Versions Compared

Old Version 2

changes.mady.by.user Ankit Khedia

Saved on

compared with

New Version 3

changes.mady.by.user Ankit Khedia

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  • Introduce a new Gluon “Fit” API that eliminates the need to code a training loop for simple model use cases, thus reduces manual errors and friction.
  • Support Fit API handlers that enable to customize the training loop for things like checkpointing, logging, early stopping and metrics inspired by Keras Callbacks.
  • Maintain backwards compatibility: the existing Gluon way to train a model will be supported and maintained - it is needed for complex models and full imperative control by the user.
  • The new Fit API will cover beginners use-cases including canonical CV and NLP models, full list is in appendix. For advanced users and complex models, the recommended path is to use the existing training loop.
  • Test coverage: 100% unit test coverage and 100% integration test coverage for the example models in Appendix .
  • Educate Gluon users via: (1) Blog post (2) Example (3) Tutorial

...