Unverified50% confidenceFactExact time
Within the same training batch, rollout groups where all attempts either fail or succeed may appear; in such cases the group-relative reward signal becomes invalid, as there is no gradient information reflecting relative quality among samples within the group.
1
Sources
50%
Confidence
Long-term
Relevance
10/7/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在稠密检索中，批内负样本训练要求批次包含有信息量的负样本，随机采样会导致训练损失快速收敛但泛化性能差79% similarUnverifiedWhen all sampled outputs in a group are incorrect, the group-relative advantage collapses to zero, causing the model to receive no effective gradient feedback in high-uncertainty scenarios.76% similarVerified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76% similarUnverifiedBehavior 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.71% similarUnverifiedWhen questions are too easy and the model can answer them correctly almost every time, the reward signal tends toward a constant, parameter updates approach zero, and training effectively stops71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/982889API
curl https://kongchang.com/api/v1/knowledge/claims/982889MCP
get_claim(id=982889)