Paper

CODE TRANSLATION USING SEQ2SEQ MODELS

While traditional methods and small pre-trained models like GPT-2 struggle with code translation, this study demonstrates that fine-tuning Mistral 7B alongside strategic data augmentation yields significantly higher-quality structural and semantic translations.