Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

Add support of plug and play fit_batch and evaluate_batch #16982

Merged
merged 10 commits into from
Dec 11, 2019

Conversation

liuzh47
Copy link
Contributor

@liuzh47 liuzh47 commented Dec 5, 2019

Description

Due to the potential estimator explosion issue introduced in issue #16930, we add support of plug and play fit_batch() and evaluate_batch() methods for the base estimator class. Concretely,
we define a new class BatchProcessor and use it to wrap the fit_batch() and evaluate_batch() methods. Our contribution can be summarized as below

  • Add a new class BatchProcessor to support plug and play fit_batch and evaluate_batch for existing estimators.
  • Add test case to test the plug and play fit_batch() and evaluate_batch() methods

Fix Issue #16930

@liuzh47 liuzh47 requested a review from szha as a code owner December 5, 2019 10:33
Copy link
Contributor

@leezu leezu left a comment

Choose a reason for hiding this comment

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

************* Module mxnet.gluon.contrib.estimator.estimator_model

python/mxnet/gluon/contrib/estimator/estimator_model.py:102:0: C0303: Trailing whitespace (trailing-whitespace)

python/mxnet/gluon/contrib/estimator/estimator_model.py:22:0: W0611: Unused import warnings (unused-import)

************* Module mxnet.gluon.contrib.estimator.estimator

python/mxnet/gluon/contrib/estimator/estimator.py:397:0: C0330: Wrong continued indentation (add 50 spaces).

                        batch, batch_axis)

                        ^                                                 | (bad-continuation)


@leezu
Copy link
Contributor

leezu commented Dec 10, 2019

@ptrendx, let's also consider this one if it's merged to master in time (expected to be merged in 12 hours after fixing the conflict)

@leezu leezu merged commit c82af38 into apache:master Dec 11, 2019
ptrendx pushed a commit to ptrendx/mxnet that referenced this pull request Dec 11, 2019
* Add support of plug and play fit_batch and evaluate_batch

* Add check for the validity of the estimator model

* Rename estimator model as batch processor

* Remove unused import

* Add documentation of the batch processor class

* refine the documentation of the batch processor

* Fix merge bugs

* fix bugs introduced during merge

* fix sanity check failures

* fix CI bugs
ptrendx added a commit that referenced this pull request Dec 11, 2019
* Include eval_net the validation model in the gluon estimator api (#16957)

* Include eval_net the validation model in the estimator api

* fix small issue

* Extend estimator.evaluate() to support event handlers (#16971)



* fix unittest failures for the new api interface

* Add comments in the code for readability

* Remove unused argument val_metrics

* merge changes with the master branch

* fix some regression errors

* fix bugs introduced in the merging phase

* Add support of plug and play fit_batch and evaluate_batch (#16982)

* Add support of plug and play fit_batch and evaluate_batch

* Add check for the validity of the estimator model

* Rename estimator model as batch processor

* Remove unused import

* Add documentation of the batch processor class

* refine the documentation of the batch processor

* Fix merge bugs

* fix bugs introduced during merge

* fix sanity check failures

* fix CI bugs

* Fix Gluon Estimator nightly test (#17042)
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.

3 participants