Unverified50% confidenceFactExact time
State perturbation uses Gaussian noise injection (sigma=0.005) to prevent the model from learning fragile 1-dimensional linear trajectories.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/9/2026
First Seen
Valid until: 12/8/2026
Sources
Related Entities
Related Claims
VerifiedThe generalization error of a model equals the squared bias plus variance plus irreducible noise.67% similarUnverifiedIf samples are too easy the model has nothing to learn; if they are too difficult, noisy learning signals may interfere with training; the most effective learning occurs in the region slightly above current capability but still within reach.66% similarUnverifiedε-prediction is prone to numerical instability at large embedding dimensions, because when α_bar_t approaches zero the scaling factor 1/sqrt(α_bar_t) tends to infinity, amplifying prediction errors.66% similarUnverifiedThe cosine annealing learning rate scheduling strategy was proposed by Loshchilov and Hutter in 2017 and smoothly decays the learning rate from its initial value to near zero following a cosine function.65% similarUnverifiedThe residual structure makes F=0 the default option—as long as the F(x) weights are close to zero, the residual block degenerates into an identity mapping, thereby resolving the optimization difficulty of SGD in learning identity mappings.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/884839API
curl https://kongchang.com/api/v1/knowledge/claims/884839MCP
get_claim(id=884839)