How to fix SageMaker training job error "SM_CHANNEL_TRAIN"?

Question

I am building a ml workflow using step function following [this](https://sagemaker-examples.readthedocs.io/en/latest/step-functions-data-science-sdk/machine_learning_workflow_abalone/machine_learning_workflow_abalone.html). However, when I start the state machine, I got error 
```
AlgorithmError: framework error ... SM_CHANNEL_TRAIN ...exit code: 1 
```
Does anyone know how to fix it? or how to set SM_CHANNEL_TRAIN?

Thank you

Answer

Assuming you are using the sagemaker python sdk, you'll have to specify the `train` channel.

The example below shows how to specify 3 channels and their respective paths to S3.
In the training container that is started, these will be translated to the environment variable `SM_CHANNEL_{channel_name.upper()}`.
I.e. `train` channel is translated to `SM_CHANNEL_TRAIN`, `test123` is translated to `SM_CHANNEL_TEST123`.

```python
from sagemaker.estimator import Estimator

s3pth = 's3://mybucket'

data = {
	'train': f'{s3pth}/train',
    'validation': f'{s3pth}/validation',
    'test': f'{s3pth}/test',
}

# starting the train job with our uploaded datasets as input
estimator.fit(
    data,
    wait=False,
    # job_name = f"{experiment_name}--{pd.Timestamp.now().strftime('%y%m%d-%H%M%S')}",
    # experiment_config = {
    #     "TrialName": trial.trial_name,
    #     "TrialComponentDisplayName": "Training",
    # },
)
```

How to fix SageMaker training job error "SM_CHANNEL_TRAIN"?

相关内容