The authors assembled a team of 50+ industry experts to create 120 human-authored and verified tasks, each organized as a ladder of five levels (600 total), covering 89 professional roles across 12 domains.
Continual Learning Exam measures AI agents' ability to improve from experience
A large-scale benchmark with 120 human-authored tasks shows that even the best agent passes less than half of levels.
Independent
Haoyu Zhao · Sitian Tu · Yuxuan Huang · Huichi Zhou · Zhangyue Yin · Jiahao Huang · +13 more
Research Digest··2 min read
The Continual Learning Exam (CLE) is a new benchmark for evaluating how LLM agents improve over time through accumulated experience.
The Daily Edition
The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.