Unverified50% confidenceBenchmarkExact time
据HyperSAE作者披露，在Gemma-2-2B模型上传统欧几里得SAE的死亡潜变量比例高达3.8%
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/13/2026
First Seen
Valid until: 11/11/2026
Sources
Related Claims
UnverifiedOn the ARC-AGI-3 benchmark, GPT-5.6 Sol scored 8%, while most other models scored below 2%; Anthropic did not run Fable due to prohibitively high costs.57% similarUnverifiedAcademic research indicates that the impact of survivorship bias on backtesting results may be as high as 1–3 percentage points of return bias per year.56% similarUnverified在二维高斯分布下，1-σ椭圆包含真实状态概率约39.4%，2-σ约86.5%，3-σ约98.9%55% similarUnverifiedGLM-4.6 has an activation ratio of approximately 9%, which is a relatively aggressive sparse design among current MoE models.54% similarUnverified基于大型GWAS的LDL-PRS在欧洲祖先群体中可达约0.3-0.4的皮尔逊相关系数，而在非欧洲祖先群体中性能衰减明显52% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/742336API
curl https://kongchang.com/api/v1/knowledge/claims/742336MCP
get_claim(id=742336)