Unverified50% confidenceTradeoffExact time
测试时计算扩展的边际收益并非线性，存在明显的收益递减区间
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
Unverified自回归缩放定律的边际收益正在递减，单纯增加模型规模和数据量带来的提升已微乎其微72% similarUnverifiedUSL中相干性开销与并发进程数N的平方成正比，而有效工作量仅与N成正比，超过临界值后导致负扩展70% similarUnverifiedThe computational cost of AI inference currently remains nearly linearly correlated with usage volume, and a significant inflection point of scale effects has yet to emerge.66% similarUnverifiedOpenAI's chart shows that at lower test-time compute levels, the model's success rate on this problem is close to zero, with occasional success only under extremely long reasoning chains.65% similarUnverified对同一测试用例执行多次调用统计成功比例得到通过率，比二元的通过/失败判断更能反映模型真实表现65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/523853API
curl https://kongchang.com/api/v1/knowledge/claims/523853MCP
get_claim(id=523853)