Skip to content

Update SingleDeviceShardingMetadata equality method to handle CPU device naming string changes. #3052

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

Open
copybara-service wants to merge 1 commit into
base: main
Choose a base branch
from
+9 −11
Open

Update SingleDeviceShardingMetadata equality method to handle CPU device naming string changes. #3052

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
8 changes: 7 additions & 1 deletion checkpoint/orbax/checkpoint/_src/metadata/sharding.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -325,7 +325,13 @@ def __repr__(self):
return f'SingleDeviceShardingMetadata(device_str={self.device_str})'

def __eq__(self, other):
return self.device_str == other.device_str
if not isinstance(other, SingleDeviceShardingMetadata):
return False
# JAX 0.10 changed CPU devices so they report as cpu:0 not TFRT_CPU_0
return (
self.device_str.replace('TFRT_CPU_', 'cpu:')
== other.device_str.replace('TFRT_CPU_', 'cpu:')
)


def from_jax_sharding(jax_sharding) -> Optional[ShardingMetadata]:
Expand Down
12 changes: 2 additions & 10 deletions checkpoint/orbax/checkpoint/_src/metadata/sharding_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,12 +93,7 @@ def test_convert_between_jax_single_device_sharding_and_sharding_metadata(
jax_sharding = jax.sharding.SingleDeviceSharding(
jax.local_devices(backend="cpu")[0]
)
# JAX used to report its cpu devices as TFRT_CPU_0
expected_single_device_sharding_metadata = (
sharding_metadata.SingleDeviceShardingMetadata(device_str="TFRT_CPU_0")
)
# ... but now uses cpu:0
expected_single_device_sharding_metadata2 = (
sharding_metadata.SingleDeviceShardingMetadata(device_str="cpu:0")
)
converted_single_device_sharding_metadata = (
Expand All @@ -109,12 +104,9 @@ def test_convert_between_jax_single_device_sharding_and_sharding_metadata(
converted_single_device_sharding_metadata,
sharding_metadata.SingleDeviceShardingMetadata,
)
self.assertIn(
self.assertEqual(
converted_single_device_sharding_metadata,
[
expected_single_device_sharding_metadata,
expected_single_device_sharding_metadata2,
],
expected_single_device_sharding_metadata,
)

# Convert from `SingleDeviceShardingMetadata` to
Expand Down
Loading