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

[Relay][Frontend][TF] Fix slice when begin or size is not Const #4372

Merged
merged 2 commits into from
Nov 21, 2019
Merged

[Relay][Frontend][TF] Fix slice when begin or size is not Const #4372

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e01f0d4
fix slice bug when input is param
lisiyuan643 Nov 19, 2019
4d5c015
use _infer_value rather than _infer_value_simulated
lisiyuan643 Nov 21, 2019
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
10 changes: 8 additions & 2 deletions python/tvm/relay/frontend/tensorflow.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -626,8 +626,14 @@ def _impl(inputs, attr, params):

def _slice():
def _impl(inputs, attr, params):
begin = _get_list_param(params, inputs[1])
size = _get_list_param(params, inputs[2])
try:
begin = _get_list_param(params, inputs[1])
except:
Copy link
Contributor

Choose a reason for hiding this comment

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

We should probably only catch the exceptions we expect to see here (like IndexError or AttributeError), and let the others go through.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Got it

begin = _infer_value_simulated(inputs[1], params).asnumpy()[0]
Copy link
Member

@yongwww yongwww Nov 19, 2019 •
edited
Loading

Choose a reason for hiding this comment

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

how about begin = _infer_value(inputs[1], params).asnumpy().tolist()?

Copy link
Contributor

Choose a reason for hiding this comment

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

I think _infer_value_simulated is better here. _infer_value assumes that all of the inputs leading up to inputs[1] are constant, _infer_value_simulated does not.

Copy link
Contributor

Choose a reason for hiding this comment

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

I'm not sure whether _infer_value_simulated should be used here. size parameter can be a data dependent expression. If we want to infer the correct value, probably we should enforce all variables in params. For symbolic begin/size slice, it will be covered in #4312.

Copy link
Member

@yongwww yongwww Nov 20, 2019 •
edited
Loading

Choose a reason for hiding this comment

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

_infer_value_simulated is not able to handle data dependent cases. Let's use _infer_value here, if _infer_value doesn't help, then we have to use dynamic strided_slice in pr 4312. @lsy643 let us know if _infer_value doesn't work for you

Copy link
Contributor Author

@lsy643 lsy643 Nov 21, 2019 •
edited
Loading

Choose a reason for hiding this comment

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

@yongwww _infer_value works for me. It seems _infer_value_simulated use _infer_value inside, so why prefer one over another?

Copy link
Contributor

Choose a reason for hiding this comment

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

_infer_value_simulated will create dummy data if some variables don't appear in params. In data dependent case, this will cause problem.

try:
size = _get_list_param(params, inputs[2])
except:
size = _infer_value_simulated(inputs[2], params).asnumpy()[0]
Copy link
Member

Choose a reason for hiding this comment

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

the same as begin

data_shape = attr['_input_shapes'][inputs[0]]
data_dim = len(data_shape)
end = size
Expand Down
15 changes: 15 additions & 0 deletions tests/python/frontend/tensorflow/test_forward.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2188,6 +2188,20 @@ def test_forward_transpose():
_test_forward_tranapose_axes_input((2, 3, 4, 5), (3, 0, 1, 2))


def _test_forward_slice_operation_input(input_value, begin_value, size_value):
input_data = np.array(input_value, dtype=np.float32)
with tf.Graph().as_default():
input_tensor = tf.placeholder(
shape=input_data.shape, dtype=input_data.dtype, name="input")
begin_tensor = tf.expand_dims(begin_value, axis=0)
size_tensor = tf.expand_dims(size_value, axis=0)
slice_tensor = tf.slice(input_tensor, begin_tensor, size_tensor, name='slice_output')
compare_tf_with_tvm([input_data], ['input:0'], 'slice_output:0')


def test_forward_slice():
_test_forward_slice_operation_input([1, 1], 0, 2)
Copy link
Member

Choose a reason for hiding this comment

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

add test cases to cover begin & size as tensor

Copy link
Contributor Author

@lsy643 lsy643 Nov 21, 2019 •
edited
Loading

Choose a reason for hiding this comment

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

@yongwww begin and size as tensor are already covered. In the _test_forward_slice_operation_input([1, 1], 0, 2) , the input 0 and input 2 will be changed into tensors after tf.expand_dimes

def _test_forward_slice_operation_input(input_value, begin_value, size_value):
    input_data = np.array(input_value, dtype=np.float32)
    with tf.Graph().as_default():
        input_tensor = tf.placeholder(
            shape=input_data.shape, dtype=input_data.dtype, name="input")
        begin_tensor = tf.expand_dims(begin_value, axis=0)
        size_tensor = tf.expand_dims(size_value, axis=0)
        slice_tensor = tf.slice(input_tensor, begin_tensor, size_tensor, name='slice_output')
        compare_tf_with_tvm([input_data], ['input:0'], 'slice_output:0')


def test_forward_ceil():
ishape = (1, 3, 10, 10)
inp_array = np.random.uniform(size=ishape).astype(np.float32)
Expand Down Expand Up @@ -2762,6 +2776,7 @@ def test_forward_add_n():
if __name__ == '__main__':

# Transforms
test_forward_slice()
test_forward_transpose()
test_forward_reshape()
test_forward_depthtospace()
Expand Down