arXiv:2608. 14425v1 Announce Type: new Abstract: LLM evaluations often use fixed sampling budgets, testing every item the same number of times even after estimates are precise.
By Toby D. Pilditch
arXiv:2602. 05395v2 Announce Type: replace-cross Abstract: A simple strategy for improving LLM accuracy, especially in math and reasoning problems, is to sample multiple responses and submit the answer most consistently reached.
By Jingkai Huang, Will Ma, Zhengyuan Zhou
The paper proposes a classification-oriented adaptive sensing method that uses posterior sampling from diffusion models. It leverages the closed-form posterior covariance of a class-conditional Gaussian mixture model to separate within-class and between-class uncertainty, estimating these terms from diffusion posterior samples via calibrated soft classifier outputs. Experiments on MNIST and CIFAR-10 demonstrate that this approach can achieve better classification accuracy for a given measurement cost compared to reconstruction-oriented methods, while also quantifying the associated reconstruction quality.
By Andriy Enttsel, Maxime Rousselot, Vincent Corlay
The paper studies how voting over multiple large language model (LLM) responses can be optimized under a fixed call budget. It introduces a recoverability threshold that quantifies the gap between discovering a correct answer and ensuring it wins the plurality vote, showing that the candidate set can only grow while the set of reachable winners can only shrink. The authors also present a gold‑free locking certificate that identifies the earliest prefix where all remaining continuations produce the same fixed‑budget output, and demonstrate empirical gains in accuracy and call efficiency through input permutation and exact locking.
By Shaoang Li, Jian Li
arXiv:2601. 07094v2 Announce Type: replace-cross Abstract: Bayesian optimization (BO) iteratively fits a Gaussian process (GP) surrogate to accumulated evaluations and selects new queries via an acquisition function.
By Jiguang Li, Hengrui Luo
arXiv:2606. 15237v1 Announce Type: cross Abstract: Ensemble classifiers are predictive models that combine the results of simpler base models, often by majority vote.
By Joseph Kalman, Amit Moscovich