Skip to content
This repository has been archived by the owner on Feb 7, 2023. It is now read-only.

Releases: koaning/whatlies

0.7.0

27 May 06:01
f369d57
Compare
Choose a tag to compare

Upgraded the version numbers for all dependencies.

0.6.5

16 Nov 22:36
154c284
Compare
Choose a tag to compare

Added support for floret embeddings.

0.6.4

06 Sep 08:51
d2c487b
Compare
Choose a tag to compare

Fixed a bug that allowed NaNs to appear in the output of an embedding.

0.6.3

17 Mar 07:19
4601962
Compare
Choose a tag to compare

Brush and Hover charts added.

0.6.0

16 Feb 21:02
a6ed809
Compare
Choose a tag to compare

Remove transformers that are never used. Made UMAP, spacy and fasttext an optional dependency.

0.5.4

07 Oct 13:31
39d6bfa
Compare
Choose a tag to compare

Added Universal Sentence Encoder.
Removed ConveRT because it was yanked.

0.5.3

24 Sep 14:37
c444307
Compare
Choose a tag to compare

Added some minor docs improvements and warnings. The main feature of this release is a patch that fixes a broken ConveRT link.

0.5.2

14 Sep 12:04
a5c5587
Compare
Choose a tag to compare
  • Added tests for matplotlib and altair.
  • Added plot_3d, allowing you to make some 3d visualisations.
  • Added assign as a nicer alternative for add_property.
  • Added a citation to an research paper on this library.
  • Removed the "helper vectors" from our transformers.

0.5.1

04 Sep 10:59
7fc7d9e
Compare
Choose a tag to compare
  • Added a guide on debiasing on the docs.
  • You can now specify the plot axes and title.
  • We've changed the internal file structure to prevent people from importing modules.
  • We've added sensible default values to "plot_interactive". We now assume that you want to plot the first two elements of a vector.
# Before
emb.transform(Pca(2)).plot_interactive('pca_0', 'pca_1')
# After
emb.transform(Pca(2)).plot_interactive()

0.5.0

24 Aug 08:22
2b154fd
Compare
Choose a tag to compare
  • Added some robustness tot he matplotlib based arrow and scatter charts.
  • Started deprecating the plot_correlation method in favor of the new plot_distance and plot_similarity methods.