Reific发布实时数据强化学习环境
摘要
YC F26项目Reific构建强化学习环境，让AI在随时间变化的实时数据流中推理；最强前沿模型平均得分均低于50%，在其环境训练的开源模型对未见任务完全解决率从25.6%升至39.5%。
原文片段
Reific - RL environments for superhuman reasoning over time Training AI to reason over live data no human could follow <https://youtu.be/ZVI-Dje5Zss> \ Reific (YC F26) builds RL environments for superhuman reasoning over time. AI has learned to reason over text and code. But most of what matters i…