This repository implements a transformer neural network architecture for predicting the remaining useful life (RUL) of lithium-ion batteries. The model processes time-series battery degradation data to estimate how many charge cycles remain before battery failure. It uses PyTorch as the deep learning framework and applies the transformer architecture (originally from NLP) to this regression task. The implementation includes support for NASA and CALCE battery datasets, with optional dropout and noise regularization parameters for improved robustness.