Class AllMiniLmL6V2QuantizedEmbeddingModel

java.lang.Object

dev.langchain4j.model.embedding.DimensionAwareEmbeddingModel

dev.langchain4j.model.embedding.onnx.AbstractInProcessEmbeddingModel

dev.langchain4j.model.embedding.onnx.allminilml6v2q.AllMiniLmL6V2QuantizedEmbeddingModel

All Implemented Interfaces:

EmbeddingModel

public class AllMiniLmL6V2QuantizedEmbeddingModel
extends AbstractInProcessEmbeddingModel

Quantized SentenceTransformers all-MiniLM-L6-v2 embedding model that runs within your Java application's process.

Maximum length of text (in tokens) that can be embedded at once: unlimited. However, while you can embed very long texts, the quality of the embedding degrades as the text lengthens. It is recommended to embed segments of no more than 256 tokens.

Embedding dimensions: 384

Uses an Executor to parallelize the embedding process. By default, uses a cached thread pool with the number of threads equal to the number of available processors. Threads are cached for 1 second.

More details here and here

Field Summary

Fields inherited from class DimensionAwareEmbeddingModel

dimension

Constructor Summary

Constructors

Constructor	Description
AllMiniLmL6V2QuantizedEmbeddingModel()	Creates an instance of an AllMiniLmL6V2QuantizedEmbeddingModel.
AllMiniLmL6V2QuantizedEmbeddingModel(Executor executor)	Creates an instance of an AllMiniLmL6V2QuantizedEmbeddingModel.

Method Summary

Modifier and Type	Method	Description
protected Integer	knownDimension()	When known (e.g., can be derived from the model name), returns the dimension of the Embedding produced by this embedding model.
protected OnnxBertBiEncoder	model()

Methods inherited from class AbstractInProcessEmbeddingModel

embedAll, loadFromJar

Methods inherited from class DimensionAwareEmbeddingModel

dimension

Methods inherited from class Object

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

Methods inherited from interface EmbeddingModel

embed, embed, modelName

Constructor Details

AllMiniLmL6V2QuantizedEmbeddingModel

public AllMiniLmL6V2QuantizedEmbeddingModel()

Creates an instance of an AllMiniLmL6V2QuantizedEmbeddingModel. Uses a cached thread pool with the number of threads equal to the number of available processors. Threads are cached for 1 second.

AllMiniLmL6V2QuantizedEmbeddingModel

public AllMiniLmL6V2QuantizedEmbeddingModel(Executor executor)

Creates an instance of an AllMiniLmL6V2QuantizedEmbeddingModel.

Parameters:

executor - The executor to use to parallelize the embedding process.

Method Details

model

protected OnnxBertBiEncoder model()

Specified by:

model in class AbstractInProcessEmbeddingModel

knownDimension

protected Integer knownDimension()

Description copied from class: DimensionAwareEmbeddingModel

When known (e.g., can be derived from the model name), returns the dimension of the Embedding produced by this embedding model. Otherwise, it returns null.

Overrides:

knownDimension in class DimensionAwareEmbeddingModel

Returns:

the known dimension of the Embedding, or null if unknown.