Unverified50% confidenceFactExact time
最大似然估计训练目标等价于最小化模型分布与训练数据经验分布之间的KL散度
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
Sources
元认知反馈：用强化学习让大模型准确表达不确定性
hackernewshackernews7/7/2026
Related Claims
Unverified批次归一化在训练模式使用当前批次统计量，在推理模式使用训练过程中累积的移动平均值78% similarVerifiedThrough higher-quality training data, more optimized training strategies, and more refined post-training alignment, smaller-scale models can fully surpass larger-scale but lower-quality-trained models on practical tasks.77% similarUnverified小模型通过知识蒸馏通常比直接在标注数据上训练精度高出2-5个百分点75% similarUnverifiedA 2017 study by Keskar et al. showed that large-batch training tends to converge to sharp minima corresponding to worse generalization, while small-batch training tends toward flat minima.74% similarUnverified参数量相似的情况下，训练数据质量分布的差异往往比架构差异更能决定特定场景的表现上限73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/319108API
curl https://kongchang.com/api/v1/knowledge/claims/319108MCP
get_claim(id=319108)