Primary source腾讯混元
View sourceEvent ·Single source
腾讯发布探索基准测试ExplorationBench
腾讯、复旦大学和清华大学联合推出探索基准测试ExplorationBench,包含两个沙盒环境AlienCode和AlienLogic,共101个任务。测试显示,获取反馈比单独思考更有效,最佳系统在四轮反馈后准确率从15.7%提升至89.0%。同一系统在不同沙盒中的表现差异巨大,准确率从5.7%到79.0%不等。
Current read
腾讯团队推出探索基准测试,评估AI系统发现规则的能力,10个前沿模型测试结果揭示重要发现。
2 sources67° AI heatLast updated 2026/9/30 13:29:01
Evidence
2 sourcesRelated source 1Hunyuan
View sourceConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.