Unverified50% confidenceFactExact time
Fine-tuning on data with narrow defects can sometimes cause models to become broadly misaligned.
1
Sources
50%
Confidence
Long-term
Relevance
9/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe Hunyuan model tends to oversimplify and smooth, with noticeable jagged artifacts, indicating that resolution is still insufficient.79% 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.78% similarUnverifiedHigh bias means the model is too simple and leads to underfitting, while high variance means the model has learned noise and leads to overfitting, and the two tend to trade off against each other.78% similarVerified若用含测试集的全量数据计算均值和标准差，会导致预处理泄漏，使评估指标虚高，无法真实反映模型泛化能力77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/961161API
curl https://kongchang.com/api/v1/knowledge/claims/961161MCP
get_claim(id=961161)