Unverified50% confidenceFactExact time
RT-2 is jointly trained on internet-scale image-text data and robot demonstration data, giving the model a certain degree of semantic generalization and commonsense reasoning ability.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRepresentative works such as Google's RT-2 and Stanford's VoxPoser have demonstrated that pre-trained vision-language models can significantly improve robots' zero-shot generalization capabilities76% similarUnverified实时在线学习要求智能体在连续流动的时间中同步完成感知、决策与参数更新，与大语言模型训练-推理分离的范式形成对比74% similarUnverifiedThe RT-X model was trained on Open X-Embodiment data and demonstrated positive transfer capabilities on unseen robots.73% similarUnverifiedRT-2使用了超过13万条机器人示教轨迹进行训练72% similarUnverified许多AI模型训练数据中仍夹杂着MV2的旧式写法，这是Modern Web Guidance技能包存在的核心价值70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1020692API
curl https://kongchang.com/api/v1/knowledge/claims/1020692MCP
get_claim(id=1020692)