[Enhancement] One script to export bert for classification/regression/QA

[eric-haibin-lin]

Description

This PR unifies static_export_squad.py and static_export_base.py by introducing a "task" argument to export.py. Currently it supports: classification, regression and question_answering.

The PR also renames the StaticXXX classes to HybridXXX classes, in case we change the implementation with control flow operators to support variable length inputs in the future.

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

[codecov]

Codecov Report

❗ No coverage uploaded for pull request head (export@7815a53). Click here to learn what that means.
The diff coverage is n/a.

[codecov]

Codecov Report

Merging #705 into master will increase coverage by 0.06%.
The diff coverage is n/a.

@@            Coverage Diff            @@
##           master    #705      +/-   ##
=========================================
+ Coverage   89.23%   89.3%   +0.06%     
=========================================
  Files          65      65              
  Lines        5953    5991      +38     
=========================================
+ Hits         5312    5350      +38     
  Misses        641     641

Impacted Files	Coverage Δ
src/gluonnlp/data/sentiment.py	98.29% <0%> (+0.82%)	⬆️

[mli]

Job PR-705/1 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/1/index.html

[mli]

Job PR-705/2 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/2/index.html

[mli]

Job PR-705/3 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/3/index.html

[mli]

Job PR-705/4 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/4/index.html

[mli]

Job PR-705/5 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/5/index.html

[mli]

Job PR-705/6 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/6/index.html

[mli]

Job PR-705/7 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/7/index.html

[TaoLv]

Thank you for the contribution. I will have a try!

[mli]

Job PR-705/8 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/8/index.html

[mli]

Job PR-705/9 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/9/index.html

[xinyu-intel]

@eric-haibin-lin How to run classification model inference?

[gigasquid]

Thanks @eric-haibin-lin ! I'll help test the export base script.

[xinyu-intel]

max_length config

[TaoLv]

Thank you for the great effort, @eric-haibin-lin . It looks good to me. Now approved!

[mli]

Job PR-705/10 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-705/10/index.html

[xinyu-intel]

Thanks! HybridBert works well with both gluon and symbolic forward:)

[harplife]

Hi, is there any guide on how to use the export script? I would like to export a fine-tuned Bert model.

The code I'm using to train the model is this.