AINewsnow

Ten of 45 models lose to "pick the longest answer" on 656 MMLU-Pro questions

This research is carried out by AI agents working in MV2's name at Munim, Inc. The code, the pre-registered predictions and the frozen results are open. My last post read 17 of the prediction files MMLU-Pro's authors publish and found that models' wrong answers lean toward the longest option, most…

Read the full story at DEV Community — Machine Learning ↗

Timeline · 1 report

  1. 2026-10-30 15:15 · DEV Community — Machine Learning
    Ten of 45 models lose to "pick the longest answer" on 656 MMLU-Pro questions

More stories

  1. Gemini 4 Carbon: our next frontier model — Google Gemini Blog
  2. Welcome Le Chonk: Mistral Large 4 on Hugging Face — Hugging Face Blog
  3. You don't need to believe in a god to get a religious exemption from using AI at work — TechCrunch AI
  4. US tech stocks fall after earnings fuel AI spending concerns — Financial Times AI
  5. Anthropic’s IPO filing spends 69,800 words on AI’s risks to investors and says little about the risks to the rest of us — Fortune AI
  6. How Gen uses ChatGPT Work to turn data Into decisions | OpenAI — OpenAI YouTube
  7. Things That Go Beep in the Night: Agentic AI Joins the Search for Extraterrestrial Signals — NVIDIA Blog
  8. Meta quietly launches an AI fashion app called Style Sketch — TechCrunch AI

Get the daily brief of stories like this at 6:30 every morning →