Skip to content

feat: expose poll interval parameter on all trainers #5859

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

Draft
joshuatowner wants to merge 1 commit into
base: master
Choose a base branch
from
Draft

feat: expose poll interval parameter on all trainers #5859

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
4 changes: 3 additions & 1 deletion sagemaker-train/src/sagemaker/train/dpo_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -181,7 +181,8 @@ def train(self,
training_dataset: Optional[Union[str, DataSet]] = None,
validation_dataset: Optional[Union[str, DataSet]] = None,
wait: bool = True,
wait_timeout: Optional[int] = None):
wait_timeout: Optional[int] = None,
poll: int = 5):
"""Execute the DPO training job.

Parameters:
Expand Down Expand Up @@ -283,6 +284,7 @@ def train(self,
wait_kwargs = {}
if wait_timeout is not None:
wait_kwargs['timeout'] = wait_timeout
wait_kwargs['poll'] = poll
_wait(training_job, **wait_kwargs)
except TimeoutExceededError as e:
logger.error("Error: %s", e)
Expand Down
3 changes: 2 additions & 1 deletion sagemaker-train/src/sagemaker/train/rlaif_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -197,7 +197,7 @@ def _validate_reward_model_id(self, reward_model_id):


@_telemetry_emitter(feature=Feature.MODEL_CUSTOMIZATION, func_name="RLAIFTrainer.train")
def train(self, training_dataset: Optional[Union[str, DataSet]] = None, validation_dataset: Optional[Union[str, DataSet]] = None, wait: bool = True, wait_timeout: Optional[int] = None):
def train(self, training_dataset: Optional[Union[str, DataSet]] = None, validation_dataset: Optional[Union[str, DataSet]] = None, wait: bool = True, wait_timeout: Optional[int] = None, poll: int = 5):
"""Execute the RLAIF training job.

Parameters:
Expand Down Expand Up @@ -301,6 +301,7 @@ def train(self, training_dataset: Optional[Union[str, DataSet]] = None, validati
wait_kwargs = {}
if wait_timeout is not None:
wait_kwargs['timeout'] = wait_timeout
wait_kwargs['poll'] = poll
_wait(training_job, **wait_kwargs)
except TimeoutExceededError as e:
logger.error("Error: %s", e)
Expand Down
3 changes: 2 additions & 1 deletion sagemaker-train/src/sagemaker/train/rlvr_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -183,7 +183,7 @@ def _process_hyperparameters(self):

@_telemetry_emitter(feature=Feature.MODEL_CUSTOMIZATION, func_name="RLVRTrainer.train")
def train(self, training_dataset: Optional[Union[str, DataSet]] = None,
validation_dataset: Optional[Union[str, DataSet]] = None, wait: bool = True, wait_timeout: Optional[int] = None):
validation_dataset: Optional[Union[str, DataSet]] = None, wait: bool = True, wait_timeout: Optional[int] = None, poll: int = 5):
"""Execute the RLVR training job.

Parameters:
Expand Down Expand Up @@ -289,6 +289,7 @@ def train(self, training_dataset: Optional[Union[str, DataSet]] = None,
wait_kwargs = {}
if wait_timeout is not None:
wait_kwargs['timeout'] = wait_timeout
wait_kwargs['poll'] = poll
_wait(training_job, **wait_kwargs)
except TimeoutExceededError as e:
logger.error("Error: %s", e)
Expand Down
3 changes: 2 additions & 1 deletion sagemaker-train/src/sagemaker/train/sft_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -180,7 +180,7 @@ def _process_hyperparameters(self):
self.hyperparameters._specs.pop('validation_data_path', None)

@_telemetry_emitter(feature=Feature.MODEL_CUSTOMIZATION, func_name="SFTTrainer.train")
def train(self, training_dataset: Optional[Union[str, DataSet]] = None, validation_dataset: Optional[Union[str, DataSet]] = None, wait: bool = True, wait_timeout: Optional[int] = None):
def train(self, training_dataset: Optional[Union[str, DataSet]] = None, validation_dataset: Optional[Union[str, DataSet]] = None, wait: bool = True, wait_timeout: Optional[int] = None, poll: int = 5):
"""Execute the SFT training job.

Parameters:
Expand Down Expand Up @@ -283,6 +283,7 @@ def train(self, training_dataset: Optional[Union[str, DataSet]] = None, validati
wait_kwargs = {}
if wait_timeout is not None:
wait_kwargs['timeout'] = wait_timeout
wait_kwargs['poll'] = poll
_wait(training_job, **wait_kwargs)
except TimeoutExceededError as e:
logger.error("Error: %s", e)
Expand Down
4 changes: 2 additions & 2 deletions sagemaker-train/tests/unit/train/test_dpo_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -420,7 +420,7 @@ def test_train_passes_wait_timeout(self, mock_training_job_create, mock_model_pa
trainer = DPOTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True, wait_timeout=600)

mock_wait.assert_called_once_with(mock_training_job, timeout=600)
mock_wait.assert_called_once_with(mock_training_job, timeout=600, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.dpo_trainer._resolve_model_and_name')
Expand Down Expand Up @@ -463,7 +463,7 @@ def test_train_without_wait_timeout_uses_default(self, mock_training_job_create,
trainer = DPOTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True)

mock_wait.assert_called_once_with(mock_training_job)
mock_wait.assert_called_once_with(mock_training_job, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.dpo_trainer._resolve_model_and_name')
Expand Down
4 changes: 2 additions & 2 deletions sagemaker-train/tests/unit/train/test_rlaif_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -596,7 +596,7 @@ def test_train_passes_wait_timeout(self, mock_training_job_create, mock_model_pa
trainer = RLAIFTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True, wait_timeout=600)

mock_wait.assert_called_once_with(mock_training_job, timeout=600)
mock_wait.assert_called_once_with(mock_training_job, timeout=600, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.common_utils.finetune_utils._get_beta_session')
Expand Down Expand Up @@ -639,7 +639,7 @@ def test_train_without_wait_timeout_uses_default(self, mock_training_job_create,
trainer = RLAIFTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True)

mock_wait.assert_called_once_with(mock_training_job)
mock_wait.assert_called_once_with(mock_training_job, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.common_utils.finetune_utils._get_beta_session')
Expand Down
4 changes: 2 additions & 2 deletions sagemaker-train/tests/unit/train/test_rlvr_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -423,7 +423,7 @@ def test_train_passes_wait_timeout(self, mock_training_job_create, mock_model_pa
trainer = RLVRTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True, wait_timeout=600)

mock_wait.assert_called_once_with(mock_training_job, timeout=600)
mock_wait.assert_called_once_with(mock_training_job, timeout=600, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.common_utils.finetune_utils._get_beta_session')
Expand Down Expand Up @@ -466,7 +466,7 @@ def test_train_without_wait_timeout_uses_default(self, mock_training_job_create,
trainer = RLVRTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True)

mock_wait.assert_called_once_with(mock_training_job)
mock_wait.assert_called_once_with(mock_training_job, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.common_utils.finetune_utils._get_beta_session')
Expand Down
4 changes: 2 additions & 2 deletions sagemaker-train/tests/unit/train/test_sft_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -434,7 +434,7 @@ def test_train_passes_wait_timeout(self, mock_training_job_create, mock_model_pa
trainer = SFTTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True, wait_timeout=600)

mock_wait.assert_called_once_with(mock_training_job, timeout=600)
mock_wait.assert_called_once_with(mock_training_job, timeout=600, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.common_utils.finetune_utils._get_beta_session')
Expand Down Expand Up @@ -477,7 +477,7 @@ def test_train_without_wait_timeout_uses_default(self, mock_training_job_create,
trainer = SFTTrainer(model="test-model", model_package_group="test-group", training_dataset="s3://bucket/train")
trainer.train(wait=True)

mock_wait.assert_called_once_with(mock_training_job)
mock_wait.assert_called_once_with(mock_training_job, poll=5)

@patch('sagemaker.train.common_utils.trainer_wait.wait')
@patch('sagemaker.train.common_utils.finetune_utils._get_beta_session')
Expand Down
Loading