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This repository has been archived by the owner on Nov 17, 2023.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

Multi-precision AdamW update op #14171

Merged
merged 8 commits into from
Feb 20, 2019
Merged

Conversation

eric-haibin-lin
Copy link
Member

@eric-haibin-lin eric-haibin-lin commented Feb 14, 2019

Description

Tested in gluon-nlp.

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

  • The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant JIRA issue created (except PRs with tiny changes)
  • Changes are complete (i.e. I finished coding on this PR)
  • All changes have test coverage:
  • Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
  • Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
  • Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
  • Code is well-documented:
  • For user-facing API changes, API doc string has been updated.
  • For new C++ functions in header files, their functionalities and arguments are documented.
  • For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
  • Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
  • To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

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

Ubuntu and others added 5 commits February 9, 2019 23:31
* softmax for fp16 with fp32 accumulator

* return AType in kernel

* add dtype

* kernel
This reverts commit 5869e0a.
@eric-haibin-lin
Copy link
Member Author

@ptrendx you might be interested

@ankkhedia
Copy link
Contributor

@eric-haibin-lin Thanks for the contribution!
@mxnet-label-bot add [pr-awaiting-review, operator]

@apeforest Could you please help review this PR

@marcoabreu marcoabreu added Operator pr-awaiting-review PR is waiting for code review labels Feb 15, 2019
@eric-haibin-lin eric-haibin-lin changed the title multi-precision AdamW update op [WIP] multi-precision AdamW update op Feb 16, 2019
@eric-haibin-lin
Copy link
Member Author

i'll add more tests.

@eric-haibin-lin eric-haibin-lin changed the title [WIP] multi-precision AdamW update op Multi-precision AdamW update op Feb 17, 2019
@eric-haibin-lin
Copy link
Member Author

@larroy @marcoabreu any idea why the test status is always pending?

@szha
Copy link
Member

szha commented Feb 19, 2019

@eric-haibin-lin what I hear is that when jenkins tries to update the status back to github, the DNS resolution may fail which causes the message to be dropped. I retriggered the test to hopefully fix the status.

@eric-haibin-lin eric-haibin-lin merged commit 349803c into apache:master Feb 20, 2019
jessr92 pushed a commit to jessr92/incubator-mxnet that referenced this pull request Feb 28, 2019
* mp adamw update

* Softmax fp16 (apache#201)

* softmax for fp16 with fp32 accumulator

* return AType in kernel

* add dtype

* kernel

* adamw with nan check

* add doc

* Revert "Softmax fp16 (apache#201)"

This reverts commit 5869e0a.

* add test

* more test for fp16

* skip update for rescale = 0
jessr92 pushed a commit to jessr92/incubator-mxnet that referenced this pull request Feb 28, 2019
* mp adamw update

* Softmax fp16 (apache#201)

* softmax for fp16 with fp32 accumulator

* return AType in kernel

* add dtype

* kernel

* adamw with nan check

* add doc

* Revert "Softmax fp16 (apache#201)"

This reverts commit 5869e0a.

* add test

* more test for fp16

* skip update for rescale = 0
jessr92 pushed a commit to jessr92/incubator-mxnet that referenced this pull request Feb 28, 2019
* mp adamw update

* Softmax fp16 (apache#201)

* softmax for fp16 with fp32 accumulator

* return AType in kernel

* add dtype

* kernel

* adamw with nan check

* add doc

* Revert "Softmax fp16 (apache#201)"

This reverts commit 5869e0a.

* add test

* more test for fp16

* skip update for rescale = 0
drivanov pushed a commit to drivanov/incubator-mxnet that referenced this pull request Mar 4, 2019
* mp adamw update

* Softmax fp16 (apache#201)

* softmax for fp16 with fp32 accumulator

* return AType in kernel

* add dtype

* kernel

* adamw with nan check

* add doc

* Revert "Softmax fp16 (apache#201)"

This reverts commit 5869e0a.

* add test

* more test for fp16

* skip update for rescale = 0
vdantu pushed a commit to vdantu/incubator-mxnet that referenced this pull request Mar 31, 2019
* mp adamw update

* Softmax fp16 (apache#201)

* softmax for fp16 with fp32 accumulator

* return AType in kernel

* add dtype

* kernel

* adamw with nan check

* add doc

* Revert "Softmax fp16 (apache#201)"

This reverts commit 5869e0a.

* add test

* more test for fp16

* skip update for rescale = 0
haohuanw pushed a commit to haohuanw/incubator-mxnet that referenced this pull request Jun 23, 2019
* mp adamw update

* Softmax fp16 (apache#201)

* softmax for fp16 with fp32 accumulator

* return AType in kernel

* add dtype

* kernel

* adamw with nan check

* add doc

* Revert "Softmax fp16 (apache#201)"

This reverts commit 5869e0a.

* add test

* more test for fp16

* skip update for rescale = 0
@eric-haibin-lin eric-haibin-lin deleted the mp-adam branch September 2, 2019 23:35
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
Operator pr-awaiting-review PR is waiting for code review
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants