Kashif Rasul

i will also add a `shift=1` if that is ok?

so the above error is due to the fact that the instance sampler cannot for reason sample windows of context and subsequent prediction length size from the dataset such that...

for which model? does this notebook work? https://github.com/zalandoresearch/pytorch-ts/blob/master/examples/Multivariate-Flow-Solar.ipynb

right that is as i mentioned because the new gluonts has changed its API so you can try to run it with an older version of gluonts or i need...

hmm ok let me fix it so that this works with the latest gluonts then...

Can you kindly downgrade gluonts as I have not updated this lib. to use the new APIs etc.

@hanlaoshi not sure if i understand, but what do you mean by " multi-scale structure" exactly? thanks!

you are not doing anything wrong. the fact is that the pygments rtf formatter only has options for `style` and and `fontface` and does not render rtf with line numbers....

I would suggest renaming the `train` method to something like `fit` as the logic of the `train` clashes with the `nn.Module`'s `train`

@KindXiaoming and @GentleFiona I believe your issues are not related to the problem raised in this issue, kindly open another issue as to not confound this problem with something else