Closed JKSenthil closed 1 year ago
This pull request was exported from Phabricator. Differential Revision: D48340818
Merging #173 (c5bf807) into main (798037d) will decrease coverage by
0.07%
. The diff coverage is19.80%
.
@@ Coverage Diff @@
## main #173 +/- ##
==========================================
- Coverage 24.67% 24.61% -0.07%
==========================================
Files 173 173
Lines 9921 10016 +95
==========================================
+ Hits 2448 2465 +17
- Misses 7473 7551 +78
Files Changed | Coverage Δ | |
---|---|---|
torcheval/metrics/synclib.py | 14.18% <9.75%> (-3.47%) |
:arrow_down: |
torcheval/metrics/toolkit.py | 28.90% <18.18%> (-0.85%) |
:arrow_down: |
tests/metrics/test_synclib.py | 17.98% <28.57%> (+2.89%) |
:arrow_up: |
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
This pull request was exported from Phabricator. Differential Revision: D48340818
This pull request was exported from Phabricator. Differential Revision: D48340818
This pull request was exported from Phabricator. Differential Revision: D48340818
This pull request was exported from Phabricator. Differential Revision: D48340818
Summary: remove all dependencies to TorchTNT, especially to avoid any recursive dependency issues in OSS
also copied
all_gather_tensors
from torchtnt into torcheval synclib. This function can be further optimized, since it currently all_gathers to get all tensor sizes, but this is already needed to be done in the synclib as well, so an all_gather can be saved here. Will do in future diffDifferential Revision: D48340818