We’re glad to announce a minor release of DeepPavlov Library v0.12.1. This release includes an update to an existing feature, as well as a number of smaller improvements and bugfixes.
- English Language support in Entity Linking: In our last release, we’ve shipped Entity Linking, a key component of KBQA, for Russian language. We’re happy to ship Entity Linking for English language. Just to remind you, this model helps you to map entities in a given text to those found in WikiData. Learn more about the model in docs.
Improvements & Bugfixes
- KBQA: Expanded a number of supported question templates in KBQA model, e.g., “Who is …”, “What is …”, “How old is …”, etc, as well as a number of bugfixes.
- Tutorials: Added link to Colab-based tutorial explaining how to configure Go-Bot using our preliminary support for the RASA DSL-based configs
- PyTorch Compatibility Improvements: Added support for using different versions of PyTorch for the shipping NVIDIA NeMo components and other DeepPavlov Library components
- Other Improvements: Added root mean square error calculation to the list of components
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 models by using our demo and our new experimental AI assistant built with DeepPavlov conversational AI stack - Dream. And keep in mind that we have a dedicated forum, where any questions concerning the framework and the models are welcome.