Joke: most models are apparently RL-trained on LisanBench these days
scaling01 · x · 2026-10-17
scaling01 jokes that "most models are RL'd on LisanBench nowadays" — a meme-benchmark jab at labs overfitting post-training to viral meme evals, poking at benchmark contamination.
More from Fun
- Tesla Roadster's cold-gas thrusters run on plain compressed air and recharge while driving — altryne · 2026-10-17
- Researcher slams OpenAI and Anthropic as grifters who invented nothing of significance — examachine · 2026-10-17
- Selling 10 iPhones could fund enough GPUs for a local AI lab — AIFlow_ML · 2026-10-17
- First Deepfake Candidate Caught in Interview: the Hand-over-Face Test — dotey · 2026-10-17
- Star Wars Moke Mocks AI Safetyists: 'Sand Arranged Wrong Kills Everyone' — NathanpmYoung · 2026-10-17
- Fable 5.5 Video Covers 40,000 Years of Art History in 15 Seconds — With a Cat Side-Quest — FinanceYF5 · 2026-10-17