I am trying to do a very basic call to an LLM via GCP VertexAI. I can’t use Google AI Studio, as that is not an option for my organization, it must be VertexAI.
My service account can connect via Google Authenticator, but I will come back to this. When I connect VertexAI Connector, and set my project id, it fails to connect. I am curious if others have faced this.
It’s also odd that we have to define scope when using a service account when using Google Authenticator, as we itemize the access via the service account already. The only options for Google Authenticator are Analytics, Drive and Sheets.
To step back, I suppose I am curious if anyone has been able to get an LLM call to work with Vertex, and if so, can you point me to an example workflow?
@Brock_Tibert An example is here but I have not tried myself
Connecting KNIME and Google Cloud is not so easy it seems too get the right combination of permissions and settings in the nodes.
My approach would be to battle it out with ChatGPT and Gemini itself positing the error message and then contacting your IT and knime support until you find the right combination. And then tell us about.
With that I was able to get some connections right but a lot of things might depend on specific settings in the service accounts and company policies.
