This respository includes a number of Analytics Zoo example models. Analytics Zoo makes it easy to build deep learning applications on Spark and BigDL, by providing an end-to-end Analytics + AI Platform. These examples are packaged for Cloudera Data Science Workbench and can be downloaded as a CDSW template.
Analytics Zoo seamlessly unites Spark, TensorFlow, Keras and BigDL programs into an integrated pipeline; the entire pipeline can then transparently scale out to a large Hadoop/Spark cluster for distributed training or inference.
Analytics Zoo provides several built-in deep learning models that you can use for a variety of problem types such as anomaly detection, object detection, image classification, etc.
For detailed documentation, please refer to the following: Intel Analytics-Zoo
Environment:
- See each indivdual example model for specific requirements
Run Model with CDSW:
- Build the Analytics Zoo Engine. See detailed information here.
- Create a folder in CDSW under your project and import the model files specified.
- Make sure the spark-defaults.conf file exists with the appropriate configuration for training the model. Add or change parameter settings as needed.