Unverified75% confidenceFactExact time
Simply pushing up the total number of parameters under a fixed TPP will actually hurt reasoning ability, whereas increasing the number of activated experts can reliably improve reasoning.
1
Sources
75%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
Rethinking Scaling Laws: Parameters Are Not the Only Answer
twitterjietang8/19/2026
Related Entities
Related Claims
UnverifiedChain-of-Thought prompting can improve reasoning quality, and contrastive prompting can reduce confirmation bias67% similarUnverified不同服务商对推理的支持方式各不相同：有的通过reasoning_effort参数控制思考深度，有的通过显式thinking开关，有的支持为思考过程设定token上限66% similarUnverifiedThe higher the reasoning effort level, the stronger the reasoning capability but the faster token consumption, representing a trade-off between cost and intelligence.65% similarUnverified提示词的清晰度直接决定输出质量，模糊或依赖特定背景知识的提示往往难以获得理想结果62% similarUnverified人类标注者在评估答案时存在信息越丰富越好的认知偏差，导致奖励模型将篇幅与质量隐性挂钩，是模型冗余的成因之一62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/796932API
curl https://kongchang.com/api/v1/knowledge/claims/796932MCP
get_claim(id=796932)