Sample, Scrutinize and Scale: Effective Inference-Time Search by Scaling Verification

• URL: http://arxiv.org/abs/2502.01839v1
• Date: Mon, 03 Feb 2025 21:31:07 GMT
• Title: Sample, Scrutinize and Scale: Effective Inference-Time Search by Scaling Verification
• Authors: Eric Zhao, Pranjal Awasthi, Sreenivas Gollapudi,
• Abstract summary: We study the scaling trends governing sampling-based search.<n>We find that simply scaling up a minimalist implementation that uses only random sampling and direct self-verification results in sustained performance improvements.<n>We identify two useful principles for improving self-verification capabilities with test-time compute.
• Score: 35.347715518778095
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Sampling-based search, a simple paradigm for utilizing test-time compute, involves generating multiple candidate responses and selecting the best one -- typically by verifying each response for correctness. In this paper, we study the scaling trends governing sampling-based search. Among our findings is that simply scaling up a minimalist implementation that uses only random sampling and direct self-verification results in sustained performance improvements that, for example, elevate the Gemini v1.5 Pro model's reasoning capabilities past that of o1-Preview on popular benchmarks. We partially attribute the scalability of sampling-based search to a phenomenon of implicit scaling, where sampling a larger pool of responses in turn improves verification accuracy. We further identify two useful principles for improving self-verification capabilities with test-time compute: (1) comparing across responses provides helpful signals about the locations of errors and hallucinations, and (2) different model output styles are useful for different contexts -- chains of thought are useful for reasoning but harder to verify. We also find that, though accurate verification can be elicited, frontier models demonstrate remarkably weak out-of-box verification capabilities and introduce a benchmark to measure progress on these deficiencies.

Related papers

• Rank1: Test-Time Compute for Reranking in Information Retrieval [45.356614696154075]
  Rank1 is the first reranking model trained to take advantage of test-time compute. We gather and open-source a dataset of more than 600,000 examples of R1 reasoning traces from queries and passages in MS MARCO.
  arXiv  Detail & Related papers  (2025-02-25T18:14:06Z)
• Diversified Sampling Improves Scaling LLM inference [31.18762591875725]
  DivSampling is a novel and versatile sampling technique designed to enhance the diversity of candidate solutions. Our theoretical analysis demonstrates that, under mild assumptions, the error rates of responses generated from diverse prompts are significantly lower compared to those produced by stationary prompts.
  arXiv  Detail & Related papers  (2025-02-16T07:37:58Z)
• Scaling Flaws of Verifier-Guided Search in Mathematical Reasoning [16.824343439487617]
  Large language models (LLMs) struggle with multi-step reasoning, where inference-time scaling has emerged as a promising strategy for performance improvement.<n>Verifier-guided search outperforms repeated sampling when sample size is limited by selecting and prioritizing valid reasoning paths.<n>As sample size increases, verifier-guided search exhibits diminishing advantages and eventually underperforms repeated sampling.
  arXiv  Detail & Related papers  (2025-02-01T02:08:49Z)
• Evaluation Agent: Efficient and Promptable Evaluation Framework for Visual Generative Models [51.067146460271466]
  Evaluation of visual generative models can be time-consuming and computationally expensive. We propose the Evaluation Agent framework, which employs human-like strategies for efficient, dynamic, multi-round evaluations. It offers four key advantages: 1) efficiency, 2) promptable evaluation tailored to diverse user needs, 3) explainability beyond single numerical scores, and 4) scalability across various models and tools.
  arXiv  Detail & Related papers  (2024-12-10T18:52:39Z)
• Optimal Algorithms for Augmented Testing of Discrete Distributions [25.818433126197036]
  We show that a predictor can indeed reduce the number of samples required for all three property testing tasks.<n>A key advantage of our algorithms is their adaptability to the precision of the prediction.<n>We provide lower bounds to indicate that the improvements in sample complexity achieved by our algorithms are information-theoretically optimal.
  arXiv  Detail & Related papers  (2024-12-01T21:31:22Z)
• Investigating the Impact of Hard Samples on Accuracy Reveals In-class Data Imbalance [4.291589126905706]
  In the AutoML domain, test accuracy is heralded as the quintessential metric for evaluating model efficacy. However, the reliability of test accuracy as the primary performance metric has been called into question. The distribution of hard samples between training and test sets affects the difficulty levels of those sets. We propose a benchmarking procedure for comparing hard sample identification methods.
  arXiv  Detail & Related papers  (2024-09-22T11:38:14Z)
• Large Language Monkeys: Scaling Inference Compute with Repeated Sampling [81.34900892130929]
  We explore inference compute as another axis for scaling, using the simple technique of repeatedly sampling candidate solutions from a model. Across multiple tasks and models, we observe that coverage scales with the number of samples over four orders of magnitude. In domains like coding and formal proofs, where answers can be automatically verified, these increases in coverage directly translate into improved performance.
  arXiv  Detail & Related papers  (2024-07-31T17:57:25Z)
• Neural Priming for Sample-Efficient Adaptation [92.14357804106787]
  We propose Neural Priming, a technique for adapting large pretrained models to distribution shifts and downstream tasks. Neural Priming can be performed at test time, even for pretraining as large as LAION-2B.
  arXiv  Detail & Related papers  (2023-06-16T21:53:16Z)
• Shortcomings of Top-Down Randomization-Based Sanity Checks for Evaluations of Deep Neural Network Explanations [67.40641255908443]
  We identify limitations of model-randomization-based sanity checks for the purpose of evaluating explanations. Top-down model randomization preserves scales of forward pass activations with high probability.
  arXiv  Detail & Related papers  (2022-11-22T18:52:38Z)
• POODLE: Improving Few-shot Learning via Penalizing Out-of-Distribution Samples [19.311470287767385]
  We propose to use out-of-distribution samples, i.e., unlabeled samples coming from outside the target classes, to improve few-shot learning. Our approach is simple to implement, agnostic to feature extractors, lightweight without any additional cost for pre-training, and applicable to both inductive and transductive settings.
  arXiv  Detail & Related papers  (2022-06-08T18:59:21Z)
• Boost Test-Time Performance with Closed-Loop Inference [85.43516360332646]
  We propose to predict hard-classified test samples in a looped manner to boost the model performance. We first devise a filtering criterion to identify those hard-classified test samples that need additional inference loops. For each hard sample, we construct an additional auxiliary learning task based on its original top-$K$ predictions to calibrate the model.
  arXiv  Detail & Related papers  (2022-03-21T10:20:21Z)
• MEMO: Test Time Robustness via Adaptation and Augmentation [131.28104376280197]
  We study the problem of test time robustification, i.e., using the test input to improve model robustness. Recent prior works have proposed methods for test time adaptation, however, they each introduce additional assumptions. We propose a simple approach that can be used in any test setting where the model is probabilistic and adaptable.
  arXiv  Detail & Related papers  (2021-10-18T17:55:11Z)
• Toward Scalable and Unified Example-based Explanation and Outlier Detection [128.23117182137418]
  We argue for a broader adoption of prototype-based student networks capable of providing an example-based explanation for their prediction. We show that our prototype-based networks beyond similarity kernels deliver meaningful explanations and promising outlier detection results without compromising classification accuracy.
  arXiv  Detail & Related papers  (2020-11-11T05:58:17Z)
• Few-shot Classification via Adaptive Attention [93.06105498633492]
  We propose a novel few-shot learning method via optimizing and fast adapting the query sample representation based on very few reference samples. As demonstrated experimentally, the proposed model achieves state-of-the-art classification results on various benchmark few-shot classification and fine-grained recognition datasets.
  arXiv  Detail & Related papers  (2020-08-06T05:52:59Z)
• Learning a Unified Sample Weighting Network for Object Detection [113.98404690619982]
  Region sampling or weighting is significantly important to the success of modern region-based object detectors. We argue that sample weighting should be data-dependent and task-dependent. We propose a unified sample weighting network to predict a sample's task weights.
  arXiv  Detail & Related papers  (2020-06-11T16:19:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.