Embedding size mismatch error

Google Cloud Build with AI Custom ML & MLOps
1

I’m working through the example VerteX AI doc matching

and am getting an embedding size mismatch error. Is this related to the approximateNeighborscount in the index_metadata being much higher than the documents specified?

nearestNeighborSearchOperationMetadata:

contentValidationStats:

  • invalidRecordCount: ‘5’

partialErrors:

  • embeddingId: ‘0’

errorMessage: Embedding field is missing

errorType: EMBEDDING_SIZE_MISMATCH

2

Yes, the “Embedding Size Mismatch” error likely arises due to a disparity between the specified embeddingSize in the index_metadata and the actual size of the embeddings in the documents.