Develops a sub-Weibull tail-aware information-theoretic framework yielding PAC-Bayes and chaining generalization bounds for RLHF and SGLD under heavy-tailed data.
Fort∈[0,m+log 2],F m(t)≤e t ≤e m+log 2 =2e m,hence, sup y≥0 ⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩ 2 ⌊ 2 θ ⌋∑︁ k=0 yθk k! −exp(y θ) ⎫⎪⎪⎪⎪⎬⎪⎪⎪⎪⎭ ≤2e m
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Tail-Aware Information-Theoretic Generalization for RLHF and SGLD
Develops a sub-Weibull tail-aware information-theoretic framework yielding PAC-Bayes and chaining generalization bounds for RLHF and SGLD under heavy-tailed data.