Skip to content

Backward function will set a flag if it released variables #21533

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

Closed
ifedan wants to merge 9 commits into from
Closed

Backward function will set a flag if it released variables #21533

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
8ba4ce8
Backward function will set a flag if it was released variables to avo…
ifedan Jun 7, 2019
119da75
Flake8 fix
ifedan Jun 7, 2019
65c17a3
Fix Flake8 issues
ifedan Jun 10, 2019
e86b768
Changed based on CR
ifedan Jun 13, 2019
240d724
Changed based on CR
ifedan Jun 13, 2019
fd60f2d
Changes based on CR
ifedan Jun 13, 2019
3dbcf3a
Changes based on CR
ifedan Jun 13, 2019
9b12aed
Changes based on CR
ifedan Jun 14, 2019
c338723
Added comments
ifedan Jun 18, 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
28 changes: 28 additions & 0 deletions test/test_autograd.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1221,6 +1221,34 @@ def test_backward_no_grad(self):
with self.assertRaises(RuntimeError):
torch.autograd.backward([b], [None])

def test_backward_twice_with_saved_values(self):
b = torch.randn(3, requires_grad=True, dtype=torch.double)
c = torch.zeros(3, dtype=torch.double)
c[[1, 2]] = b[[1, 1]]
Copy link
Contributor

@zou3519 zou3519 Jun 13, 2019

Choose a reason for hiding this comment

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

It's not completely obvious that this indexing expression causes a TensorList to be saved for backwards and that is what we're fixing in this PR (that backward twice with a saved TensorList works as expected). I don't see any other functions in derivatives.yaml that saves a TensorList, though, so I don't have ideas on how to make this test better.

c.backward(torch.tensor([1, 1, 1], dtype=torch.double))
self.assertRaisesRegex(RuntimeError, 'Specify retain_graph=True',
lambda: c.backward(torch.tensor([1, 1, 1], dtype=torch.double)))

def test_backward_twice_retained_graph_with_saved_values(self):
b = torch.randn(3, requires_grad=True, dtype=torch.double)
c = torch.zeros(3, dtype=torch.double)
c[[1, 2]] = b[[1, 1]]
c.backward(torch.tensor([1, 1, 1], dtype=torch.double), retain_graph=True)
c.backward(torch.tensor([1, 1, 1], dtype=torch.double))

def test_backward_twice_without_saved_values(self):
b = torch.randn(3, requires_grad=True, dtype=torch.double)
c = b + 1
c.backward(torch.tensor([1, 1, 1], dtype=torch.double))
c.backward(torch.tensor([1, 1, 1], dtype=torch.double))

def test_backward_twice_retained_graph_without_saved_values(self):
b = torch.randn(3, requires_grad=True, dtype=torch.double)
c = torch.zeros(3, dtype=torch.double)
c[[1, 2]] = b[[1, 1]]
c.backward(torch.tensor([1, 1, 1], dtype=torch.double), retain_graph=True)
c.backward(torch.tensor([1, 1, 1], dtype=torch.double))

def test_next_functions(self):
x = torch.randn(5, 5, requires_grad=True)
y = torch.randn(5, 5, requires_grad=True)
Expand Down
8 changes: 8 additions & 0 deletions tools/autograd/gen_autograd_functions.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@

FUNCTION_DEFINITION = CodeTemplate("""\
variable_list ${op}::apply(variable_list&& grads) {
${asserts}
IndexRangeGenerator gen;
${compute_index_ranges}
variable_list grad_inputs(gen.size());
Expand Down Expand Up @@ -126,6 +127,7 @@ def process_function(func):
release_variables = []
saved_list_sizes = []
unpack = []
asserts = []

env['compute_index_ranges'] = []
for arg in func['args_with_derivatives']:
Expand All @@ -146,8 +148,13 @@ def save_arg(arg, is_output):
unpack.append('auto {} = {}_.unpack({});'.format(name, name, ptr))
elif arg['type'] == 'TensorList':
saved_variables.append('std::vector<SavedVariable> {}_;'.format(name))
saved_variables.append('bool {}_released_ = false;'.format(name))
# Just clear() is sufficient, we don't need to loop and clear each variable.
# Because the SavedVariable owns a tensor and a grad_fn, removing the SavedVariable makes them go away as well.
release_variables.append('{}_.clear();'.format(name))
release_variables.append('{}_released_ = true;'.format(name))
unpack.append('auto {} = unpack_list({}_);'.format(name, name))
asserts.append('TORCH_CHECK(!{}_released_, ERR_BACKWARD_TWICE);'.format(name))
elif arg['type'] == 'IntArrayRef':
saved_variables.append('std::vector<int64_t> {};'.format(name))
elif arg['type'] == 'int64_t':
Expand All @@ -162,6 +169,7 @@ def save_arg(arg, is_output):
env['saved_variables'] = saved_variables
env['release_variables'] = release_variables
env['saved_list_sizes'] = saved_list_sizes
env['asserts'] = asserts

if uses_retain_variables(func):
env['will_release_variables'] = WILL_RELEASE_VARIABLES.substitute()
Expand Down