CrewAI Memory [Vertex AI]

Hi, I can’t find a solution to use memory within VertexAI authentication
I’ve tried multiple things like changing the names of the values on the vertex_embedder_config dictionary, removing the api_key row and I’m still stuck on the same error:

ERROR: Error during short_term save: Expected Embedings to be non-empty list or numpy array, got in add.

vertex_embedder_config = {

    "provider": "vertexai",

    "config": {

        "project_id": vertex_ai_project,

        "region": vertex_ai_location,
        "api_key": service_account_key_string

        \# textembedding-gecko,  text-multilingual-embedding-002, models/text-embedding-004

        "model_name": "textembedding-gecko"

    } 

}

crew_email_rag_storage = ShortTermMemory(

        storage = RAGStorage(

        type = 'short_term',

        embedder_config=vertex_embedder_config,

        path=crew_email_memory_path

    )

    )

crew_email_rag_storage_entity = EntityMemory(

    storage=RAGStorage(

    type = 'short_term',

    embedder_config=vertex_embedder_config,

    \# type="short_term", # Ou "long_term" se aplicável

    path=crew_email_memory_path

)

)
crew_email_long_term_memory = LongTermMemory(

        storage=LTMSQLiteStorage(db_path=crew_email_memory_path)

    )

self.crew_email = Crew(

            agents=\[self.agent….\],

            tasks=\[self.task….\],

            memory = True,

            short_term_memory= crew_email_rag_storage,

            entity_memory= crew_email_rag_storage_entity,

            long_term_memory=crew_email_long_term_memory,

            cache=True,

            verbose=False

        )