experiments
/
STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Removing v1 optimizer tests from ctl_correctness_test to speed it up 

(v2, eager, v1_optimizer) combinations don't make much sense.

I've also tried reducing the batch size, but it doesn't seem to have any difference.

PiperOrigin-RevId: 309297422
Change-Id: I76fc456a8d94e0ace6dc5259104f706e3fe56dc6
This commit is contained in:
Ran Chen 2020-04-30 14:14:07 -07:00 committed by TensorFlower Gardener
parent 336c431b3b
commit e0701fcb76
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tensorflow/python/distribute/ctl_correctness_test.py
View File

@ -224,7 +224,7 @@ class TestDistributionStrategyDnnCorrectness(test.TestCase,
@combinations.generate( @combinations.generate(
combinations.combine( combinations.combine(
distribution=strategy_combinations.all_strategies, distribution=strategy_combinations.all_strategies,
optimizer_fn=strategy_combinations.optimizers_v1_and_v2, optimizer_fn=strategy_combinations.optimizers_v2,
mode=['eager'], mode=['eager'],
iteration_type=['iterator', 'dataset'], iteration_type=['iterator', 'dataset'],
inside_func=[False, True], inside_func=[False, True],