Package com.helix.core.ml

Class OnnxModelExecutor

java.lang.Object

com.helix.core.ml.OnnxModelExecutor

public final class OnnxModelExecutor
extends Object

Static runtime bridge for executing ONNX model inference from compiled JVM bytecode and interpreters.

Provides high-performance static dispatch methods invoked directly by compiled rule classes via INVOKESTATIC, delegating to the globally active OnnxSessionPool.

Method Summary

Modifier and Type	Method	Description
static double	evaluate(String modelName, ExecutionContext context)	Evaluates the named model using its default output tensor and class index.
static double	evaluate(String modelName, String outputTensorName, int outputIndex, ExecutionContext context)	Evaluates the named model with custom output tensor and index mapping.
static OnnxInferenceResult	evaluateWithResult(String modelName, ExecutionContext context)	Evaluates the named model and returns the full telemetry OnnxInferenceResult.
static OnnxSessionPool	getSessionPool()	Retrieves the active session pool.
static void	reset()	Resets the active session pool reference.
static void	setSessionPool(OnnxSessionPool pool)	Registers the active session pool for runtime evaluation.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

setSessionPool

public static void setSessionPool(OnnxSessionPool pool)

Registers the active session pool for runtime evaluation.

Parameters:

pool - session pool instance

getSessionPool

public static OnnxSessionPool getSessionPool()

Retrieves the active session pool.

Returns:

current active session pool, or null if uninitialized

reset

public static void reset()

Resets the active session pool reference.

evaluate

public static double evaluate(String modelName,
 ExecutionContext context)

Evaluates the named model using its default output tensor and class index.

Parameters:

modelName - model identifier

context - runtime execution context

Returns:

probability or score value as a double

evaluate

public static double evaluate(String modelName,
 String outputTensorName,
 int outputIndex,
 ExecutionContext context)

Evaluates the named model with custom output tensor and index mapping.

Parameters:

modelName - model identifier

outputTensorName - target output tensor name

outputIndex - target class or output index

context - runtime execution context

Returns:

extracted score value as a double

evaluateWithResult

public static OnnxInferenceResult evaluateWithResult(String modelName,
 ExecutionContext context)

Evaluates the named model and returns the full telemetry OnnxInferenceResult.

Parameters:

modelName - model identifier

context - runtime execution context

Returns:

detailed inference result