Message': "module 'text2vec-transformers': invalid combination of properties

Support
1

I have the following collection that is receiving an error: invalid combination of properties

client.collections.create(
    "Blogs",
    description="this is testing db to understand weaviate",
    vectorizer_config=Configure.Vectorizer.text2vec_transformers(
        inference_url="http://t2v-transformers:8080", vectorize_collection_name=False
    ),
    properties=[
        Property(name="idCSV", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="gender", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="age", data_type=DataType.INT, skip_vectorization=True),
        Property(name="topic", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="sign", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="date", data_type=DataType.DATE, skip_vectorization=True),
        Property(name="text", data_type=DataType.TEXT_ARRAY, skip_vectorization=False),
    ],
)

Note: This is the v4 python api.

I have seen another issue that was similar. However, the above collection is vectorizing on the ‘text’ attribute.

How do I vectorize on a single attribute? How important is the vectorization of the collection name?

Thank you very much in advance.

2

Found the answer. configuring text2vec_transformer is a little different.

Here is a working updated version.

client.collections.create(
    "Blogs",
    description="this is testing db to understand weaviate",
    vectorizer_config=[
        Configure.NamedVectors.text2vec_transformers(
            name="text_vector",
            source_properties=["text"],
            vectorize_collection_name=False,
        )
    ],
    properties=[
        Property(name="idCSV", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="gender", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="age", data_type=DataType.INT, skip_vectorization=True),
        Property(name="topic", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="sign", data_type=DataType.TEXT, skip_vectorization=True),
        Property(name="date", data_type=DataType.DATE, skip_vectorization=True),
        Property(name="text", data_type=DataType.TEXT_ARRAY, skip_vectorization=False),
    ],
)
3

hi @Samuel ! Welcome to our community! :hugs:

What is the server version you are running?

Both codes gave me the same expected error. :thinking:

The problem here is that it must have at least one text element other than the array.

Because you have defined to not vectorize the collection name, you must provide one as a property.

If you change the module to openai, it will raise a more detailed error:

UnexpectedStatusCodeError: Collection may not have been created properly.! Unexpected status code: 422, with response body: {‘error’: [{‘message’: “module ‘text2vec-openai’: invalid properties: didn’t find a single property which is of type string or text and is not excluded from indexing. In addition the class name is excluded from vectorization as well, meaning that it cannot be used to determine the vector position. To fix this, set ‘vectorizeClassName’ to true if the class name is contextionary-valid. Alternatively add at least contextionary-valid text/string property which is not excluded from indexing”}]}.

Let me know if this helps!

THanks!

4

I see. Removing the following line worked.

Question: Why must the DB require one element other the array to work? I’m quite new at vectorDBs.

5

Hi!

While checking the code, it looks like this is a guardrail to make sure there is something vectorizable to pass to the embeddings service.

However, it only seems to check for a text data type, but a text_array should pass that validation as well.