The project's objective is to create a transformer-based model for code-to-code translation, focusing on converting Java code to JavaScript. This initiative leverages machine learning and natural language techniques to automate the process of translating code between programming languages. The chosen transformer-based model, a cutting-edge neural network architecture, has demonstrated significant success in various natural language processing tasks, particularly in code translation.

To achieve this, the project will employ a dataset comprising a parallel corpus of Java and JavaScript code provided by MuST-CoST. The success and performance of the model will be assessed using BLEU scores, a metric optimized for evaluating the quality of code translation. The anticipated outcome is a tool that streamlines the translation of Java code to JavaScript, benefiting developers seeking an efficient and accurate conversion process.