def test_optimizer_state(self, tmp_path: Path, xor_trial_controller: Callable) -> None:
def make_trial_controller_fn(
workloads: workload.Stream, load_path: Optional[str] = None
) -> det.TrialController:
hparams = {**self.hparams, "optimizer": "adam"}
return xor_trial_controller(hparams, workloads, load_path=load_path)
utils.optimizer_state_test(make_trial_controller_fn, tmp_path)
def test_optimizer_state(self, tmp_path: pathlib.Path) -> None:
def make_trial_controller_fn(
workloads: workload.Stream,
load_path: typing.Optional[str] = None) -> det.TrialController:
return utils.make_trial_controller_from_trial_implementation(
trial_class=pytorch_xor_model.XORTrialOptimizerState,
hparams=self.hparams,
workloads=workloads,
load_path=load_path,
trial_seed=self.trial_seed,
)
utils.optimizer_state_test(make_trial_controller_fn, tmp_path)
def test_optimizer_state(self, tmp_path: Path,
xor_trial_controller: Callable) -> None:
def make_trial_controller_fn(
workloads: workload.Stream,
load_path: Optional[str] = None) -> det.TrialController:
return xor_trial_controller(
self.hparams,
workloads,
batches_per_step=100,
load_path=load_path,
trial_seed=self.trial_seed,
)
utils.optimizer_state_test(make_trial_controller_fn, tmp_path)
