The ML Example Partition is a utility node that partitions a set of examples, see ML Example. This set of examples may consist of unlabeled examples, for example, generated by ML Attribute Generate and ML Pose Generate, but it could also be a set of labeled examples created using ML Example.

Each part of the partition is represented by a packed primitive, each of which contains an example set that is a subset of the input example set. The total set of examples stored in the parts of the partition are the same as those stored in the input example set, without duplicates.

See Machine Learning documentation for more general information.

This node is useful for generating ML training data in a hybrid TOP-SOP setup. Each work item of the TOP network may correspond to a part in the partition, and the elements in each part can be processed by a for-construct in SOPs. This approach allows the cost of invoking work items through PDG to be amortized by processing multiple examples per PDG work item. It also allows memory use and generation speed to be balanced against ech other. It limits the part size limits memory use while increasing the part size reduces the relative work-item cooking overhead per sample.