Looking for a way to vectorize a data object using WCS internal vectorizer module

I’m a Weaviate beginner, using java to create a desktop app and Weaviate Cloud Services (WCS) to digest text strings, so that a semantic search can be done for a question asked later on.

At the moment I’ve successfully connected to WCS, created a schema/class, and added some test data, however I’m not able to get it to generate the vector representation.

  WeaviateClass clazz = WeaviateClass.builder()
			.description("A document")
			.vectorizer("text2vec")//does nothing
			.properties(new ArrayList<Property>()
					add(Property.builder().dataType(new ArrayList<String>()
					}).description("text content").name("content").build());

I do not want to use a third party embedding service (like text2vec-openai), because I may need to do this for a large volume of data later.

I see these two statements which sort of imply this is not possible, unless I download and run weviate in a docker instance.

  • The text2vec-transformers module is not available on the WCS.
  • The text2vec-contextionary module is not available on the WCS.

Anyone know if built-in vectorization is possible with WCS ?

Update: I’ve successfully used the tensor flow java API and the “universal-sentence-encoder_4” model to create embeddings/vectors offline.