3 ms·
Could you explain for a dum-dum?
by rrr_oh_man 2y ago
Could you explain for a dum-dum?
- karalala 2y agoResults of xlstm are promising but will need larger scale experiments. However they completely messed up benchmarking experiments for various RNN models which in their papers claim comparable and even better performance than base transformer.
- AIsore 2y agoThese experiments seem pretty large already though, no? How are you so sure they messed up benchmarking? Is the code out already?