Unverified50% confidenceOpinionExact time
Evaluating within the training distribution easily overestimates model capability, while out-of-distribution benchmarks require the model to generalize to unseen scene and interaction combinations, better exposing the gap between 'looking right' and 'being physically right'.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
Unverified参数量相似的情况下，训练数据质量分布的差异往往比架构差异更能决定特定场景的表现上限79% similarVerifiedAlignment training optimizes the model's statistical behavior in known scenarios rather than establishing formally verifiable security boundaries, and may fail when faced with rare instruction combinations or role-playing frameworks outside the training distribution.77% similarVerifiedBehavioral cloning suffers from the distribution shift problem: during training the policy only encounters the state distribution covered by expert trajectories, and during deployment the accumulation of small deviations causes the actual state distribution visited to gradually diverge from the training distribution.76% similarUnverified对于MoE模型，路由层中两个专家得分极为接近时，微小数值差异可能翻转专家选择结果，导致分布式训练崩溃74% similarUnverifiedFine-tuning embedding models with contrastive learning can improve matching effectiveness in recruitment scenarios, with the key being the use of hard negatives (such as positions in the same field but at a different seniority level) rather than random negatives.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1012797API
curl https://kongchang.com/api/v1/knowledge/claims/1012797MCP
get_claim(id=1012797)