Odyssey-3 Pro刷新Physics-IQ基准,具身智能展现泛化纠错能力
Huge release from @odysseyml.
推荐理由
Odyssey-3 Pro在物理理解与具身智能上取得突破,特别是模型在无监督情况下展现的纠错泛化能力值得关注,建议关注其技术报告细节。
Huge release from @odysseyml.
@odysseyml 发布了重磅更新。
Odyssey-3 Pro sets a new top score on Physics-IQ Verified, a benchmark that asks models to continue videos of real physics experiments.
Odyssey-3 Pro 在 Physics-IQ Verified 基准测试中创下新高,该基准要求模型能够续播真实物理实验的视频。
The robotics results stood out to me. With tens of hours of demos, the robot arm recovered from a missed grasp, a behavior that never appeared in those demos.
机器人领域的成果令我印象深刻。仅凭数十小时的演示数据,机械臂就能从抓取失误中恢复,而这一行为从未出现在那些演示数据中。
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力