Changelog¶
All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
[Unreleased]¶
…
0.0.5 (2021-05-20)¶
Fixed¶
Fix clamping on
MaskedAttention
andMultitokenAvgEmbed
to small value less than 1. That’s the proper behavior to re-scale attentions that sum to less than 1 and ignore ones that sum to 0. This was only causing a minor decrease in F1 score, though.
Added¶
Add –use_gpu option to CLI (before it would always use a GPU if available)
Colab notebooks (see README)
Conda compatibility (see README)
Changed¶
Simplify
fix_pool_weights
code. Same behavior.
0.0.4 (2021-04-20)¶
Fixed¶
Fixed
field_mask
onFieldsEmbedNet
by clamping values to 1. Before, this mask was multiplying field embeddings by the field length in tokens. Now, the correct behavior is implemented: multiply by 0 the empty fields, and by 1 the non-empty fields. This was only causing a minor decrease in F1 score, though.
0.0.3 (2021-04-20)¶
Added¶
Example on how to do pairwise matching of candidate pairs at
notebooks/End-to-End-Matching-Example.ipynb
.Enable return of
field_embedding_dict
fromBlockerNet
for assisting pairwise matching. Usereturn_field_embeddings
parameter.Enable return of attention scores for interpretation from
MultitokenAttentionEmbed
. Use_forward
method.
Changed¶
Use of
LayerNorm
inEntityAvgPoolNet
instead ofF.normalize
, it’s less “esoteric”.Zeroing of empty field embeddings in
FieldsEmbedNet
instead ofBlockerNet
.
0.0.2 (2021-04-06)¶
Added¶
Documentation.
example-data/
in repo.
Changed¶
Simpler API for validation and test.
Better naming of various API objects and methods.
Consider -1 in min_epochs since epochs start from 0.
Upgrade pytorch-metric-learning to 0.9.98.
0.0.1 (2021-03-30)¶
First release on PyPI.