Interface EmbeddingRequestParameters

All Known Implementing Classes:

DefaultEmbeddingRequestParameters, OpenAiEmbeddingRequestParameters

@Experimental
public interface EmbeddingRequestParameters

The per-call parameters of an EmbeddingRequest, such as dimensions() or inputType(). Provider integrations can extend this interface (via DefaultEmbeddingRequestParameters) to add provider-specific parameters.

Every value carried by these parameters is identified by an EmbeddingParameter token. The set of tokens actually populated is exposed via presentParameters(); an EmbeddingModel rejects a request carrying a parameter it does not list in EmbeddingModel.supportedParameters(), so a parameter a model cannot apply is never silently ignored.

Since:

1.18.0

See Also:

• DefaultEmbeddingRequestParameters

Field Summary

Fields

Modifier and Type	Field	Description
static final EmbeddingParameter<Integer>	DIMENSIONS	A token identifying the number of dimensions the resulting embeddings should have, for providers that support reducing the output dimensionality.
static final EmbeddingRequestParameters	EMPTY	Empty parameters: nothing is populated, so presentParameters() is empty.
static final EmbeddingParameter<EmbeddingInputType>	INPUT_TYPE	A token identifying the EmbeddingInputType (query vs document), for providers that encode queries and documents differently.
static final EmbeddingParameter<String>	MODEL_NAME	A token identifying the name of the model to use for this request.

Method Summary

Modifier and Type	Method	Description
static DefaultEmbeddingRequestParameters.Builder<?>	builder()
Integer	dimensions()
EmbeddingInputType	inputType()
String	modelName()
EmbeddingRequestParameters	overrideWith(EmbeddingRequestParameters that)	Creates a new EmbeddingRequestParameters by combining the current parameters with the specified ones.
<T> T	parameter(EmbeddingParameter<T> parameter)	Returns the value of the given parameter, or null if it is not populated.
Set<EmbeddingParameter<?>>	presentParameters()	Returns the set of parameters that are actually populated (non-null) in this instance.

Field Details

MODEL_NAME

static final EmbeddingParameter<String> MODEL_NAME

A token identifying the name of the model to use for this request.

DIMENSIONS

static final EmbeddingParameter<Integer> DIMENSIONS

A token identifying the number of dimensions the resulting embeddings should have, for providers that support reducing the output dimensionality.

INPUT_TYPE

static final EmbeddingParameter<EmbeddingInputType> INPUT_TYPE

A token identifying the EmbeddingInputType (query vs document), for providers that encode queries and documents differently.

EMPTY

static final EmbeddingRequestParameters EMPTY

Empty parameters: nothing is populated, so presentParameters() is empty.

Method Details

modelName

String modelName()

dimensions

Integer dimensions()

inputType

EmbeddingInputType inputType()

parameter

<T> T parameter(EmbeddingParameter<T> parameter)

Returns the value of the given parameter, or null if it is not populated.

Type Parameters:

T - the value type.

Parameters:

parameter - the parameter token.

Returns:

the value, or null.

presentParameters

Set<EmbeddingParameter<?>> presentParameters()

Returns the set of parameters that are actually populated (non-null) in this instance. This is what an EmbeddingModel validates against its EmbeddingModel.supportedParameters().

Returns:

the populated parameters; never null.

overrideWith

EmbeddingRequestParameters overrideWith(EmbeddingRequestParameters that)

Creates a new EmbeddingRequestParameters by combining the current parameters with the specified ones. Values from the specified parameters override values from the current parameters when there is overlap. Neither instance is modified.

Parameters:

that - the parameters whose values will override the current ones.

Returns:

a new combined EmbeddingRequestParameters instance.

builder

static DefaultEmbeddingRequestParameters.Builder<?> builder()