Radiology benchmark RadLE is 100% human-verified; RadLE 2.1 results coming at RSNA
DrDatta_AIIMS · x · 2026-11-19
Responding to talk about flawed evals, radiologist DrDatta (AIIMS) shares that Radiology's Last Exam (RadLE) from CRASHLabAI is a 100% human-verified benchmark where two radiologists agreed on every unambiguous case, curated from 10k+ images.
- He declines to share updated results hastily until his radiologist team confirms they are 100% true
- RadLE 2.1 results are delayed a few more days, with a hint they'll be unveiled at RSNA 2026
More from Research
- MIT CSAIL paper: frozen LLM agents design their own self-improvement without being told how — omarsar0 · 2026-11-19
- Anthropic Says Claude Solved Two Open Problems in Black Hole Physics — ghumare64 · 2026-11-19
- Shu Ding's essay 'Metaphors': protocols, dimensional analysis, and Curry-Howard — shuding · 2026-11-19
- Why your fine-tuning job takes longer than the FLOPs say — Hugging Face Blog · 2026-11-19
- CyberArena benchmark: Lithos defense agent wins 3-0, blocking 12 of 12 attacks — JiaZhihao · 2026-11-19
- Tianqi Chen's team launches TIRx Rust, a cross-language ML compiler base for agents — JiaZhihao · 2026-11-19