フリー問題

問題文

You want to record the accuracy achieved by each training run, the learning rate that produced it, and the resulting model files, so that all of this appears in the Azure Machine Learning workspace without adding cloud-specific code to your training script. Which approach should you use?

選択肢

1. Write the values to a text file in the compute instance's local disk
2. Add the values as tags on the compute cluster resource
3. Store the values in the datastore that holds the training data
4. Use MLflow tracking in the training script