keras-team / keras-hub

Pretrained model hub for Keras 3
Apache License 2.0
797 stars 243 forks source link

Port XLM-RoBERTa to follow Bert model changes #405

Closed mattdangerw closed 2 years ago

mattdangerw commented 2 years ago

We had a few larger changes to the bert API: https://github.com/keras-team/keras-nlp/pull/387 https://github.com/keras-team/keras-nlp/pull/390

As well as some minor cleanups: https://github.com/keras-team/keras-nlp/pull/397 https://github.com/keras-team/keras-nlp/pull/398 https://github.com/keras-team/keras-nlp/pull/399

We should mirror these where applicable to the XLMRoberta model code. Since we do not yet host pretrained weight for this model, we can just raise a NotImplementedError in from_preset for now.

jbischof commented 2 years ago

See #406 for example PR, thanks in advance!

abheesht17 commented 2 years ago

Taking this up since this depends on RoBERTa, anyway. Making changes in the same RoBERTa PR