Q8_0BlockTensorData

Implementation of Q8_0TensorData backed by a packed byte array.

Memory layout per block:

initialShape

the logical shape of the tensor (in elements, not blocks)

packedData

the raw packed block data

Constructors

constructor(initialShape: Shape, data: ByteArray)

open override val blockCount: Int

Number of Q8_0 blocks in the tensor.

open override val blockSize: Int

Number of logical elements per block.

The physical encoding describing the block layout.

open override val packedData: ByteArray

Raw packed data containing all blocks.

open override val shape: Shape

The shape descriptor that defines the dimensionality and size of this tensor data.

open override fun dequantizeBlock(blockIdx: Int, output: FloatArray, outputOffset: Int = 0)

Dequantize a single block to float values.

open operator override fun get(vararg indices: Int): Byte

Retrieves an element at the specified multidimensional indices.

open override fun getBlockScale(blockIdx: Int): Float

Get the scale factor for a specific block.

open override fun getCode(blockIdx: Int, elementIdx: Int): Byte

Get a quantized code value within a block (0..31).

open operator override fun set(vararg indices: Int, value: Byte)

Setter