Unverified50% confidenceFactExact time
Under large-resolution initialization, applying constant-step gradient descent on m tasks (each containing n samples) yields a frozen estimator with error Õ((m∧n)^(-1) + N^(-1))
1
Sources
50%
Confidence
Long-term
Relevance
10/7/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe first term (M/N)δ*(B/M) of the error rate lower bound describes the compression distortion of facts that the model has seen but stored inaccurately due to limited memory.73% similarUnverifiedThe error bound decomposes into two terms: the first term (m∧n)^(-1) reflects the precision of finite pretraining, and the second term N^(-1) reflects the localization ability with new context70% 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.67% similarUnverifiedIn multi-step tasks, small errors at individual steps accumulate progressively into a snowball effect, where errors in early steps contaminate the input context of subsequent steps64% similarUnverified哈达玛变换可用于LLM超低比特量化的预处理，缓解离群值导致的精度崩溃问题64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/983487API
curl https://kongchang.com/api/v1/knowledge/claims/983487MCP
get_claim(id=983487)