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This repository has been archived by the owner on Nov 17, 2023.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

Backport #17002, #17068 and #17114 to 1.6 branch #17137

Merged
merged 3 commits into from
Dec 20, 2019
Merged

Backport #17002, #17068 and #17114 to 1.6 branch #17137

merged 3 commits into from
Dec 20, 2019

Commits on Dec 20, 2019

  1. Improve the speed of the pointwise fusion graph pass (apache#17114) 

    * Debug the long startup time

* Optimize backward fusion

* Figure out why the fusion pass is called twice

* Cleaning

* Small optimization
    @ptrendx
    ptrendx committed Dec 20, 2019
    Configuration menu
    Copy the full SHA
    be9b20c View commit details
    Browse the repository at this point in the history

  2. [BUGFIX] Fix trainer param order (apache#17068) 

    * fix trainer param order

* Update trainer.py

* Update trainer.py

* Update trainer.py
    @eric-haibin-lin @ptrendx
    eric-haibin-lin authored and ptrendx committed Dec 20, 2019
    Configuration menu
    Copy the full SHA
    67b5cc5 View commit details
    Browse the repository at this point in the history

  3. [reproducibility] multi_sum_sq review, AtomicAdd removal (apache#17002) 

    * Update multi_sum_sq to avoid AtomicAdd

* Add specific test for multi_sum_sq

* Add a determism test and lint issues

* better test for cheching op is deterministic

* Follow MXNet letters case format

* Reduce dimensions of tensors in the test
    @MoisesHer @ptrendx
    MoisesHer authored and ptrendx committed Dec 20, 2019
    Configuration menu
    Copy the full SHA
    7d20ac4 View commit details
    Browse the repository at this point in the history