Unverified50% confidenceFactExact time
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
1
Sources
50%
Confidence
Long-term
Relevance
8/5/2026
First Seen
Sources
Related Claims
VerifiedGradient descent iteratively adjusts weights in the opposite direction of the loss function's gradient, causing the prediction error to converge progressively.69% 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67% similarUnverifiedSmaller models often show exponential rather than linear degradation in performance on long reasoning chain tasks.65% similarUnverifiedNegative constraints narrow the probability space by explicitly excluding undesired outputs, are equally important as positive instructions, and can reduce deviations caused by the model's creative improvisation.64% similarUnverified结构化、类型强约束的工具响应可以显著降低模型在解析阶段产生误解的概率62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/690911API
curl https://kongchang.com/api/v1/knowledge/claims/690911MCP
get_claim(id=690911)