Text embedding modules in tensorflow hub
- Mentors
- Vojtech Bardiovsky
- Organization
- TensorFlow
Project's main goal is adding new techniques and tools for text embedding modules in TensorFlow Hub.It is divided into three parts:
- Improve the TensorFlow Hub experience by creating demos and tools for exporting text embeddings to non-English languages (also in TF2),
- Release a proof of concept ULMFit (Universal Language Model Fine Tuning for Text Classification) module,
- Add a tool for showing how embedding changes after fine-tuning a module.