Skip to content

Release 9.3.0

Latest

Choose a tag to compare

@miguelgrinberg miguelgrinberg released this 04 Feb 11:40
· 46 commits to main since this release
c79031c

New packDenseVector helper function #1499

The packDenseVector function can be used to pack dense vectors for efficient bulk ingesting.

Example usage:

use Elastic\Elasticsearch\Helper\Vectors;

$params = ['body' => []];
// ...
$params['body'][] = ['index' => ['_index' => $index]];
$params['body'][] = [
    'docid' => $doc['docid'],
    'title' => $doc['title'],
    'text' => $doc['text'],
    'emb' => Vectors::packDenseVector($doc['emb']),
];
// ...
$response = $client->bulk($params);

Packed dense vectors can provide significant bulk ingest performance improvements.

Cat.circuitBreaker (new API)

Esql.getView (new experimental API)

Esql.putView (new experimental API)

Esql.deleteView (new experimental API)

Indices.getSample (new experimental API)

Indices.getSampleConfiguration (new experimental API)

Indices.getAllSampleConfiguration (new experimental API)

Indices.putSampleConfiguration (new experimental API)

Indices.deleteSampleConfiguration (new experimental API)

Indices.getSampleStats (new experimental API)

Inference.putGroq (new API)

Inference.putNvidia (new API)

Inference.putOpenshiftAi (new API)

Ml.openJob

  • Added the timeout parameter (int|string)

Ml.stopDatafeed

  • Added the close_job parameter (bool)

knnSearch (removed API)

  • This endpoint is not supported since release 9.0 and was inadvertently left in this client.