Cohere's Multicultural Riddle Benchmark Enters Human Evaluation Phase
Cohere_Labs · x · 2026-08-01
Cohere Labs announced that its multicultural riddle benchmark project has entered the human evaluation phase, with over 61,000 model responses already evaluated.
Aiming to build one of the most multilingual and multicultural human evaluation datasets in NLP, the project is actively recruiting native speakers. They are specifically looking for contributors proficient in various dialects of Arabic, Luganda, Lumasaba, Swahili, and Spanish, alongside other languages.
Related event: Cohere's Multicultural Riddle Benchmark Enters Human Evaluation(2 posts)→
More from Research
- It's not gradient descent: burny_tech corrects the diagnosis to RLHF mode collapse — burny_tech · 2026-10-16
- A beginner-friendly introduction to causal inference kicks off, Part 1 — Nasereliver · 2026-10-16
- FBI method trains for Pass@K with one sample per update at 12.3% of the cost — EmtiyazKhan · 2026-10-16
- Berkeley researchers sample humanoid robot policies directly from weight space with diffusion — berkeley_ai · 2026-10-16
- OneWM: unified causal training shows world and action modeling improve each other — IAAI233 · 2026-10-16
- Baidu's Virtual Head Attention tames the KV cache, FLOPs, quality triangle in long-context LLMs — baidu · 2026-10-16