Benchmarking Intelligence Efficiency of LM Inference - View it on GitHub