VertexModelGardenLlama using publisher model?

[wonhs91]

llm = VertexModelGardenLlama(
  model=MODEL,
  credentials=credentials,
  location=REGION,
  project=PROJECT_ID
)

I try to use llama3.1 model like above, and it throws the following error

HTTPStatusError: Error response 400 while fetching
  "error": {
    "code": 400,
    "message": "Project `XXXXXX` is not allowed to use Publisher Model `projects/XXXXXX/locations/REGION/publishers/meta/models/MODEL`",
    "status": "FAILED_PRECONDITION"
  }

The curl command using the same settings (model, credentials, locations, project_id) works. So I don't think it is the server issue.

[lkuligin]

could you share a full snippet, please?
it's working fine on my side.

[nndnha]

@lkuligin

llm = VertexModelGardenLlama(model="meta/llama-3.1-405b-instruct-maas")

ValidationError: 1 validation error for VertexModelGardenLlama
  Value error, Model meta/llama-3.1-405b-instruct-maas is not supported yet! [type=value_error, input_value={'model_name': 'meta/llam... 'default_metadata': ()}, input_type=dict]

langchain-google-vertexai==2.0.7

[jaroslav-bellus]

Could someone please add meta/llama-3.1-8b-instruct-maas, meta/llama-3.1-70b-instruct-maas, and meta/llama-3.1-405b-instruct-maas to _LLAMA_MODELS? Thanks.

cc: @lkuligin