Unverified50% confidenceFactExact time
The square-root learning rate scaling rule only preserves the per-sample first-order drift and does not preserve the true convergence rate accumulated over multi-step iteration.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhen the spectral radius of the combined operator is less than 1, the iterative process converges stably to a fixed point.70% similarUnverifiedUnder the square-root learning rate scaling rule, the noise magnitude shrinks as 1/√k as the batch increases, so scaling the learning rate by √k exactly maintains the gradient's signal-to-noise ratio unchanged.70% similarUnverifiedWhen individually tuning dedicated learning rates for each control group and repeating experiments with three random seeds, the advantage of repairing only the encoder disappears, reaching parity with the other approaches.69% similarUnverifiedWith a fixed learning_rate, when batch_size scales linearly with the number of environments and num_epoch remains unchanged, the total displacement pushing the policy toward a new distribution in a single iteration may increase, manifesting as violent oscillation or even collapse of rewards in the early stages of training.68% similarUnverifiedNumerical discretization error arises from decomposing the theoretically continuous reverse-time stochastic differential equation process into a finite number of discrete time steps for execution, with fewer steps leading to larger per-step approximation errors.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1027460API
curl https://kongchang.com/api/v1/knowledge/claims/1027460MCP
get_claim(id=1027460)