3 ms·

And like all deep learning forecasting models thus far, it makes for a nice paper but is not worth anyone using for a real problem. Much slower than the classic

And like all deep learning forecasting models thus far, it makes for a nice paper but is not worth anyone using for a real problem. Much slower than the classical methods it fails to beat.

p1esk 2y ago
That’s what people said about CV models in 2011.
- claytonjy 2y ago
  That's fair, but they stopped saying it about CV models in 2012. We've been saying this about foundational forecasting models since...2019 at least, probably earlier. But it is a harder problem!