The feedback phase will be open until Aug, 2023. Comments are invited to the
pull request linked.
Summary:
The proposal introduces the DynamicEmbedding layer to the Keras ecosystem, providing a native solution for handling colossal-scale problems in recommendation systems. The proposed solution facilitates automatic vocabulary building and updates, and dynamic embedding updates corresponding to evolving input patterns and vocabulary changes.
Thanks for your review and participation!
Divya