AREnets-0.23.1

Changelog

Full Changelog

Implemented enhancements:

• Tutorial on how we embed model and attention #53

v0.23.1 (2023-01-17)

Full Changelog

Implemented enhancements:

• Colab experience -- watch output in csv instead of csv.gz #65
• JSONL data in examples -- reduced input jsonl for a simplicity purposes. #64
• PandasCsvReader -- consider any compression if exists by default #62
• Sample ID -- considering int type as well #61
• Saving ids for hidden states #60
• BasePredictProvider -- allow output customization [codalab and user experience] #57
• Examples has wrong entities annotations #54
• RCNN and BiLSTM with Z-Yang Self-Attention -- Attention Hidden sides might be a part of the config #51
• PZhou and ZYang attentions -- clarify that output is actually a rnn_output which reduces the time series #50
• PZhou and ZYang configs -- using different BasicLSTM cells by default. #49
• CNN with self attention #48
• ENTITY_VALUES, ENTITY_TYPES -- might be removed #46
• BaseIDProvider -- this class should not provide the connection of opinions and samples [Refactoring ids] #43

Fixed bugs:

• RCNN attention -- Incorrect parameter mentioning for backward LSTM #47

Closed issues:

• s_ind and t_ind might be optional as well #63
• id field is very annoing in usage when bagging is disabled [Feedback] #56
• 'sent-ind' and 'doc-id' don't required #45
• Tensorflow warnings #5

Merged pull requests:

• Merge with 0.23.1 #66 (nicolay-r)

* This Changelog was automatically generated by github_changelog_generator