LLM Digest

Story

arxiv_cs_ai · Nov 2, 2026 · paper

Source brief

LAND: Learning an Adaptive Number of Draws for Bayesian Best-of-N Selection

arxiv.orgNov 2, 2026
original source linked

In brief

Best-of-N decoding improves generation quality by sampling multiple candidates and selecting the best, but it allocates equal compute to sources regardless of how much they benefit from additional sampling: the expected benefit is small for sources with low-variance score distrib

Read the original at arxiv.org →Open in live feed Read that day’s brief

LAND: Learning an Adaptive Number of Draws for Bayesian Best-of-N Selection

Earlier in this thread 4 items

TradeWar: A Benchmark for Execution-Grounded Trade Negotiation with Adaptive LLM Agents

When Does Test-Time Search Help World Action Models? An Empirical Audit of Candidate Selection and Simulator Integrity

Game-Theoretic Solvers with Bayesian Opponent Modeling and Moral Rhetoric Framing for GLEE

Channel Selection for Mixture-of-Experts Decoding Needs No Calibration Data