Skip to content

v1.13.0

Latest
Latest

Choose a tag to compare

View all tags
@gauchm gauchm released this 14 Jan 19:07
· 4 commits to master since this release
v1.13.0
74b6fb8
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Setup changes

As of #279, NeuralHydrology switched from using conda environments to uv. This has several advantages (e.g., it's much faster to create environments, and we'll be able to get up-to-date PyTorch packages after the pytorch conda channel deprecation).

You can continue to use existing conda environments, but for new setups we recommend using uv as described in the updated installation instructions.

Added

Model/training

  • Added xLSTM backbone to neuralhydrology (#259)
  • Added early stopping and dynamic learning rate scheduling
  • Feature to include embedding networks as model component in MTS-LSTM (#232)

Datasets

  • Added CAMELS IND to datasetzoo (#250)
  • Added CAMELS DE to datasetzoo (#253)
  • Added AORC hourly forcing support (#262)
  • Introduced dataset registry to allow datasets defined outside of neuralhydrology (#215)

Other

  • Added tests for uncertainty estimation (#238, #239)

Fixed

  • Fixed broken CAMELS links (#266, #267)
  • Fixed typehint in Basedataset collate_fn. (#246)
  • Fixed pandas warning and numpy 2.4 incompatibility (#244, #280)
  • Fix transformer typo for dropout config in samplingutils.py (#264)
  • Fixed samplingutils to work with new x_d logic. (#235)
  • Fixed continue run from last epoch when config file with run_dir is specified. (#236)
  • Fixed nh_results_ensemble for non-daily timescales (#268, #271)
Assets 2
Loading