Added Horovod distributed backend (#1529)

* Initial commit of Horovod distributed backend implementation

* Update distrib_data_parallel.py

* Update distrib_data_parallel.py

* Update tests/models/test_horovod.py

Co-Authored-By: Jirka Borovec <Borda@users.noreply.github.com>

* Update tests/models/test_horovod.py

Co-Authored-By: Jirka Borovec <Borda@users.noreply.github.com>

* Fixed tests

* Added six

* tests

* Install tox for GitHub CI

* Retry tests

* Catch all exceptions

* Skip cache

* Remove tox

* Restore pip cache

* Remove the cache

* Restore pip cache

* Remove AMP

Co-authored-by: William Falcon <waf2107@columbia.edu>
Co-authored-by: Jirka Borovec <Borda@users.noreply.github.com>
Co-authored-by: J. Borovec <jirka.borovec@seznam.cz>
This commit is contained in:
Travis Addair
2020-04-22 17:39:08 -04:00
committed by GitHub
co-authored by Jirka Borovec William Falcon J. Borovec
parent 4d24032ea5
commit 7024177f7d
18 changed files with 597 additions and 27 deletions
+9 -2
View File
@@ -41,9 +41,15 @@ jobs:
if: runner.os == 'macOS'
run: |
brew install libomp # https://github.com/pytorch/pytorch/issues/20030
brew install openmpi # Horovod on macOS requires OpenMPI, Gloo not currently supported
- name: Setup Windows
if: runner.os == 'windows'
run: |
python -c "lines = [line for line in open('requirements-extra.txt').readlines() if not line.startswith('horovod')] ; open('requirements-extra.txt', 'w').writelines(lines)"
# TODO: remove after https://github.com/pytorch/pytorch/issues/32186 is resolved
- name: Setup Windows
- name: Setup Windows on Latest
if: runner.os == 'windows' && matrix.requires == 'latest'
run: |
python -c "req = open('requirements.txt').read().replace('torch>=1.1', 'torch<1.5') ; open('requirements.txt', 'w').write(req)"
@@ -75,11 +81,12 @@ jobs:
run: |
# python -m pip install --upgrade --user pip
pip install -r requirements.txt -U -f https://download.pytorch.org/whl/torch_stable.html -q
pip install -r ./tests/requirements.txt -q
HOROVOD_BUILD_ARCH_FLAGS="-mfma" pip install -r ./tests/requirements.txt -q
# pip install tox coverage
python --version
pip --version
pip list
shell: bash
- name: Tests
# env: