Project MPG: towards a generalized performance benchmark for LLM capabilities

Open in new window