zhousiyi f926650c64 if AbstractFunction comparison succeed in NewContext, then the evaluator should use the same one, otherwise one of the evaluator will not be evaluated. ... if funcgraph or metafuncgraph call it recursively, then anf_node should be used as tracking_id to discriminate the first occurcance and the recursive occurance. add anf_node to PrimitiveAbstractClosure hash() to reduce cost of GetEvaluatorFor(). ignore the tracking_id to make cse work.		2020-08-03 07:57:07 +00:00
..
mindspore_test_framework	Rewrite tensor's __bool__ for pynative mode	2020-07-23 16:51:46 +08:00
perf_test	Add optimization pass to remove redundant Select, fix uninitiated parameter in test_bert_train.py script	2020-07-28 01:04:29 -04:00
st	!3803 add gpu klDivLoss op	2020-08-03 10:07:34 +08:00
ut	if AbstractFunction comparison succeed in NewContext, then the evaluator should use the same one, otherwise one of the evaluator will not be evaluated.	2020-08-03 07:57:07 +00:00
vm_impl	set output value for dynamic graph	2020-07-23 17:05:47 +08:00
.gitignore	initial version	2020-03-27 22:54:54 +08:00
CMakeLists.txt	initial version	2020-03-27 22:54:54 +08:00
__init__.py	initial version	2020-03-27 22:54:54 +08:00
dataset_mock.py	fix get daataset size error	2020-07-24 21:45:42 +08:00
ops_common.py	clean pylint	2020-05-18 16:42:35 +08:00
runtest.sh	initial version	2020-03-27 22:54:54 +08:00
summary_utils.py	add more ut and st for SummaryCollector	2020-06-30 17:02:03 +08:00
train_step_wrap.py	pylint waring clean	2020-05-13 11:30:27 +08:00