Unverified50% confidenceOpinionExact time
In credit approval, anomalous applications are out-of-distribution cases insufficiently covered by the training data, and the more confident the model is, the higher the cost of error.
1
Sources
50%
Confidence
Long-term
Relevance
10/17/2026
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75% similarUnverified端到端模型的核心优势在于避免模块间信息损失和误差累积，但代价是可解释性低、对训练数据分布敏感、难以应对长程推理任务74% similarVerifiedThe probability of hallucination increases significantly when contradictory information exists in the training data or when the model is asked about domains that are sparsely represented in the training data.73% similarUnverifiedBias introduced by a single defective data point can leak into seemingly unrelated capability dimensions, causing the model to simultaneously regress across multiple independent metrics such as deceptiveness, compliance, and risk aversion.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1055034API
curl https://kongchang.com/api/v1/knowledge/claims/1055034MCP
get_claim(id=1055034)