Unverified50% confidenceFactExact time
The classical baseline variance-reduction technique exploits the property that subtracting any quantity that does not depend on the action does not change the expected gradient.
1
Sources
50%
Confidence
Long-term
Relevance
9/20/2026
First Seen
Sources
Related Entities
Related Claims
Unverified从回报中减去任何不依赖于动作的基线函数b(s)不会改变梯度估计的期望值但可以降低方差75% similarUnverified为几个百分点的榜单差距进行定向优化是本末倒置，评估模型时不应盲目迷信榜单排名72% similarUnverified在实践中优先明确约束与停止条件这两个负向维度，比精心描述正向目标更能快速压制冗余输出72% similarUnverified传统梯度下降方法（如SGD、Adam）要求目标函数可微且连续，在面对离散、非凸、多模态问题时容易陷入局部最优72% similarVerifiedGradient descent iteratively adjusts weights in the opposite direction of the loss function's gradient, causing the prediction error to converge progressively.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/937387API
curl https://kongchang.com/api/v1/knowledge/claims/937387MCP
get_claim(id=937387)