Summary
Deep learning is a powerful technique that you can use to train sophisticated machine learning models. Azure Databricks provides a suitable platform for deep learning model training because of its scalability and ability to work with large volumes of data in a data lake.
You can use any deep learning library that runs in PySpark, including PyTorch and TensorFlow; and you can distribute training tasks for these frameworks across cluster nodes by using Horovod.
Tip
For more information about deep learning on Azure Databricks, see Best practices for deep learning on Azure Databricks.