fix get_scheduler when name is warmup_stable_decay (#31128)
fix get_scheduler args
This commit is contained in:
@@ -540,6 +540,9 @@ def get_scheduler(
|
|||||||
if name == SchedulerType.INVERSE_SQRT:
|
if name == SchedulerType.INVERSE_SQRT:
|
||||||
return schedule_func(optimizer, num_warmup_steps=num_warmup_steps)
|
return schedule_func(optimizer, num_warmup_steps=num_warmup_steps)
|
||||||
|
|
||||||
|
if name == SchedulerType.WARMUP_STABLE_DECAY:
|
||||||
|
return schedule_func(optimizer, num_warmup_steps=num_warmup_steps, **scheduler_specific_kwargs)
|
||||||
|
|
||||||
# All other schedulers require `num_training_steps`
|
# All other schedulers require `num_training_steps`
|
||||||
if num_training_steps is None:
|
if num_training_steps is None:
|
||||||
raise ValueError(f"{name} requires `num_training_steps`, please provide that argument.")
|
raise ValueError(f"{name} requires `num_training_steps`, please provide that argument.")
|
||||||
|
|||||||
@@ -36,6 +36,7 @@ if is_torch_available():
|
|||||||
get_inverse_sqrt_schedule,
|
get_inverse_sqrt_schedule,
|
||||||
get_linear_schedule_with_warmup,
|
get_linear_schedule_with_warmup,
|
||||||
get_polynomial_decay_schedule_with_warmup,
|
get_polynomial_decay_schedule_with_warmup,
|
||||||
|
get_scheduler,
|
||||||
get_wsd_schedule,
|
get_wsd_schedule,
|
||||||
)
|
)
|
||||||
|
|
||||||
@@ -176,6 +177,27 @@ class ScheduleInitTest(unittest.TestCase):
|
|||||||
lrs_2 = unwrap_and_save_reload_schedule(scheduler, self.num_steps)
|
lrs_2 = unwrap_and_save_reload_schedule(scheduler, self.num_steps)
|
||||||
self.assertListEqual(lrs_1, lrs_2, msg=f"failed for {scheduler_func} in save and reload")
|
self.assertListEqual(lrs_1, lrs_2, msg=f"failed for {scheduler_func} in save and reload")
|
||||||
|
|
||||||
|
def test_get_scheduler(self):
|
||||||
|
test_params = [
|
||||||
|
{
|
||||||
|
"name": "warmup_stable_decay",
|
||||||
|
"optimizer": self.optimizer,
|
||||||
|
"num_warmup_steps": 2,
|
||||||
|
"scheduler_specific_kwargs": {"num_stable_steps": 1, "num_decay_steps": 3},
|
||||||
|
},
|
||||||
|
{
|
||||||
|
"name": "warmup_stable_decay",
|
||||||
|
"optimizer": self.optimizer,
|
||||||
|
"num_warmup_steps": 2,
|
||||||
|
"num_training_steps": 10,
|
||||||
|
"scheduler_specific_kwargs": {"num_stable_steps": 1, "num_decay_steps": 3},
|
||||||
|
},
|
||||||
|
{"name": "cosine", "optimizer": self.optimizer, "num_warmup_steps": 2, "num_training_steps": 10},
|
||||||
|
]
|
||||||
|
|
||||||
|
for param in test_params:
|
||||||
|
self.assertTrue(get_scheduler(**param), msg=f"failed for {param['name']} in get_scheduler")
|
||||||
|
|
||||||
|
|
||||||
class LambdaScheduleWrapper:
|
class LambdaScheduleWrapper:
|
||||||
"""See https://github.com/huggingface/transformers/issues/21689"""
|
"""See https://github.com/huggingface/transformers/issues/21689"""
|
||||||
|
|||||||
Reference in New Issue
Block a user