Unverified50% confidenceFactExact time
The core idea of contrastive decoding is to run a strong model and a weak model simultaneously and subtract their output logits to amplify the strong model's distinctive tendencies relative to the weak model.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe core mechanism of DoubtLess is: let an auxiliary small model read the CoT already generated by the large model under two prompts, concise and verbose, respectively, then multiply the difference between the logit vectors under the two prompts by an intensity coefficient and add it to the logits of the main model's decoding at each step.74% similarUnverifiedHaving more capable models help align less capable models may be a viable path to solving the alignment scalability problem.69% similarUnverifiedThe core logic of a cascade routing strategy is 'never invoke a large model for what a small model can handle,' triggering a higher-cost model only when the small model's output confidence falls below a threshold.69% similarUnverified嵌入模型的选择对 RAG 系统性能的影响甚至超过生成模型68% similarUnverifiedSimple models (such as linear regression) typically exhibit high bias and low variance, while complex models (such as deep neural networks) exhibit low bias and high variance68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1028116API
curl https://kongchang.com/api/v1/knowledge/claims/1028116MCP
get_claim(id=1028116)