API Reference — `dna/layers/MachineLearnedBehaviorExtReader`

`class MachineLearnedBehaviorExtReader : public virtual MachineLearnedBehaviorReader`

Read-only accessors to the neural network extension data associated with a rig — the per-model operation graph (operation sets, operation types, static parameters, and dependency indices) that MachineLearnedBehaviorReader itself does not expose.

When to use this

Use this when you need to evaluate or inspect the actual ML operation graph for a rig — which operations run in which set, what type each operation is, and what it depends on. Implementors should inherit from Reader itself, not from this class directly, since it is meant to be composed into the full reader interface.

Method groups

Group	Methods
ML type / operation set counts	getMLTypeCount, getMLOperationSetCount, getMLOperationCount
Operation data	getMLOperationType, getMLOperationParameters
Operation dependencies	getMLOperationDependencyOperationIndices (and related dependency-index accessors)

Example

// reader implements Reader, which composes MachineLearnedBehaviorExtReader
const auto mlTypeCount = reader->getMLTypeCount();
for (std::uint16_t mlType = 0u; mlType < mlTypeCount; ++mlType) {
    const auto setCount = reader->getMLOperationSetCount(mlType);
    for (std::uint16_t setIdx = 0u; setIdx < setCount; ++setIdx) {
        const auto opCount = reader->getMLOperationCount(mlType, setIdx);
        for (std::uint16_t opIdx = 0u; opIdx < opCount; ++opIdx) {
            auto opType = reader->getMLOperationType(mlType, setIdx, opIdx);
            // dispatch evaluation based on opType
        }
    }
}

Parameters

Name	Type	Description
`mlTypeIndex`	`std::uint16_t`	required — position in the zero-indexed array of ML models; must be less than `getMLTypeCount`.
`mlOperationSetIndex`	`std::uint16_t`	required — position of the operation set within the selected ML model; must be less than `getMLOperationSetCount`.
`mlOperationIndex`	`std::uint16_t`	required — position of the operation within the selected set; must be less than `getMLOperationCount`.

Returns

std::uint16_t / MachineLearnedBehaviorOperationType / ConstArrayView<std::uint32_t> — counts, the operation's type, or a view over its static parameters/dependency indices, depending on the accessor called.

Watch out for

Operations within a single operation set are safe to execute in parallel, but operation sets themselves must be executed sequentially, one after another.
Each subsequent operation set can access the results of earlier sets, which is the basis for the dependency-index accessors.
Index bounds are not checked automatically: mlTypeIndex must be less than getMLTypeCount(), mlOperationSetIndex less than getMLOperationSetCount(mlTypeIndex), and mlOperationIndex less than getMLOperationCount(mlTypeIndex, mlOperationSetIndex). Passing an out-of-range index produces undefined behavior.
Dependency indices from getMLOperationDependencyOperationSetIndices and getMLOperationDependencyOperationIndices form paired lists: element i from each list together identify one (operation-set, operation) dependency pair. Process both arrays together under the same index.

API Reference — dna/layers/MachineLearnedBehaviorExtReader

class MachineLearnedBehaviorExtReader : public virtual MachineLearnedBehaviorReader