This study systematically evaluated the performance of models not only on custom novel dataset but also on the publicly available Bangla handwritten character dataset called the Grapheme dataset. This research