Skip to content
This repository has been archived by the owner on Jan 15, 2024. It is now read-only.

[FEATURE]add XNLI task #717

Merged
merged 18 commits into from
May 28, 2019
Merged

[FEATURE]add XNLI task #717

merged 18 commits into from
May 28, 2019

Conversation

paperplanet
Copy link
Member

@paperplanet paperplanet commented May 21, 2019

Description

Reproduce XNLI task in chinese.
Data can be downloaded from https://ernie.bj.bcebos.com/task_data.tgz (https://github.com/PaddlePaddle/LARK/tree/develop/ERNIE)

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

  • reproduce XNLI Task in chinese, validation metrics:accuracy:0.7755

Comments

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

accuracy results:
dev set:
INFO:root:Best model at epoch 2. Validation metrics:accuracy:0.7827
INFO:root:Now we are doing testing on test with gpu(5).
INFO:root:Time cost=25.46s, throughput=197.05 samples/s

test set:
INFO:root:validation metrics:accuracy:0.7762
INFO:root:Time cost=25.31s, throughput=198.16 samples/s
INFO:root:Now we are doing testing on test with gpu(3).
INFO:root:Time cost=24.40s, throughput=205.59 samples/s

@paperplanet paperplanet requested a review from szha as a code owner May 21, 2019 04:02
@codecov
Copy link

codecov bot commented May 21, 2019

Codecov Report

❗ No coverage uploaded for pull request head (feature/add-xnli-task@ba5679f). Click here to learn what that means.
The diff coverage is n/a.

@codecov
Copy link

codecov bot commented May 21, 2019

Codecov Report

Merging #717 into master will not change coverage.
The diff coverage is n/a.

Impacted file tree graph

@@           Coverage Diff           @@
##           master     #717   +/-   ##
=======================================
  Coverage   90.56%   90.56%           
=======================================
  Files          65       65           
  Lines        6071     6071           
=======================================
  Hits         5498     5498           
  Misses        573      573

Copy link
Member

@eric-haibin-lin eric-haibin-lin left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nice work! Some comments:

scripts/bert/baidu_ernie_data.py Show resolved Hide resolved
scripts/bert/baidu_ernie_data.py Outdated Show resolved Hide resolved
scripts/bert/dataset.py Outdated Show resolved Hide resolved
scripts/bert/dataset.py Show resolved Hide resolved
scripts/bert/index.rst Outdated Show resolved Hide resolved
Copy link
Member

@eric-haibin-lin eric-haibin-lin left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the update. A few more comment.

scripts/bert/baidu_ernie_data.py Outdated Show resolved Hide resolved
scripts/bert/dataset.py Outdated Show resolved Hide resolved
scripts/bert/baidu_ernie_data.py Outdated Show resolved Hide resolved
scripts/bert/baidu_ernie_data.py Show resolved Hide resolved
scripts/bert/index.rst Outdated Show resolved Hide resolved
scripts/bert/baidu_ernie_data.py Outdated Show resolved Hide resolved
scripts/bert/baidu_ernie_data.py Outdated Show resolved Hide resolved
@mli
Copy link
Member

mli commented May 25, 2019

Job PR-717/13 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-717/13/index.html

@mli
Copy link
Member

mli commented May 25, 2019

Job PR-717/14 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-717/14/index.html

@mli
Copy link
Member

mli commented May 27, 2019

Job PR-717/15 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-717/15/index.html

@mli
Copy link
Member

mli commented May 27, 2019

Job PR-717/16 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-717/16/index.html

@mli
Copy link
Member

mli commented May 27, 2019

Job PR-717/17 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-717/17/index.html

@szha szha merged commit 87abcb5 into dmlc:master May 28, 2019
paperplanet added a commit to paperplanet/gluon-nlp that referenced this pull request Jun 9, 2019
* add XNLI task

* update XNLI train cmd in index.rst

* update XNLI train cmd in index.rst

* rename label_project to label_alias

* add docstring for label_alias

* automatically download baidu ernie task data

* add some comments

* update xnli in index.rst

* fix xnli data comments

* update train command for xnli data

* fix xnli test case

* fix xnli test case

* fix xnli test case

* fix lint warning

* try to fix baidu_ernie_data import issue

* try to fix baidu_ernie_data import issue

* update XNLI finetune parameters in index.rst
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants