In the REST APIs, users can specify the resources that the pod would use when running the lm-eval job