count_token for embedding model is failing for 'rest' transport with a 400 BadRequest error.

[vi3k6i5]

Thanks for stopping by to let us know something could be better!

Environment details

• OS type and version: Mac Os, latest
• Python version: 3.9.6
• pip version: 21.2.4
• google-cloud-aiplatform: 1.69.0

Steps to reproduce

1. Initialise the env over "Rest" transport.
2. Get pretrained embedding model
3. call count_token mothod on the model object.

Code example

vertexai.init(
    project='project',
    api_endpoint='<api_endpoint>',
    transport='rest',
    credentials=get_credentials(),
)
embedding_model = TextEmbeddingModel.from_pretrained(model_name='text-embedding-004')
resp = embedding_model.count_token(['Hello world'])

Stack trace

...
raise core_exception.from_http_response(response)
google.api_core.exception.BadRequest: 400 POST https://<api_endpoint>/v1beta1/projects/<project>/locations/us-central1/publishers/google/models/test-embedding-004@default:countTokens

Thanks!

[jaycee-li]

Hi @vi3k6i5 , REST transport is working on my side (see the following code). Maybe it's an issue with your api endpoint?

import vertexai
from vertexai.language_models import TextEmbeddingModel

vertexai.init(
   project='my-project',
   api_transport='rest',
)
embedding_model = TextEmbeddingModel.from_pretrained(model_name='text-embedding-004')
resp = embedding_model.count_tokens(['Hello world'])