Verified70% confidenceFactExact time
仅用成功轨迹训练的模型往往存在分布外泛化的严重缺陷，遇到工具返回错误时便丧失纠偏能力
4
Sources
70%
Confidence
Long-term
Relevance
7/16/2026
First Seen
Sources
Related Claims
UnverifiedBehavior Cloning (BC) suffers from the covariate shift problem: during training the model only sees the state distribution of expert trajectories, and at deployment small errors can drive the robot into states not covered by the training set, leading to cascading failures.81% similarUnverified级联失败是指单次工具调用错误会污染后续上下文，导致模型在错误的中间状态上继续推理，最终使整个任务偏轨80% similarUnverified小模型在分布内高性能的同时存在分布外脆弱性，一旦推理时的工具描述措辞、参数命名或用户查询风格偏离训练集，成功率可能显著下滑80% similarVerifiedAfter geometric transformation, the transformation parameters must be recorded and the predicted boxes must be inverse-transformed to align them; ignoring this step will cause semantic misalignment in the consistency loss and harm training effectiveness79% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/527727API
curl https://kongchang.com/api/v1/knowledge/claims/527727MCP
get_claim(id=527727)