Very interesting results. xLSTMs works as a vision backbone - especially for large images (as could be expected). arxiv.org/abs/2406.04303