LLM Digest

Story

arxiv_cs_ai · Nov 10, 2026 · paper

Source brief

BARS: Budget-Aware Sequential Sampling with Refusal for Majority Identification under Unknown Distributions

arxiv.orgNov 10, 2026
original source linked

In brief

Repeated sampling with majority selection is an effective means of enforcing self-consistency in large language models. Recently, a variant has also been introduced that terminates early to save costs when the output distribution is favorable. But it still cannot yield a correct

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

BARS: Budget-Aware Sequential Sampling with Refusal for Majority Identification under Unknown Distributions

Earlier in this thread 4 items

On Turn-Level Best-of-N Sampling for Scaling LLM Agents

Real-time zero-knowledge LLM identification from physical telemetry

Seq@K: Evaluating Language Models Under Sequential Feedback

When Visual Evidence Is Insufficient: How Instructions, Post-Training, and Internal Activations Shape Refusal in Multimodal Language Models