Unverified50% confidenceFactExact time
GAE通过引入λ参数在偏差和方差之间折中，λ=1退化为蒙特卡洛估计，λ=0退化为单步TD估计，实践中λ=0.95左右通常能取得良好平衡
1
Sources
50%
Confidence
Long-term
Relevance
8/5/2026
First Seen
Sources
Related Claims
UnverifiedGradient descent updates parameters through the iterative formula θ = θ - α∇J(θ), where α is the learning rate and ∇J(θ) is the vector of partial derivatives of the loss function with respect to the parameters.61% similarUnverifiedThe error introduced by uniform quantization approximately follows a uniform distribution, with a variance of Δ²/12 (where Δ is the quantization step size).60% similarUnverifiedFocal Loss adds a modulating factor (1-p_t)^γ before the standard cross-entropy loss, with γ typically set to 2, causing the model to focus on hard examples.60% similarUnverifiedThe gradient descent update rule is v → v − η∇C; since η > 0 and ‖∇C‖² ≥ 0, each update is guaranteed to decrease the cost, ΔC ≤ 0.60% similarUnverifiedThe advantage function A(s,a) = Q(s,a) - V(s); in practical PPO implementations, Generalized Advantage Estimation (GAE) is typically used, with typical settings of λ = 0.95 and γ = 0.99.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/691005API
curl https://kongchang.com/api/v1/knowledge/claims/691005MCP
get_claim(id=691005)