[Suggestion] NBEATS modification for covariate datasets: NBEATSx was released

[pnmartinez]

Hi @jdb78 ,

Some days ago, a paper introducing "NBEATSx", a covariate-ready NBEATS was released (https://arxiv.org/abs/2104.05522).

The authors have published the implementation here: https://github.com/cchallu/nbeatsx.

I think it is something that you should definitely have in your radar, and maybe over time add it to the framework.

Cheers!

[jdb78]

Excellent :) Think we should get this into the repo. Do you want to give the PR a stab?

[pnmartinez]

Hi Jan, Same as in the other issue: throughout the week I'll take some time to open the PR. Cheers!

…

________________________________ De: Jan Beitner ***@***.***> Enviado: jueves, 29 de abril de 2021 13:30 Para: jdb78/pytorch-forecasting ***@***.***> Cc: Pablo ***@***.***>; Author ***@***.***> Asunto: Re: [jdb78/pytorch-forecasting] [Suggestion] NBEATS modification for covariate datasets: NBEATSx was released (#471) Excellent :) Think we should get this into the repo. Do you want to give the PR a stab? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#471 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AHEB2L2TYCUL7FTFRSKZWJDTLE7M7ANCNFSM43ZTGQMQ>.

[enertel]

Did this ever get in? We've done some work on using NBEATSx internally (forked from the paper's repo) and will be bringing in lightning-flash soon which uses pytorch-forecasting under the hood. Might be a good opportunity to replace the existing NBEATS backbone.