Our model achieved a final training RMSE loss of 0.448, validation loss of 0.4418, and test loss of 0.4379, in

/ProteiNN achieves low RMSE losses (0.448 train, 0.4418 validation, 0.4379 test) indicating strong generalization