Generate Embeddings from text using Batch processing

The code sample showcases how to use a pre-trained model to batch generate embeddings for a list of text inputs, and store them in a specified location.

Code sample

Java

Before trying this sample, follow the Java setup instructions in the Vertex AI quickstart using client libraries. For more information, see the Vertex AI Java API reference documentation.

To authenticate to Vertex AI, set up Application Default Credentials. For more information, see Set up authentication for a local development environment. 


import com.google.cloud.aiplatform.v1.BatchPredictionJob;
import com.google.cloud.aiplatform.v1.GcsDestination;
import com.google.cloud.aiplatform.v1.GcsSource;
import com.google.cloud.aiplatform.v1.JobServiceClient;
import com.google.cloud.aiplatform.v1.JobServiceSettings;
import com.google.cloud.aiplatform.v1.LocationName;
import java.io.IOException;

public class EmbeddingBatchSample {

  public static void main(String[] args) throws IOException, InterruptedException {
    // TODO(developer): Replace these variables before running the sample.
    String project = "YOUR_PROJECT_ID";
    String location = "us-central1";
    // inputUri: URI of the input dataset.
    // Could be a BigQuery table or a Google Cloud Storage file.
    // E.g. "gs://[BUCKET]/[DATASET].jsonl" OR "bq://[PROJECT].[DATASET].[TABLE]"
    String inputUri = "gs://cloud-samples-data/generative-ai/embeddings/embeddings_input.jsonl";
    // outputUri: URI where the output will be stored.
    // Could be a BigQuery table or a Google Cloud Storage file.
    // E.g. "gs://[BUCKET]/[OUTPUT].jsonl" OR "bq://[PROJECT].[DATASET].[TABLE]"
    String outputUri = "gs://YOUR_BUCKET/embedding_batch_output";
    String textEmbeddingModel = "text-embedding-005";

    embeddingBatchSample(project, location, inputUri, outputUri, textEmbeddingModel);
  }

  // Generates embeddings from text using batch processing.
  // Read more: https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/batch-prediction-genai-embeddings
  public static BatchPredictionJob embeddingBatchSample(
      String project, String location, String inputUri, String outputUri, String textEmbeddingModel)
      throws IOException {
    BatchPredictionJob response;
    JobServiceSettings jobServiceSettings =  JobServiceSettings.newBuilder()
        .setEndpoint("us-central1-aiplatform.googleapis.com:443").build();
    LocationName parent = LocationName.of(project, location);
    String modelName = String.format("projects/%s/locations/%s/publishers/google/models/%s",
        project, location, textEmbeddingModel);

    // Initialize client that will be used to send requests. This client only needs to be created
    // once, and can be reused for multiple requests.
    try (JobServiceClient client = JobServiceClient.create(jobServiceSettings)) {
      BatchPredictionJob batchPredictionJob =
          BatchPredictionJob.newBuilder()
              .setDisplayName("my embedding batch job " + System.currentTimeMillis())
              .setModel(modelName)
              .setInputConfig(
                  BatchPredictionJob.InputConfig.newBuilder()
                      .setGcsSource(GcsSource.newBuilder().addUris(inputUri).build())
                      .setInstancesFormat("jsonl")
                      .build())
              .setOutputConfig(
                  BatchPredictionJob.OutputConfig.newBuilder()
                      .setGcsDestination(GcsDestination.newBuilder()
                          .setOutputUriPrefix(outputUri).build())
                      .setPredictionsFormat("jsonl")
                      .build())
              .build();

      response = client.createBatchPredictionJob(parent, batchPredictionJob);

      System.out.format("response: %s\n", response);
      System.out.format("\tName: %s\n", response.getName());
    }
    return response;
  }
}

Python

Before trying this sample, follow the Python setup instructions in the Vertex AI quickstart using client libraries. For more information, see the Vertex AI Python API reference documentation.

To authenticate to Vertex AI, set up Application Default Credentials. For more information, see Set up authentication for a local development environment. 

import vertexai

from vertexai.preview import language_models

# TODO(developer): Update & uncomment line below
# PROJECT_ID = "your-project-id"
vertexai.init(project=PROJECT_ID, location="us-central1")
input_uri = (
    "gs://cloud-samples-data/generative-ai/embeddings/embeddings_input.jsonl"
)
# Format: `"gs://your-bucket-unique-name/directory/` or `bq://project_name.llm_dataset`
output_uri = OUTPUT_URI

textembedding_model = language_models.TextEmbeddingModel.from_pretrained(
    "textembedding-gecko@003"
)

batch_prediction_job = textembedding_model.batch_predict(
    dataset=[input_uri],
    destination_uri_prefix=output_uri,
)
print(batch_prediction_job.display_name)
print(batch_prediction_job.resource_name)
print(batch_prediction_job.state)
# Example response:
# BatchPredictionJob 2024-09-10 15:47:51.336391
# projects/1234567890/locations/us-central1/batchPredictionJobs/123456789012345
# JobState.JOB_STATE_SUCCEEDED

What's next

To search and filter code samples for other Google Cloud products, see the Google Cloud sample browser.