[Refactor] Refactor BERT with new data preprocessing

[zburning]

Description

New data preprocessing.
Refactor BERT squad script.
Add XLNet squad script.
Update & add corresponding results.

Checklist

Essentials

• PR's title starts with a category (e.g. [BUGFIX], [MODEL], [TUTORIAL], [FEATURE], [DOC], etc)
• Changes are complete (i.e. I finished coding on this PR)
• All changes have test coverage
• Code is well-documented

Changes

• Feature1, tests, (and when applicable, API doc)
• Feature2, tests, (and when applicable, API doc)

Comments

• If this change is a backward incompatible change, why must this change be made.
• Interesting edge cases to note here

[mli]

Job PR-1124/7 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-1124/7/index.html

[leezu]

@zburning please resolve the conflicts. As the Bert results match reported performance and some XLNet results still show a gap, how about removing the XLNet changes from this PR to unblock this PR?

[mli]

Job PR-1124/9 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-1124/9/index.html

[mli]

Job PR-1124/10 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-1124/10/index.html

[eric-haibin-lin]

I think we need to add an API nlp.data.list_datasets() to list all available datasets in gluonnlp. Otherwise every time a new model is added, we need to revise the choice list in the script..

[leezu]

Please delete scripts/bert/test-682b5d15.bpe. I think it's included by mistake.

[mli]

Job PR-1124/13 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-1124/13/index.html

[mli]

Job PR-1124/14 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-1124/14/index.html

[mli]

Job PR-1124/15 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-1124/15/index.html

[mli]

Job PR-1124/16 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-1124/16/index.html

[leezu]

Thank you!