# generation_fast.py
import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

class CodeGenerator:
    def __init__(self, model_name="S-Dreamer/PyCodeT5"):
        self.tokenizer = AutoTokenizer.from_pretrained(model_name)
        self.model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        self.model.to(self.device)

    def generate_code(self, nl_input, max_length=512, num_beams=5, early_stopping=True):
        inputs = self.tokenizer(nl_input, return_tensors="pt").to(self.device)
        outputs = self.model.generate(
            **inputs,
            max_length=max_length,
            num_beams=num_beams,
            early_stopping=early_stopping,
            no_repeat_ngram_size=2, # Prevents repetition
            length_penalty=1.0, # Adjust length penalty
            temperature=1.0, # Adjust temperature for diversity
        )
        generated_code = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
        return generated_code

if __name__ == "__main__":
    generator = CodeGenerator()
    nl_input = "Write a Python function to reverse a string."
    generated_code = generator.generate_code(nl_input)
    print(generated_code)