Deep learning

This section includes example notebooks using two of the most popular deep learning libraries, TensorFlow and PyTorch.

Because deep learning models are data- and computation-intensive, distributed training can be important. This section also includes information about and examples of distributed deep learning using Horovod and spark-tensorflow-distributor.

• Best practices for deep learning on Databricks
  • Resource and environment management
  • Best practices for loading data
  • Best practices for training deep learning models
  • Best practices for inference
• Get started with TensorFlow Keras in Databricks
  • Requirements
  • Single node and distributed training
  • Example notebook
• TensorFlow
  • Single node and distributed training
  • Example notebook
  • TensorBoard
• PyTorch
  • Single node and distributed training
  • Example notebook
  • Install PyTorch
• Distributed training
  • Horovod
  • spark-tensorflow-distributor
• Deep Learning Pipelines migration guide
  • Read images
  • Transfer learning
  • Distributed hyperparameter tuning
  • Distributed inference
  • Deploy models as SQL UDFs