Unverified50% confidenceOpinionExact time
FLOPs对应模型的智能/推理能力，参数对应模型的知识/存储能力，两者可在一定程度上解耦
1
Sources
50%
Confidence
Long-term
Relevance
8/27/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFLOPs（浮点运算次数）是衡量模型计算量的核心指标77% similarUnverifiedKnowledge distillation technology enables small models to extract a considerable proportion of capability from large models, undermining the validity of fixed compute thresholds as a proxy metric for capability.72% similarUnverified在推理阶段追加FLOPs能带来显著的能力提升70% similarUnverifiedSkills 机制通过上下文注入在不修改模型权重的前提下引导模型输出，相比一次性 Prompt Engineering 强调可复用性、版本管理和模块化组合67% similarUnverifiedThere is a positive correlation between a model's calibration capability and model scale; larger models can more finely distinguish between states of knowledge.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/810404API
curl https://kongchang.com/api/v1/knowledge/claims/810404MCP
get_claim(id=810404)