Embedding

public abstract class Embedding

Represents the embedding for a given embedder head. Typically used in embedding tasks.

One and only one of the two 'floatEmbedding' and 'quantizedEmbedding' will contain data, based on whether or not the embedder was configured to perform scala quantization.

Summary

Public constructors
`Embedding()`

Public methods
`static Embedding`	`create( float[] floatEmbedding, byte[] quantizedEmbedding, int headIndex, Optional<String> headName )` Creates an `Embedding` instance.
`static Embedding`	`createFromProto(EmbeddingsProto.Embedding proto)` Creates an `Embedding` object from an `EmbeddingsProto.Embedding` protobuf message.
`abstract float[]`	`floatEmbedding()` Floating-point embedding.
`abstract int`	`headIndex()` The index of the embedder head these entries refer to.
`abstract Optional<String>`	`headName()` The optional name of the embedder head, which is the corresponding tensor metadata name.
`abstract byte[]`	`quantizedEmbedding()` Quantized embedding.

Public constructors

Embedding

public Embedding()

Public methods

create

public static Embedding create(
    float[] floatEmbedding,
    byte[] quantizedEmbedding,
    int headIndex,
    Optional<String> headName
)

Creates an Embedding instance.

Parameters
`float[] floatEmbedding`	the floating-point embedding
`byte[] quantizedEmbedding`	the quantized embedding.
`int headIndex`	the index of the embedder head.
`Optional<String> headName`	the optional name of the embedder head.

createFromProto

public static Embedding createFromProto(EmbeddingsProto.Embedding proto)

Creates an Embedding object from an EmbeddingsProto.Embedding protobuf message.

Parameters
`EmbeddingsProto.Embedding proto`	the `EmbeddingsProto.Embedding` protobuf message to convert.

floatEmbedding

public abstract float[] floatEmbedding()

Floating-point embedding.

Empty if the embedder was configured to perform scalar quantization.

headIndex

public abstract int headIndex()

The index of the embedder head these entries refer to. This is useful for multi-head models.

headName

public abstract Optional<String> headName()

The optional name of the embedder head, which is the corresponding tensor metadata name.

quantizedEmbedding

public abstract byte[] quantizedEmbedding()

Quantized embedding.

Empty if the embedder was not configured to perform scalar quantization.