Olivier Sprangers comments

Results 191 comments of


                                            Olivier Sprangers

[DistributionLoss] Add a "return samples" option

> distributionloss generates samples paths that are then used by neuralforecast to compute and return quantiles. What I seek is the option to get them directly rather than the quantiles....

Query on forecast

@Mohan16071996 Apologies for the late reply. Confirming @khizer-kt answer - does that help you?

build(deps): bump the ci-dependencies group across 1 directory with 2 updates

@dependabot rebase

build(deps): bump the ci-dependencies group across 1 directory with 2 updates

@dependabot recreate

[FEAT] Reduce minimum required size for finetuning

> Can you add a small test finetuning with horizon + 1 samples? Check, good point, added

[FEAT] Reduce minimum required size for finetuning

Ray in conjunction with latest Pyarrow gives an error (see [this issue](https://github.com/ray-project/ray/issues/54722)). Fixed the pyarrow version to

[FEAT] Reduce minimum required size for finetuning

A lot of CI tests are failing, and we also are only testing for very old Python versions.

500 error code from API call

@tracykteal @jmoralez Is this still open / relevant?

Recommend a versatile and easy-to-implement loss function

Thanks for the suggestion! Are you (one of) the authors? I like the simplicity - as I understand your loss function is a weighted sum of the plain mean squared...

FEAT: Support exog features in TimeMixer

@JQGoh 1.7.4 was [already released today,](https://github.com/Nixtla/neuralforecast/releases/tag/v1.7.4) we took out the futr_exogenous functionality of `TimeMixer` for now, but we can make it part of the next release.