Skip to content

Move CheckpointConnector.fault_tolerant_auto_save_path out of CheckpointConnector.hpc_resume_path #11092

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

Merged
merged 4 commits into from
Dec 21, 2021
Merged

Move CheckpointConnector.fault_tolerant_auto_save_path out of CheckpointConnector.hpc_resume_path #11092

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ab056db
first commit
jjenniferdai Dec 15, 2021
d335efe
_ hpc and auto-save paths
jjenniferdai Dec 17, 2021
ceddfc8
update
jjenniferdai Dec 17, 2021
6f95a73
fault_tolerant_auto_save_path --> fault_tolerant_auto_resume_path
jjenniferdai Dec 18, 2021
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
20 changes: 12 additions & 8 deletions pytorch_lightning/trainer/connectors/checkpoint_connector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,25 +49,29 @@ def __init__(self, trainer: "pl.Trainer", resume_from_checkpoint: Optional[_PATH
self._loaded_checkpoint: Dict[str, Any] = {}

@property
def hpc_resume_path(self) -> Optional[str]:
def _hpc_resume_path(self) -> Optional[str]:
if not os.path.isdir(self.trainer.weights_save_path):
return None
dir_path_hpc = str(self.trainer.weights_save_path)
max_version = self.max_ckpt_version_in_folder(dir_path_hpc, "hpc_ckpt_")
if max_version is not None:
return os.path.join(dir_path_hpc, f"hpc_ckpt_{max_version}.ckpt")
auto_save_checkpoint = os.path.join(dir_path_hpc, ".pl_auto_save.ckpt")
if os.path.exists(auto_save_checkpoint):
return auto_save_checkpoint

@property
def _fault_tolerant_auto_resume_path(self) -> Optional[str]:
auto_saved_path = os.path.join(str(self.trainer.weights_save_path), ".pl_auto_save.ckpt")
if os.path.exists(auto_saved_path):
return auto_saved_path

def resume_start(self, checkpoint_path: Optional[_PATH] = None) -> None:
"""Attempts to pre-load the checkpoint file to memory, with the source path determined in this priority:

1. from HPC weights if found
2. from `checkpoint_path` file if provided
3. don't restore
2. from fault-tolerant auto-saved checkpoint if found
3. from `checkpoint_path` file if provided
4. don't restore
"""
self.resume_checkpoint_path = self.hpc_resume_path or checkpoint_path
self.resume_checkpoint_path = self._hpc_resume_path or self._fault_tolerant_auto_resume_path or checkpoint_path
checkpoint_path = self.resume_checkpoint_path
if not checkpoint_path:
return
Expand Down Expand Up @@ -162,7 +166,7 @@ def restore_model(self) -> None:

# TODO: remove this in v1.8.
# call hpc specific hook
if self.hpc_resume_path is not None:
if self._hpc_resume_path is not None:
model.on_hpc_load(self._loaded_checkpoint)

# restore model state_dict
Expand Down
2 changes: 1 addition & 1 deletion tests/trainer/connectors/test_checkpoint_connector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -133,7 +133,7 @@ def test_hpc_max_ckpt_version(tmpdir):
trainer.save_checkpoint(tmpdir / "hpc_ckpt_3.ckpt")
trainer.save_checkpoint(tmpdir / "hpc_ckpt_33.ckpt")

assert trainer.checkpoint_connector.hpc_resume_path == str(tmpdir / "hpc_ckpt_33.ckpt")
assert trainer.checkpoint_connector._hpc_resume_path == str(tmpdir / "hpc_ckpt_33.ckpt")
assert trainer.checkpoint_connector.max_ckpt_version_in_folder(tmpdir) == 33
assert trainer.checkpoint_connector.max_ckpt_version_in_folder(tmpdir / "not" / "existing") is None

Expand Down