Hello #C031M8RLC66! We just made a new release of neuralforecast with some pretty important changes: New features - New model: TimeXer, a Transformer-based model specifically designed to handle exogenous features - All losses compatible with all types of models (e.g. univariate/multivariate, direct/recurrent) OR appropriate protection added. - DistributionLoss now supports the use of

quantiles

in

predict

, allowing for easy quantile retrieval for all

DistributionLosses

. - Mixture losses (GMM, PMM and NBMM) now support learned weights for weighted mixture distribution outputs. - Mixture losses now support the use of

quantiles

in

predict

, allowing for easy quantile retrieval. - Improved stability of

ISQF

by adding softplus protection around some parameters instead of using

.abs

. - Unified API for any quantile or any confidence level during predict for both point and distribution losses. Enhancements - Improve docstrings of all models - Minor bug fix in TFT: we can omit specifying an RNN type and the static covariate encoder will still work. - Fitting with an invalid validation size now print a nice error message - Add bfloat16 support - Recurrent models can now produce forecasts recursively or directly. - IQLoss now gives monotonic quantiles - MASE loss now works Breaking Changes - Unify API - RMoK uses the

revin_affine

parameter instead of

revine_affine

. This was a typo in the previous version. - All models now inherit the

BaseModel

class. This changes how we implement new models in neuralforecast. - Recurrent models now require an

input_size

parameter. -

TCN

and

DRNN

are now window models, not recurrent models - We cannot load a recurrent model from a previous version to v3.0.0 Bug Fixes - Multivariate models do not error when predicting when

n_series

>

batch_size

- Insample prediction works with series of varying lengths Documentation - Big overhaul of the documentation to remove old and deprecated code. - Add example of modifying the default

configure_optimizers()

behavior (use of

ReduceLROnPlateau

scheduler) This release solves many of your pain points and it adds features that were aksed for a long time. Big thanks to @Olivier for his amazing contribution to this release, as well as to all our users for taking the time to raise issues and ask questions. We'll keep working on improving neuralforecast!

amazing! > IQLoss now gives monotonic quantiles Just confirming that MQLoss still does not guarantee monotonic quantiles? (wasn't sure if MQLoss was implied here) EDIT: found the relevant PR, looks like its just IQLoss!

Correct! MQLoss could be added in the future though, but given the additional complexity of possibly having few quantiles I was a bit hesitant of just doing a sort over the outputs (with IQLoss we can just resample many quantiles and sort them thereafter)

Hi @Marco! I have a question about adding historical exogenous data for forecasting—do you know when this feature will be available?

Historical exogenous has been supported for years, check the models that support "H" here.

yes but I am asking about TimeXER, and i fount this issue: https://github.com/Nixtla/neuralforecast/issues/1311, I have the same case

I didn't read TimeXer anywhere in your question (?) You can use another TF based model that supports historical exogenous for the time being, and you're free to implement the functionality yourself and submit a PR 🙂