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch

def load_model():
    model_name = "Salesforce/codet5-base"
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
    model.eval()

    device = "cuda" if torch.cuda.is_available() else "cpu"
    model.to(device)
    return tokenizer, model, device

def generate_explanation(prompt, tokenizer, model, device):
    inputs = tokenizer(prompt, return_tensors="pt", truncation=True).to(device)
    output = model.generate(
        **inputs,
        decoder_start_token_id=tokenizer.pad_token_id,
        max_new_tokens=256,
        temperature=0.7
    )
    return tokenizer.decode(output[0], skip_special_tokens=True)