incremental fine-tune an existing fine-tuned Gemini model in Vertex AI

Google Cloud Build with AI Custom ML & MLOps
1

I want to fine-tune a model on top of existing fine-tuned model. I am unable to do so.

python Vertex AI is throwing the following error

InvalidArgument**: 400 Base model projects/{project id}/locations/us-central1/endpoints/{endpoint id} is not supported.**

How can fine-tune Gemini models incrementally?

2

Hi @harshpreetsingh,

Welcome to Google Cloud Community!

It seems that you are facing an issue where the base model’s endpoint is not compatible with incremental fine-tuning.

Here are some potential ways to address your issue:

  • Check the model’s endpoint: Make sure the endpoint you’re using is correct and compatible with fine-tuning. It’s a good idea to verify the endpoint ID and project ID in your code to ensure they are accurate.
  • Verify model compatibility: Since not all models support this feature, ensure that the base model you’re using is compatible with incremental fine-tuning. Checking compatibility in advance will help prevent issues during the fine-tuning process and ensure optimal results.
  • Update your libraries: Make sure that you are using the latest version of the Vertex AI SDK for Python, as updates may address and resolve such issues. Keeping your libraries up to date guarantees compatibility with new features and improvements, minimizing the chances of encountering errors.