Researchers have made significant progress in developing neural debuggers for Python by training large language models on execution traces. This approach enables the models to predict line-by-line execution of entire programs, effectively transforming them into neural interpreters. However, real-world debugging scenarios often involve setting breakpoints and stepping through code, rather than executing programs sequentially. To address this, a new neural debugger is being proposed, which would allow developers to leverage the capabilities of large language models to debug Python code more efficiently. The debugger would learn from execution traces and predict program behavior, facilitating more effective debugging. This development has the potential to revolutionize the way developers debug Python code, making the process more efficient and accurate. The introduction of such a debugger would be a significant step forward, enabling developers to tap into the power of large language models to improve their coding workflows1.