Related papers: Trust but Verify: Prover-Verifier Deliberation for…

Large Language Model (LLM) reasoning for complex tasks inherently involves a trade-off between solution accuracy and computational efficiency. The subsequent step of verification, while intended to improve performance, further complicates…

We present a principled approach to provide LLM-based evaluation with a rigorous guarantee of human agreement. We first propose that a reliable evaluation method should not uncritically rely on model preferences for pairwise evaluation, but…

Advances in training, post-training, and inference-time methods have enabled frontier reasoning models to win gold medals in math competitions and settle challenging open problems. Gaining trust in the responses of these models requires…

It's better to say "I can't answer" than to answer incorrectly. This selective prediction ability is crucial for NLP systems to be reliably deployed in real-world applications. Prior work has shown that existing selective prediction…

In many real-world scenarios, a single Large Language Model (LLM) may encounter contradictory claims-some accurate, others forcefully incorrect-and must judge which is true. We investigate this risk in a single-turn, multi-agent debate…

Data attribution methods aim to answer useful counterfactual questions like "what would a ML model's prediction be if it were trained on a different dataset?" However, estimation of data attribution models through techniques like empirical…

Multimodal large language models (MLLMs) achieve ever-stronger performance on visual-language tasks. Even as traditional visual question answering (VQA) benchmarks approach saturation, reliable deployment requires satisfying low error…

Speculative decoding is a promising approach for accelerating large language models. The primary idea is to use a lightweight draft model to speculate the output of the target model for multiple subsequent timesteps, and then verify them in…

Verifiers can improve language model capabilities by scoring and ranking responses from generated candidates. Currently, high-quality verifiers are either unscalable (e.g., humans) or limited in utility (e.g., tools like Lean). While LM…

As statistical analyses become more central to science, industry and society, there is a growing need to ensure correctness of their results. Approximate correctness can be verified by replicating the entire analysis, but can we verify…

The reliability of large language models remains a critical challenge, particularly due to their susceptibility to hallucinations and factual inaccuracies during text generation. Existing solutions either underutilize models'…

A trustworthy real-world prediction system should produce well-calibrated confidence scores; that is, its confidence in an answer should be indicative of the likelihood that the answer is correct, enabling deferral to an expert in cases of…

Large language models (LLMs) are increasingly used for tasks that implicitly reduce to Boolean satisfiability (SAT), yet their reasoning ability on SAT remains unclear. We present a systematic study of LLMs on 2-SAT and 3-SAT, together with…

Colluding language-model agents can hide coordination in messages that remain policy-compliant at the surface level. We present CLBC, a protocol where generation and admission are separated: a message is admitted to transcript state only if…

Standard evaluation metrics for machine learning -- accuracy, precision, recall, and AUROC -- assume that all errors are equivalent: a confident incorrect prediction is penalized identically to an uncertain one. For discrete commitment…

Test-time computation has become a primary driver of progress in large language model (LLM) reasoning, but it is increasingly bottlenecked by expensive verification. In many reasoning systems, a large fraction of verifier calls are spent on…

Same-model self-verification, prompting a model to audit its own predicted answer, is a plausible confidence signal for selective prediction, but its practical value remains unclear once strong likelihood-based baselines are taken…

In this paper we demonstrate a technique for developing high performance applications with strong correctness guarantees. We use a theorem prover to derive a high-level specification of the application that includes correctness invariants…

Large language models have achieved remarkable success on final-answer mathematical problems, largely due to the ease of applying reinforcement learning with verifiable rewards. However, the reasoning underlying these solutions is often…

A typical desideratum for quantifying the uncertainty from a classification model as a prediction set is class-conditional singleton set calibration. That is, such sets should map to the output of well-calibrated selective classifiers,…