Unverified50% confidenceBenchmarkExact time
Representational depth can predict the magnitude of behavioral effects, achieving a Spearman correlation of 0.761 (p < 0.001) across all 17 models.
1
Sources
50%
Confidence
Long-term
Relevance
10/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAs the model scales from 8B to 70B, the predictive power of fixed output vector similarity for behavior actually worsens, with a paired mean correlation change of -0.080 (95% confidence interval [-0.127, -0.035]).67% similarUnverifiedThe same ordering—relation information preceding entity information—is observed across all 16 model–threshold combinations (threshold range 0.2 to 0.5).66% similarUnverifiedScale AI的研究团队发现，部分模型在被污染的基准上的表现比在新编写的同等难度题目上高出10-15个百分点64% similarUnverified参数量在1亿以下的轻量模型经领域数据微调后可达95%以上意图分类准确率，单次推理延迟控制在5-20ms以内62% similarUnverifiedMainstream GPU inference solutions based on the NVIDIA H100 can typically only achieve tens to one or two hundred tokens per second on models of equivalent scale.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/964072API
curl https://kongchang.com/api/v1/knowledge/claims/964072MCP
get_claim(id=964072)