Leveraging LLM Inconsistency to Boost Pass@k Performance

• URL: http://arxiv.org/abs/2505.12938v2
• Date: Tue, 20 May 2025 14:22:15 GMT
• Title: Leveraging LLM Inconsistency to Boost Pass@k Performance
• Authors: Uri Dalal, Meirav Segal, Zvika Ben-Haim, Dan Lahav, Omer Nevo,
• Abstract summary: Large language models (LLMs) achieve impressive abilities in numerous domains, but exhibit inconsistent performance in response to minor input changes.<n>We introduce a novel method for leveraging models' inconsistency to boost Pass@k performance.<n>Specifically, we present a "Variator" agent that generates k variants of a given task and submits one candidate solution for each one.
• Score: 3.797421474324735
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Large language models (LLMs) achieve impressive abilities in numerous domains, but exhibit inconsistent performance in response to minor input changes. Rather than view this as a drawback, in this paper we introduce a novel method for leveraging models' inconsistency to boost Pass@k performance. Specifically, we present a "Variator" agent that generates k variants of a given task and submits one candidate solution for each one. Our variant generation approach is applicable to a wide range of domains as it is task agnostic and compatible with free-form inputs. We demonstrate the efficacy of our agent theoretically using a probabilistic model of the inconsistency effect, and show empirically that it outperforms the baseline on the APPS dataset. Furthermore, we establish that inconsistency persists even in frontier reasoning models across coding and cybersecurity domains, suggesting our method is likely to remain relevant for future model generations.

Related papers

• Continuous Bayesian Model Selection for Multivariate Causal Discovery [22.945274948173182]
  We propose a scalable algorithm leveraging a continuous relaxation of the discrete model selection problem.<n>We employ the Causal Gaussian Process Conditional Density Estimator (CGP-CDE) as a Bayesian non-parametric model.<n>This matrix is then optimised using the marginal likelihood and an acyclicity regulariser, giving the maximum a posteriori causal graph.
  arXiv  Detail & Related papers  (2024-11-15T12:55:05Z)
• Explanatory Model Monitoring to Understand the Effects of Feature Shifts on Performance [61.06245197347139]
  We propose a novel approach to explain the behavior of a black-box model under feature shifts. We refer to our method that combines concepts from Optimal Transport and Shapley Values as Explanatory Performance Estimation.
  arXiv  Detail & Related papers  (2024-08-24T18:28:19Z)
• Fine-Tuning with Divergent Chains of Thought Boosts Reasoning Through Self-Correction in Language Models [63.36637269634553]
  We present a novel method of further improving performance by requiring models to compare multiple reasoning chains. We find that instruction tuning on DCoT datasets boosts the performance of even smaller, and therefore more accessible, language models.
  arXiv  Detail & Related papers  (2024-07-03T15:01:18Z)
• Amortizing intractable inference in large language models [56.92471123778389]
  We use amortized Bayesian inference to sample from intractable posterior distributions. We empirically demonstrate that this distribution-matching paradigm of LLM fine-tuning can serve as an effective alternative to maximum-likelihood training. As an important application, we interpret chain-of-thought reasoning as a latent variable modeling problem.
  arXiv  Detail & Related papers  (2023-10-06T16:36:08Z)
• Model-based Offline Policy Optimization with Adversarial Network [0.36868085124383626]
  We propose a novel Model-based Offline policy optimization framework with Adversarial Network (MOAN) Key idea is to use adversarial learning to build a transition model with better generalization. Our approach outperforms existing state-of-the-art baselines on widely studied offline RL benchmarks.
  arXiv  Detail & Related papers  (2023-09-05T11:49:33Z)
• Improving Transferability of Adversarial Examples via Bayesian Attacks [84.90830931076901]
  We introduce a novel extension by incorporating the Bayesian formulation into the model input as well, enabling the joint diversification of both the model input and model parameters. Our method achieves a new state-of-the-art on transfer-based attacks, improving the average success rate on ImageNet and CIFAR-10 by 19.14% and 2.08%, respectively.
  arXiv  Detail & Related papers  (2023-07-21T03:43:07Z)
• Lightweight reranking for language model generations [26.942659041383596]
  We present a novel approach for reranking Large Language Models (LLMs) generations. Unlike other techniques that might involve additional inferences or training a specialized reranker, our approach relies on easy to compute pairwise statistics. We show strong improvements for selecting the best k generations for code generation tasks as well as robust improvements for the best generation for the tasks of autoformalization, summarization, and translation.
  arXiv  Detail & Related papers  (2023-07-11T17:51:48Z)
• Making Substitute Models More Bayesian Can Enhance Transferability of Adversarial Examples [89.85593878754571]
  transferability of adversarial examples across deep neural networks is the crux of many black-box attacks. We advocate to attack a Bayesian model for achieving desirable transferability. Our method outperforms recent state-of-the-arts by large margins.
  arXiv  Detail & Related papers  (2023-02-10T07:08:13Z)
• Model Selection for Bayesian Autoencoders [25.619565817793422]
  We propose to optimize the distributional sliced-Wasserstein distance between the output of the autoencoder and the empirical data distribution. We turn our BAE into a generative model by fitting a flexible Dirichlet mixture model in the latent space. We evaluate our approach qualitatively and quantitatively using a vast experimental campaign on a number of unsupervised learning tasks and show that, in small-data regimes where priors matter, our approach provides state-of-the-art results.
  arXiv  Detail & Related papers  (2021-06-11T08:55:00Z)
• Attentional Prototype Inference for Few-Shot Segmentation [128.45753577331422]
  We propose attentional prototype inference (API), a probabilistic latent variable framework for few-shot segmentation. We define a global latent variable to represent the prototype of each object category, which we model as a probabilistic distribution. We conduct extensive experiments on four benchmarks, where our proposal obtains at least competitive and often better performance than state-of-the-art prototype-based methods.
  arXiv  Detail & Related papers  (2021-05-14T06:58:44Z)
• MASSIVE: Tractable and Robust Bayesian Learning of Many-Dimensional Instrumental Variable Models [8.271859911016719]
  We propose a general and efficient causal inference algorithm that accounts for model uncertainty. We show that, as long as some of the candidates are (close to) valid, without knowing a priori which ones, they collectively still pose enough restrictions on the target interaction to obtain a reliable causal effect estimate.
  arXiv  Detail & Related papers  (2020-12-18T10:06:55Z)
• Meta-Learned Confidence for Few-shot Learning [60.6086305523402]
  A popular transductive inference technique for few-shot metric-based approaches, is to update the prototype of each class with the mean of the most confident query examples. We propose to meta-learn the confidence for each query sample, to assign optimal weights to unlabeled queries. We validate our few-shot learning model with meta-learned confidence on four benchmark datasets.
  arXiv  Detail & Related papers  (2020-02-27T10:22:17Z)

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.