My Benchmarks

Every benchmark and eval I've built or proposed: model behaviour and AI safety, speech recognition, image generation and local inference. Each row links the code repo and the test prompts or dataset. Read live from benchmarks.json.