Record Class QwenVlProcessor.ProcessedMultimodal

java.lang.Object

java.lang.Record

smile.llm.model.qwen.QwenVlProcessor.ProcessedMultimodal

Record Components:

inputIds - expanded token ids including vision pads.

mmTokenTypeIds - 0=text, 1=image, 2=video.

pixelValues - packed patches, or null when text-only.

imageGridThw - image grids.

videoGridThw - video grids.

mrope - interleaved mRoPE planes.

Enclosing class:

QwenVlProcessor

public static record QwenVlProcessor.ProcessedMultimodal(int[] inputIds, int[] mmTokenTypeIds, Tensor pixelValues, int[][] imageGridThw, int[][] videoGridThw, InterleavedMRope.MropePositions mrope)
extends Record

Result of multimodal preprocessing for one prompt.

Constructor Summary

Constructors

Constructor	Description
ProcessedMultimodal(int[] inputIds, int[] mmTokenTypeIds, Tensor pixelValues, int[][] imageGridThw, int[][] videoGridThw, InterleavedMRope.MropePositions mrope)	Creates an instance of a ProcessedMultimodal record class.

Method Summary

Modifier and Type	Method	Description
final boolean	equals(Object o)	Indicates whether some other object is "equal to" this one.
final int	hashCode()	Returns a hash code value for this object.
boolean	hasVision()	Returns whether vision tensors are present.
int[][]	imageGridThw()	Returns the value of the imageGridThw record component.
int[]	inputIds()	Returns the value of the inputIds record component.
int[]	mmTokenTypeIds()	Returns the value of the mmTokenTypeIds record component.
InterleavedMRope.MropePositions	mrope()	Returns the value of the mrope record component.
Tensor	pixelValues()	Returns the value of the pixelValues record component.
QwenVlProcessor.ProcessedMultimodal	to(Device device, ScalarType dtype)	Moves pixel values to device (no-op when null).
final String	toString()	Returns a string representation of this record class.
int[][]	videoGridThw()	Returns the value of the videoGridThw record component.

Methods inherited from class Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Constructor Details

ProcessedMultimodal

public ProcessedMultimodal(int[] inputIds,
 int[] mmTokenTypeIds,
 Tensor pixelValues,
 int[][] imageGridThw,
 int[][] videoGridThw,
 InterleavedMRope.MropePositions mrope)

Creates an instance of a ProcessedMultimodal record class.

Parameters:

inputIds - the value for the inputIds record component

mmTokenTypeIds - the value for the mmTokenTypeIds record component

pixelValues - the value for the pixelValues record component

imageGridThw - the value for the imageGridThw record component

videoGridThw - the value for the videoGridThw record component

mrope - the value for the mrope record component

Method Details

hasVision

public boolean hasVision()

Returns whether vision tensors are present.

Returns:

true when vision tensors are present.

to

public QwenVlProcessor.ProcessedMultimodal to(Device device,
 ScalarType dtype)

Moves pixel values to device (no-op when null).

Parameters:

device - target device.

dtype - compute dtype.

Returns:

this, with updated pixelValues ownership transferred.

toString

public final String toString()

Returns a string representation of this record class. The representation contains the name of the class, followed by the name and value of each of the record components.

Specified by:

toString in class Record

Returns:

a string representation of this object

hashCode

public final int hashCode()

Returns a hash code value for this object. The value is derived from the hash code of each of the record components.

Specified by:

hashCode in class Record

Returns:

a hash code value for this object

equals

public final boolean equals(Object o)

Indicates whether some other object is "equal to" this one. The objects are equal if the other object is of the same class and if all the record components are equal. All components in this record class are compared with Objects::equals(Object,Object).

Specified by:

equals in class Record

Parameters:

o - the object with which to compare

Returns:

true if this object is the same as the o argument; false otherwise.

inputIds

public int[] inputIds()

Returns the value of the inputIds record component.

Returns:

the value of the inputIds record component

mmTokenTypeIds

public int[] mmTokenTypeIds()

Returns the value of the mmTokenTypeIds record component.

Returns:

the value of the mmTokenTypeIds record component

pixelValues

public Tensor pixelValues()

Returns the value of the pixelValues record component.

Returns:

the value of the pixelValues record component

imageGridThw

public int[][] imageGridThw()

Returns the value of the imageGridThw record component.

Returns:

the value of the imageGridThw record component

videoGridThw

public int[][] videoGridThw()

Returns the value of the videoGridThw record component.

Returns:

the value of the videoGridThw record component

mrope

public InterleavedMRope.MropePositions mrope()

Returns the value of the mrope record component.

Returns:

the value of the mrope record component