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

Jump to bottom

Fix mixed precision check in TestCase.run_layer_test: compare with output_spec dtype instead of hardcoded float16 #19297

Merged
merged 7 commits into from
Mar 14, 2024
Merged

Fix mixed precision check in TestCase.run_layer_test: compare with output_spec dtype instead of hardcoded float16 #19297

Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
94d5584
Fix mixed precision check: compare with output spec dtype instead of …
shkarupa-alex Mar 12, 2024
d2f8f02
Revert "Fix mixed precision check: compare with output spec dtype ins…
shkarupa-alex Mar 12, 2024
94e36e8
Restore changes
shkarupa-alex Mar 12, 2024
dbfde7f
Trying to reformat code
shkarupa-alex Mar 13, 2024
e15035a
Restore formatted code
shkarupa-alex Mar 13, 2024
3cc3854
Fix formatting
shkarupa-alex Mar 14, 2024
8ecb5c5
Fix black latest wrong formatting
shkarupa-alex Mar 14, 2024
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
21 changes: 18 additions & 3 deletions keras/testing/test_case.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -417,14 +417,29 @@ def data_generator():

if run_mixed_precision_check:
layer = layer_cls(**{**init_kwargs, "dtype": "mixed_float16"})
input_spec = tree.map_structure(
lambda spec: KerasTensor(
spec.shape,
dtype=(
layer.compute_dtype
if layer.autocast
and backend.is_float_dtype(spec.dtype)
else spec.dtype
),
),
keras_tensor_inputs,
)
if isinstance(input_data, dict):
output_data = layer(**input_data, **call_kwargs)
output_spec = layer.compute_output_spec(**input_spec)
else:
output_data = layer(input_data, **call_kwargs)
for tensor in tree.flatten(output_data):
output_spec = layer.compute_output_spec(input_spec)
for tensor, spec in zip(
tree.flatten(output_data), tree.flatten(output_spec)
):
dtype = standardize_dtype(tensor.dtype)
if is_float_dtype(dtype):
self.assertEqual(dtype, "float16")
self.assertEqual(dtype, spec.dtype)
for weight in layer.weights:
dtype = standardize_dtype(weight.dtype)
if is_float_dtype(dtype):
Expand Down