AGI Eval

SmartGPT: Major Benchmark Broken - 89.0% on MMLU + Exam's Many Errors

AI Explained

A benchmark that the speaker has personally taught topics for.