Add an XLMRobertaMaskedLM task model

[mattdangerw]

In #653 we added a masked language modeling task for RoBERTa. We can make a similar change for the XLMRoberta model.

• Update XLMRobertaTokenizer to expect a mask token.
• Add a XLMRobertaMaskedLMPreprocessor preprocessor layer and tests.
• Add a XLMRobertaMaskedLM task model and tests.
• Update keras_nlp/models/__init__.py to export XLMRobertaMaskedLM and XLMRobertaMaskedLMPreprocessor.

[soma2000-lang]

I would love to work on this issue!

[soma2000-lang]

/assign

[kanpuriyanawab]

Hi @soma2000-lang are you still working on it? if not then I would take this one up !

[soma2000-lang]

@shivance yes ,I had also opened a pr for it,but had to close it for some reason.Will reopen another one.

[jayam30]

Hey!
I would like to take this issue up, if others are not @soma2000-lang

[soma2000-lang]

@jayam30 already opened a pr

[kanpuriyanawab]

Hi @mattdangerw , it's already been 1month+ since this issue has gone stale. Please assign me this issue, I'm already having a crack at it to see how challenging it is. So far the tests pass in local.

If you allow, I could open a PR. Thanks !

[soma2000-lang]

@shivance I am working on it but the tests are failing currently ,working on that.