[SCRIPT] Add ESIM for text matching

[vanewu]

Description

Build an Enhancing LSTM model for Natural Language Inference.
The model does not currently contain the TreeLSTM mentioned in the original paper. Training scripts and tests are in progress.

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

[hhexiy]

In my experiment using a standard 1 layer FFN is good enough. The current fc_encoder gives NaN results somehow.

[vanewu]

@hhexiy The reason for this is that I previously mistyped the softmax axis of weight2. Many bugs have been fixed in the new submission based on your comments and original papers. I will do a full test on the data mentioned in the paper later. Thank you very much.

[vanewu]

@hhexiy
Thanks for the comment. The model code is okay on a competition task. But it has not been run on public data. Testing and calibration work is in progress. I will check and test it according to your proposal, and ping you after the work is completed.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request head (ESIM@7bcd4ba). Click here to learn what that means.
The diff coverage is n/a.

[codecov]

Codecov Report

Merging #689 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #689   +/-   ##
=======================================
  Coverage   90.38%   90.38%           
=======================================
  Files          66       66           
  Lines        6378     6378           
=======================================
  Hits         5765     5765           
  Misses        613      613

[mli]

Job PR-689/2 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-689/2/index.html

[mli]

Job PR-689/3 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-689/3/index.html

[mli]

Job PR-689/4 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-689/4/index.html

[mli]

Job PR-689/5 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-689/5/index.html

[mli]

Job PR-689/6 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-689/6/index.html

[szha]

Some items left:

• report performance and upload log to dmlc/web-data
• update model zoo page to include the description of the model, the command, and link to the log

[eric-haibin-lin]

@szha move the remaining items to github issue to track ?