Blog DeepPavlov

Announcing DeepPavlov Library 0.15.0 release

2021-05-18 16:20 Products Updates Community
We are happy to share our first release in 2021, v0.15.0! 

Back in October 2020 we shared our perspective on 2020-2021 years and promised to begin releasing new components. While we’ve shipped some of them as examples since that time, this is a special release for us as we start a refactoring process of the DeepPavlov Library!

This release includes an update to an existing feature, as well as a number of smaller improvements and bugfixes. In addition to that, we’ve begun our first ever Spring Cleaning!

Improvements & Bug Fixes


  • Сomponent-based Config Requirements Generation
In order to simplify maintenance and dependency management of the provided components we’ve decided to create a central Requirements Registry that stores requirements information for all shipped components.

  • NER AutoModel feature
NER AutoModel feature allows you to use any (adopted by huggingface) transformer for sequence classification tasks including named-entity recognition. You can pick the suitable transformer by modifying TRANSFORMER variable in the corresponding configuration file.

  • Spring Cleaning: 748 added and 7830 removed code lines
In Q1, we went through a refactoring process of the DeepPavlov Library; starting with this release, we’ve deprecated several ML models, which will be no longer supported by us in this and future releases:
  1. bert_as_summarizer, 
  2. seq2seq_go_bot, 
  3. hyperparameter optimization by neural evolution, 
  4. all deeppavlov.deprecated components.
 
If you feel a need in one or more of these models we welcome community contributors and would be happy to see them being supported by our lovely community.

We encourage you to begin building your Conversational AI systems with our DeepPavlov Library on GitHub and let us know what you think! Feel free to test our BERT-based models by using our demo. And keep in mind that we have a dedicated forum, where any questions concerning the framework and the models are welcome.


Follow @deeppavlov on Twitter.