Hi everyone,

I am using KServe to deploy models and also utilizing the explainer functionality. I noticed that the explainer pod primarily relies on API calls to the predictor pod to retrieve predictions. However, for certain metrics available in libraries like ART (e.g., Loss Sensitivity), direct access to the model itself is required.

Is there a way to ensure that the model is also available in the explainer pod, so that such metrics can be computed directly without needing to make calls to the predictor API? Any guidance on how to achieve this or examples of similar configurations would be greatly appreciated.

Thanks in advance for your help!