Skip to content

Save variables with keys in the checkpoint #46

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
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Save variables with keys in the checkpoint #46

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
14 changes: 12 additions & 2 deletions recml/core/training/keras_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -118,6 +118,7 @@ def __init__(
max_checkpoints_to_keep: int = 5,
checkpoint_save_interval_epochs: int = 1,
rng_seed: int = core.DEFAULT_RNG_SEED,
legacy_format: bool = True,
):
"""Initializes the instance."""

Expand Down Expand Up @@ -148,12 +149,14 @@ def __init__(
model_dir, core.TRAINING_COMPLETE_MARKER_FILE
)
self._checkpoint_dir = os.path.join(model_dir, core.CHECKPOINT_DIR)
self._legacy_format = legacy_format

if keras.backend.backend() == "jax":
self._checkpoint_manager = keras_utils.KerasOrbaxCheckpointManager(
checkpoint_dir=self._checkpoint_dir,
max_to_keep=max_checkpoints_to_keep,
save_interval_epochs=checkpoint_save_interval_epochs,
legacy_format=self._legacy_format,
)
self._train_callbacks = [
keras_utils.EpochSummaryCallback(
Expand Down Expand Up @@ -314,13 +317,18 @@ def __init__(
self,
checkpoint_dir: str,
epoch: int,
legacy_format: bool,
):
self._checkpoint_dir = checkpoint_dir
self._epoch = epoch
self._legacy_format = legacy_format

def on_test_begin(self, logs: Mapping[str, Any] | None = None):
keras_utils.restore_keras_model(
model, self._checkpoint_dir, step=self._epoch
model,
self._checkpoint_dir,
step=self._epoch,
legacy_format=self._legacy_format,
)

history = None
Expand All @@ -329,7 +337,9 @@ def on_test_begin(self, logs: Mapping[str, Any] | None = None):
timeout=self._continuous_eval_timeout,
timeout_fn=timeout_fn,
):
restore_callback = _RestoreCallback(self._checkpoint_dir, epoch)
restore_callback = _RestoreCallback(
self._checkpoint_dir, epoch, self._legacy_format
)
[tb_cbk] = [
cbk
for cbk in self._eval_callbacks
Expand Down
16 changes: 16 additions & 0 deletions recml/core/training/keras_trainer_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -94,6 +94,22 @@ def test_keras_task_and_trainer(self, mode: str):
):
self.assertEqual(history.history["num_params/trainable"][0], 2)

def test_new_checkpoint_format(self):
if keras.backend.backend() != "jax":
self.skipTest("Only supported on the Jax backend.")
trainer = keras_trainer.KerasTrainer(
distribution=keras.distribution.DataParallel(),
train_steps=5,
steps_per_eval=3,
steps_per_loop=2,
model_dir=self.create_tempdir().full_path,
continuous_eval_timeout=5,
legacy_format=False,
)
experiment = core.Experiment(_KerasTask(), trainer)
core.run_experiment(experiment, core.Experiment.Mode.TRAIN)
core.run_experiment(experiment, core.Experiment.Mode.CONTINUOUS_EVAL)


if __name__ == "__main__":
absltest.main()
Loading