tessl/maven-dev-langchain4j--langchain4j-vertex-ai
LangChain4j integration for Google Vertex AI models including chat, language, embedding, image, and scoring capabilities
—
Pending
Quality
Pending
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
api.mddocs/models/embedding/
Embedding Model API Reference
Class Definition
public class VertexAiEmbeddingModel extends DimensionAwareEmbeddingModel {
public Response<List<Embedding>> embedAll(List<TextSegment> segments);
public String modelName();
public List<Integer> calculateTokensCounts(List<TextSegment> segments);
public static Builder builder();
public static final Integer DEFAULT_MAX_SEGMENTS_PER_BATCH = 250;
public static final Integer DEFAULT_MAX_TOKENS_PER_BATCH = 20000;
}Methods
/**
* Embed multiple text segments into vectors.
*
* @param segments List of text segments to embed
* @return Response containing list of embeddings with metadata
*/
public Response<List<Embedding>> embedAll(List<TextSegment> segments);
/**
* Get the model name being used.
*
* @return The model name string
*/
public String modelName();
/**
* Calculate token counts for text segments.
*
* @param segments List of text segments
* @return List of token counts corresponding to each segment
*/
public List<Integer> calculateTokensCounts(List<TextSegment> segments);
/**
* Create a new builder for configuring a VertexAiEmbeddingModel.
*
* @return A new Builder instance
*/
public static Builder builder();Builder Methods
Required
public Builder endpoint(String endpoint); // API endpoint
public Builder project(String project); // GCP project ID
public Builder location(String location); // GCP region
public Builder publisher(String publisher); // Model publisher
public Builder modelName(String modelName); // Model name/versionOptional
public Builder maxRetries(Integer maxRetries); // Default: 2
public Builder maxSegmentsPerBatch(Integer maxSegmentsPerBatch); // Default: 250
public Builder maxTokensPerBatch(Integer maxTokensPerBatch); // Default: 20,000
public Builder taskType(TaskType taskType); // Task optimization
public Builder titleMetadataKey(String titleMetadataKey); // Default: "title"
public Builder autoTruncate(Boolean autoTruncate); // Default: false
public Builder outputDimensionality(Integer outputDimensionality); // Custom dimension
public Builder credentials(GoogleCredentials credentials); // Custom auth
public VertexAiEmbeddingModel build();TaskType Enum
public enum TaskType {
RETRIEVAL_QUERY, // Query for retrieval tasks
RETRIEVAL_DOCUMENT, // Document for retrieval tasks
SEMANTIC_SIMILARITY, // Semantic similarity comparison
CLASSIFICATION, // Text classification
CLUSTERING, // Text clustering
QUESTION_ANSWERING, // Question answering
FACT_VERIFICATION, // Fact verification
CODE_RETRIEVAL_QUERY // Code retrieval query
}VertexAiEmbeddingModelName Enum
public enum VertexAiEmbeddingModelName {
MULTIMODALEMBEDDING("multimodalembedding", 1408),
TEXT_EMBEDDING_004("text-embedding-004", 768),
TEXT_EMBEDDING_PREVIEW_0815("text-embedding-preview-0815", 768),
TEXT_MULTILINGUAL_EMBEDDING_002("text-multilingual-embedding-002", 768),
TEXTEMBEDDING_GECKO_MULTILINGUAL_001("textembedding-gecko-multilingual@001", 768),
TEXTEMBEDDING_GECKO_001("textembedding-gecko@001", 768),
TEXTEMBEDDING_GECKO_002("textembedding-gecko@002", 768),
TEXTEMBEDDING_GECKO_003("textembedding-gecko@003", 768);
public String toString();
public Integer dimension();
public static Integer knownDimension(String modelName);
}Deprecated Constructor
/**
* @deprecated Since version 1.2.0, use builder() instead
*/
@Deprecated
public VertexAiEmbeddingModel(
String endpoint,
String project,
String location,
String publisher,
String modelName,
Integer maxRetries,
Integer maxSegmentsPerBatch,
Integer maxTokensPerBatch,
TaskType taskType,
String titleMetadataKey,
Integer outputDimensionality,
Boolean autoTruncate
);Parameter Details
endpoint
- Required: Yes
- Format:
{region}-aiplatform.googleapis.com:443(note: different from other models) - Examples:
us-central1-aiplatform.googleapis.com:443
maxSegmentsPerBatch
- Type: Integer
- Default: 250
- Description: Maximum segments per API call
- Constant:
DEFAULT_MAX_SEGMENTS_PER_BATCH
maxTokensPerBatch
- Type: Integer
- Default: 20,000
- Description: Maximum total tokens per batch
- Constant:
DEFAULT_MAX_TOKENS_PER_BATCH
taskType
- Type: TaskType enum
- Required: No
- Purpose: Optimizes embeddings for specific use cases
- Usage:
RETRIEVAL_DOCUMENT: For indexing documentsRETRIEVAL_QUERY: For search queries- Others: Specific task optimization
titleMetadataKey
- Type: String
- Default:
"title" - Description: Metadata key for document titles
- Note: Model uses titles for better embeddings when available
autoTruncate
- Type: Boolean
- Default:
false - Description: Auto-truncate texts exceeding model limits
- Behavior: If false, throws exception for oversized texts
outputDimensionality
- Type: Integer
- Required: No
- Description: Custom embedding dimension (model-dependent)
- Note: Not all models support all dimensions
Response Types
Response<List<Embedding>>
public class Response<T> {
public T content();
public TokenUsage tokenUsage();
public FinishReason finishReason();
}Embedding
public class Embedding {
public float[] vector();
public int dimension();
}Batching Behavior
The model automatically batches requests based on:
- maxSegmentsPerBatch: Max segments in one API call
- maxTokensPerBatch: Max total tokens in one batch
If input exceeds limits, automatically splits into multiple API calls.
Error Handling
- Automatically retries transient failures up to
maxRetriestimes - If
autoTruncateis true, truncates oversized texts - If
autoTruncateis false, throws exception for oversized texts
Thread Safety
Thread-safe, can be reused across threads. Builder is not thread-safe.
Performance
- Reuse model instances
- Use appropriate
taskTypefor best quality - Adjust batch sizes based on text lengths
- Monitor token counts via
calculateTokensCounts()
Install with Tessl CLI
npx tessl i tessl/maven-dev-langchain4j--langchain4j-vertex-ai