Skip to content

Releases: automl/DEHB

v0.1.1

01 Apr 17:15
f6f1992
Compare
Choose a tag to compare

[0.1.1] - 2024-04-01

Added

  • Improved logging and state saving
  • Checkpointing and restarting an optimization run (#31)
  • Clear communication, that warmstarting via tell is currently not supported
  • Add class specific random number generators for better reproducibility

Changes

  • Interface changes for run, removing unnecessary logging frequency parameters, since they have been moved to the constructor

v0.1.0

17 Feb 15:33
0ce86c6
Compare
Choose a tag to compare

Added

  • Configuration IDs to improve logging and reproducability (#62)
  • Ask and Tell interface (#36)
  • Examples for Ask and Tell interface

Changes

  • Interface changes (renamed budget to fidelity) for clearer interface

v0.0.7

23 Aug 10:07
Compare
Choose a tag to compare

[0.0.7] - 2023-08-23

Added

  • Support for Constant hyperparameters (#52)
  • Unittests to test conversion from vector to configuration
  • Landing page of documentation

v0.0.6

04 Aug 10:10
Compare
Choose a tag to compare

Version 0.0.6 - 2023-08-04

Added

  • Unittest and documentation setup via pytest and mkdocs respectively
  • Pre-commit pipeline for style-checking via ruff, mypy and black
  • Better logging for min_budget >= max_budget (#33)
  • CONTRIBUTING.md for future contributions

Fixed

  • Use of deprecated numpy method np.int in dehb setup (#41)
  • If condition for proper Client cleanup, since it was never True (#45)
  • Data leak in example 01_Optimizing_RandomForest_using_DEHB (#23)

Changed

  • README.md to feature badges