topic Re: Gemini though Mosaic Gateway in Machine Learning

Gemini though Mosaic Gateway

roman_belkin — Mon, 21 Oct 2024 20:11:49 GMT

I am trying to configure the Gemini Vertex API in Databricks. In simple Python code, everything works fine, which indicates that I have correctly set up the API and credentials.

Error message:
{"error_code":"INVALID_PARAMETER_VALUE","message":"INVALID_PARAMETER_VALUE: Failed to generate access token for Google VertexAI. Please check the private key for the external model gemini-pro."}

I tried using different model names, but I didn't have any luck.

Re: Gemini though Mosaic Gateway

AnthSaou_48676 — Tue, 19 Nov 2024 16:31:01 GMT

have you solved this problem? i am still running into the same issue

Re: Gemini though Mosaic Gateway

roman_belkin — Tue, 19 Nov 2024 19:21:49 GMT

No, it seems they gave up

Re: Gemini though Mosaic Gateway

sam-p — Mon, 10 Mar 2025 17:04:48 GMT

With support from a helpful Databricks employee, we found out that the problem was that the `private_key` / `private_key_plaintext` field needs to be the entire JSON object that GCP creates for the service account not just the private key string from it as the documentation lead us to believe