LINGOLY-TOO: Disentangling Memorisation from Reasoning with Linguistic Templatisation and Orthographic Obfuscation

• URL: http://arxiv.org/abs/2503.02972v3
• Date: Fri, 07 Mar 2025 09:31:42 GMT
• Title: LINGOLY-TOO: Disentangling Memorisation from Reasoning with Linguistic Templatisation and Orthographic Obfuscation
• Authors: Jude Khouja, Karolina Korgul, Simi Hellsten, Lingyi Yang, Vlad Neacsu, Harry Mayne, Ryan Kearns, Andrew Bean, Adam Mahdi,
• Abstract summary: We introduce a framework for producing linguistic reasoning problems that reduces the effect of memorisation in model performance estimates.<n>We apply this framework to develop LINGOLY-TOO, a challenging benchmark for linguistic reasoning.
• Score: 1.2576388595811496
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Assessing the reasoning capabilities of large language models (LLMs) is susceptible to overestimation due to data exposure of evaluation benchmarks. We introduce a framework for producing linguistic reasoning problems that reduces the effect of memorisation in model performance estimates and apply this framework to develop LINGOLY-TOO, a challenging benchmark for linguistic reasoning. By developing orthographic templates, we dynamically obfuscate the writing systems of real languages to generate numerousquestion variations. These variations preserve the reasoning steps required for each solution while reducing the likelihood of specific problem instances appearing in model training data. Our experiments demonstrate that frontier models, including Claud 3.7 Sonnet, o1-preview and DeepSeek R1, struggle with advanced reasoning. Our analysis also shows that LLMs exhibit noticeable variance in accuracy across permutations of the same problem, and on average perform better on questions appearing in their original orthography. Our findings highlight the opaque nature of response generation in LLMs and provide evidence that prior data exposure contributes to over estimating the reasoning capabilities of frontier models.

Related papers

• Beyond Chains of Thought: Benchmarking Latent-Space Reasoning Abilities in Large Language Models [0.0]
  Large language models (LLMs) can perform reasoning computations both internally within their latent space and externally. This study introduces a benchmark designed to quantify model-internal reasoning in different domains.
  arXiv  Detail & Related papers  (2025-04-14T18:15:27Z)
• ExpliCa: Evaluating Explicit Causal Reasoning in Large Language Models [75.05436691700572]
  We introduce ExpliCa, a new dataset for evaluating Large Language Models (LLMs) in explicit causal reasoning. We tested seven commercial and open-source LLMs on ExpliCa through prompting and perplexity-based metrics. Surprisingly, models tend to confound temporal relations with causal ones, and their performance is also strongly influenced by the linguistic order of the events.
  arXiv  Detail & Related papers  (2025-02-21T14:23:14Z)
• Investigating the Robustness of Deductive Reasoning with Large Language Models [7.494617747914778]
  Large Language Models (LLMs) have been shown to achieve impressive results for many reasoning-based Natural Language Processing (NLP) tasks.<n>It remains unclear to which extent LLMs, in both informal and autoformalisation methods, are robust on logical deduction tasks.
  arXiv  Detail & Related papers  (2025-02-04T17:16:51Z)
• JustLogic: A Comprehensive Benchmark for Evaluating Deductive Reasoning in Large Language Models [51.99046112135311]
  We introduce JustLogic, a synthetically generated deductive reasoning benchmark for rigorous evaluation of Large Language Models.<n>JustLogic is highly complex, capable of generating a diverse range of linguistic patterns, vocabulary, and argument structures.<n>Our experimental results reveal that most state-of-the-art (SOTA) LLMs perform significantly worse than the human average.
  arXiv  Detail & Related papers  (2025-01-24T15:49:10Z)
• Exploring Robustness of LLMs to Sociodemographically-Conditioned Paraphrasing [7.312170216336085]
  We take a broader approach to explore a wider range of variations across sociodemographic dimensions.<n>We extend the SocialIQA dataset to create diverse paraphrased sets conditioned on sociodemographic styles.<n>We find that demographic-specific paraphrasing significantly impacts the performance of language models.
  arXiv  Detail & Related papers  (2025-01-14T17:50:06Z)
• Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision [120.40788744292739]
  We propose a two-player paradigm that separates the roles of reasoning and critique models. We first propose AutoMathCritique, an automated and scalable framework for collecting critique data. We demonstrate that the critique models consistently improve the actor's performance on difficult queries at test-time.
  arXiv  Detail & Related papers  (2024-11-25T17:11:54Z)
• SIaM: Self-Improving Code-Assisted Mathematical Reasoning of Large Language Models [54.78329741186446]
  We propose a novel paradigm that uses a code-based critic model to guide steps including question-code data construction, quality control, and complementary evaluation. Experiments across both in-domain and out-of-domain benchmarks in English and Chinese demonstrate the effectiveness of the proposed paradigm.
  arXiv  Detail & Related papers  (2024-08-28T06:33:03Z)
• LLMs are Superior Feedback Providers: Bootstrapping Reasoning for Lie Detection with Self-Generated Feedback [33.14770105185958]
  Large Language Models (LLMs) excel at generating human-like dialogues and comprehending text. We propose a bootstrapping framework that leverages self-generated feedback to enhance LLM reasoning capabilities for lie detection. We investigate the application of the proposed framework for detecting betrayal and deception in Diplomacy games, and compare it with feedback from professional human players.
  arXiv  Detail & Related papers  (2024-08-25T18:47:55Z)
• Self-training Large Language Models through Knowledge Detection [26.831873737733737]
  Large language models (LLMs) often necessitate extensive labeled datasets and training compute to achieve impressive performance across downstream tasks. This paper explores a self-training paradigm, where the LLM autonomously curates its own labels and selectively trains on unknown data samples. Empirical evaluations demonstrate significant improvements in reducing hallucination in generation across multiple subjects.
  arXiv  Detail & Related papers  (2024-06-17T07:25:09Z)
• Prompting or Fine-tuning? Exploring Large Language Models for Causal Graph Validation [0.0]
  This study explores the capability of Large Language Models to evaluate causality in causal graphs. Our study compares two approaches: (1) prompting-based method for zero-shot and few-shot causal inference and, (2) fine-tuning language models for the causal relation prediction task.
  arXiv  Detail & Related papers  (2024-05-29T09:06:18Z)
• The Power of Question Translation Training in Multilingual Reasoning: Broadened Scope and Deepened Insights [108.40766216456413]
  We propose a question alignment framework to bridge the gap between large language models' English and non-English performance. Experiment results show it can boost multilingual performance across diverse reasoning scenarios, model families, and sizes. We analyze representation space, generated response and data scales, and reveal how question translation training strengthens language alignment within LLMs.
  arXiv  Detail & Related papers  (2024-05-02T14:49:50Z)
• DPP-Based Adversarial Prompt Searching for Lanugage Models [56.73828162194457]
  Auto-regressive Selective Replacement Ascent (ASRA) is a discrete optimization algorithm that selects prompts based on both quality and similarity with determinantal point process (DPP) Experimental results on six different pre-trained language models demonstrate the efficacy of ASRA for eliciting toxic content.
  arXiv  Detail & Related papers  (2024-03-01T05:28:06Z)
• L2CEval: Evaluating Language-to-Code Generation Capabilities of Large Language Models [102.00201523306986]
  We present L2CEval, a systematic evaluation of the language-to-code generation capabilities of large language models (LLMs) We analyze the factors that potentially affect their performance, such as model size, pretraining data, instruction tuning, and different prompting methods. In addition to assessing model performance, we measure confidence calibration for the models and conduct human evaluations of the output programs.
  arXiv  Detail & Related papers  (2023-09-29T17:57:00Z)
• CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing [139.77117915309023]
  CRITIC allows large language models to validate and amend their own outputs in a manner similar to human interaction with tools. Comprehensive evaluations involving free-form question answering, mathematical program synthesis, and toxicity reduction demonstrate that CRITIC consistently enhances the performance of LLMs.
  arXiv  Detail & Related papers  (2023-05-19T15:19:44Z)
• Data Augmentation for Spoken Language Understanding via Pretrained Language Models [113.56329266325902]
  Training of spoken language understanding (SLU) models often faces the problem of data scarcity. We put forward a data augmentation method using pretrained language models to boost the variability and accuracy of generated utterances.
  arXiv  Detail & Related papers  (2020-04-29T04:07:12Z)

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.