@finlaydag33k that's the kind of stuff i mean when i say that it's a shame that GPUs are only used for lame neural net stuff instead for real parallel algorithms :)
@finlaydag33k
also the machine learning is always an approximation with _unknown uncertainty_. with other stochastic or heuristic algorithms or you could at least give an uncertainty value. here it's usually just "trust the machine". fine with enhancing images, not the thing i'd trust my life with or base decisions on.
@bonifartius Aye.
Machine Learning accelerators are fun for *very specific* workloads... But majority of the "regular consumers", don't really care about it beyond a buzzword.