xlstm s/m model and mingru - minlstm for parallel concurrency or threads using stacked layers or columns like the cerebral cortex, parallel or distributed concurrent networks, let's try everything except attention :)