Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix exponential op #58029

Merged
merged 16 commits into from
Oct 26, 2023
Merged

fix exponential op #58029

Show file tree
Hide file tree
Changes from 7 commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
7621a5e
fix exponential op
xingmingyyj Oct 11, 2023
f6f3344
更新 op_compat.yaml
xingmingyyj Oct 11, 2023
546a461
fix parser
xingmingyyj Oct 14, 2023
221f686
Merge branch 'develop' into fix_exponential_op
xingmingyyj Oct 14, 2023
72b2232
更新 utils.cc
xingmingyyj Oct 14, 2023
0bf972d
fix_exponential_op
xingmingyyj Oct 16, 2023
ce68891
add flag FLAGS_FLAGS_NEW_IR_NO_CHECK
xingmingyyj Oct 24, 2023
8e6aedc
add flag FLAGS_FLAGS_NEW_IR_NO_CHECK
xingmingyyj Oct 24, 2023
dfe7f18
add flag FLAGS_FLAGS_NEW_IR_NO_CHECK
xingmingyyj Oct 24, 2023
3cc8c13
add flag FLAGS_FLAGS_NEW_IR_NO_CHECK
xingmingyyj Oct 24, 2023
1efb4d1
add flag FLAGS_FLAGS_NEW_IR_NO_CHECK
xingmingyyj Oct 24, 2023
e12c550
Update new_ir_op_test_no_check_list
xingmingyyj Oct 24, 2023
30b1c3f
Update op_test.py
xingmingyyj Oct 25, 2023
d0d968e
Update op_test.py
xingmingyyj Oct 25, 2023
dc48f14
Update CMakeLists.txt
xingmingyyj Oct 25, 2023
e356fab
fix getenv
xingmingyyj Oct 26, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions paddle/fluid/pir/dialect/operator/utils/utils.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@ const std::unordered_set<std::string> LegacyOpList = {
"pd_op.c_sync_calc_stream_",
"pd_op.c_sync_comm_stream_",
"pd_op.dpsgd",
"pd_op.exponential",
"pd_op.send_v2",
"pd_op.recv_v2",
"pd_op.c_allreduce_sum",
Expand Down
4 changes: 2 additions & 2 deletions paddle/phi/api/yaml/op_compat.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -987,8 +987,8 @@
extra :
attrs : [bool use_mkldnn = false, bool use_cudnn = false]

- op : exponential_
backward : exponential__grad
- op : exponential_ (exponential)
backward : exponential__grad (exponential_grad)
inputs :
x : X
outputs :
Expand Down
4 changes: 4 additions & 0 deletions test/legacy_test/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1387,6 +1387,10 @@ foreach(IR_OP_TEST ${NEW_IR_OP_TESTS})
set_tests_properties(
${IR_OP_TEST} PROPERTIES ENVIRONMENT
"FLAGS_NEW_IR_OPTEST_WHITE_LIST=True")
if(${IR_OP_TEST} STREQUAL "test_exponential_op")
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

没有扩展性，建议改为列表实现，类似 #55857

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

已修改

set_tests_properties(${IR_OP_TEST}
PROPERTIES ENVIRONMENT "FLAGS_NEW_IR_NO_CHECK=True")
endif()
else()
message(STATUS "NewIR OpTest: not found ${IR_OP_TEST} in legacy_test")
endif()
Expand Down
5 changes: 5 additions & 0 deletions test/legacy_test/op_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1439,6 +1439,8 @@ def _check_ir_output(self, place, program, feed_map, fetch_list, outs):
check_method = lambda x, y, z: np.testing.assert_allclose(
x, y, err_msg=z, atol=1e-6, rtol=1e-6
)
if os.getenv("FLAGS_FLAGS_NEW_IR_NO_CHECK", None):
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这个flag为什么和上面CMake里设置的不一样？而且他能生效？

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

已修改 这里感觉应该单测失败才对

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

我怀疑这种写法是不是有问题，它会不会总是生效的？可以验证下这个flag对别的单测是不是也起效果了

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

确实有问题,if os.getenv("FLAGS_NEW_IR_NO_CHECK", None)只会检查这个环境变量是否存在，如果是0的话也会执行if后面的逻辑，现在改为if os.getenv("FLAGS_NEW_IR_NO_CHECK", None)==“True”只有得到环境变量为“True”才会执行

check_method = lambda x, y, err_msg: None

for i in range(len(outs)):
check_method(
Expand Down Expand Up @@ -3373,6 +3375,9 @@ def _check_ir_grad_output(
x, y, err_msg=z, atol=1e-6, rtol=1e-6
)

if os.getenv("FLAGS_FLAGS_NEW_IR_NO_CHECK", None):
check_method = lambda x, y, err_msg: None

for i in range(len(new_gradients)):
check_method(
gradients[i],
Expand Down
1 change: 1 addition & 0 deletions test/white_list/new_ir_op_test_white_list
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -79,6 +79,7 @@ test_elementwise_mul_op
test_elementwise_pow_op
test_erfinv_op
test_expand_v2_op
test_exponential_op
test_eye_op
test_fill_any_op
test_fill_constant_batch_size_like
Expand Down