Forget What You Know about LLMs Evaluations - LLMs are Like a Chameleon

• URL: http://arxiv.org/abs/2502.07445v1
• Date: Tue, 11 Feb 2025 10:43:36 GMT
• Title: Forget What You Know about LLMs Evaluations - LLMs are Like a Chameleon
• Authors: Nurit Cohen-Inger, Yehonatan Elisha, Bracha Shapira, Lior Rokach, Seffi Cohen,
• Abstract summary: Large language models (LLMs) often appear to excel on public benchmarks, but these high scores may mask an overreliance on dataset-specific surface cues. We introduce the Chameleon Benchmark Overfit Detector (C-BOD), a meta-evaluation framework that distorts benchmark prompts. By rephrasing inputs while preserving semantic content and labels, C-BOD exposes whether a model's performance is driven by memorized patterns.
• Score: 11.753349115726952
• License:
• Abstract: Large language models (LLMs) often appear to excel on public benchmarks, but these high scores may mask an overreliance on dataset-specific surface cues rather than true language understanding. We introduce the Chameleon Benchmark Overfit Detector (C-BOD), a meta-evaluation framework that systematically distorts benchmark prompts via a parametric transformation and detects overfitting of LLMs. By rephrasing inputs while preserving their semantic content and labels, C-BOD exposes whether a model's performance is driven by memorized patterns. Evaluated on the MMLU benchmark using 26 leading LLMs, our method reveals an average performance degradation of 2.15% under modest perturbations, with 20 out of 26 models exhibiting statistically significant differences. Notably, models with higher baseline accuracy exhibit larger performance differences under perturbation, and larger LLMs tend to be more sensitive to rephrasings indicating that both cases may overrely on fixed prompt patterns. In contrast, the Llama family and models with lower baseline accuracy show insignificant degradation, suggesting reduced dependency on superficial cues. Moreover, C-BOD's dataset- and model-agnostic design allows easy integration into training pipelines to promote more robust language understanding. Our findings challenge the community to look beyond leaderboard scores and prioritize resilience and generalization in LLM evaluation.

Related papers

• Beyond the Singular: The Essential Role of Multiple Generations in Effective Benchmark Evaluation and Analysis [10.133537818749291]
  Large language models (LLMs) have demonstrated significant utilities in real-world applications. Benchmark evaluations are crucial for assessing the capabilities of LLMs.
  arXiv  Detail & Related papers  (2025-02-13T03:43:33Z)
• Preference Leakage: A Contamination Problem in LLM-as-a-judge [69.96778498636071]
  Large Language Models (LLMs) as judges and LLM-based data synthesis have emerged as two fundamental LLM-driven data annotation methods. In this work, we expose preference leakage, a contamination problem in LLM-as-a-judge caused by the relatedness between the synthetic data generators and LLM-based evaluators.
  arXiv  Detail & Related papers  (2025-02-03T17:13:03Z)
• The Vulnerability of Language Model Benchmarks: Do They Accurately Reflect True LLM Performance? [1.3810901729134184]
  Large Language Models (LLMs) excel at standardized tests while failing to demonstrate genuine language understanding and adaptability. Our systematic analysis of NLP evaluation frameworks reveals pervasive vulnerabilities across the evaluation spectrum. We lay the groundwork for new evaluation methods that resist manipulation, minimize data contamination, and assess domain-specific tasks.
  arXiv  Detail & Related papers  (2024-12-02T20:49:21Z)
• IdentifyMe: A Challenging Long-Context Mention Resolution Benchmark [22.238377215355545]
  We introduce IdentifyMe, a new benchmark for mention resolution presented in a multiple-choice question (MCQ) format. We observe a significant performance gap between the state-of-the-art sub-10B open models vs. closed ones. The highest-scoring model, GPT-4o, achieves 81.9% accuracy, highlighting the strong referential capabilities of state-of-the-art LLMs.
  arXiv  Detail & Related papers  (2024-11-12T01:05:55Z)
• The Good, The Bad, and The Greedy: Evaluation of LLMs Should Not Ignore Non-Determinism [39.392450788666814]
  Current evaluations of large language models (LLMs) often overlook non-determinism. greedy decoding generally outperforms sampling methods for most evaluated tasks. Smaller LLMs can match or surpass larger models such as GPT-4-Turbo.
  arXiv  Detail & Related papers  (2024-07-15T06:12:17Z)
• DARG: Dynamic Evaluation of Large Language Models via Adaptive Reasoning Graph [70.79413606968814]
  We introduce Dynamic Evaluation of LLMs via Adaptive Reasoning Graph Evolvement (DARG) to dynamically extend current benchmarks with controlled complexity and diversity. Specifically, we first extract the reasoning graphs of data points in current benchmarks and then perturb the reasoning graphs to generate novel testing data. Such newly generated test samples can have different levels of complexity while maintaining linguistic diversity similar to the original benchmarks.
  arXiv  Detail & Related papers  (2024-06-25T04:27:53Z)
• On the Worst Prompt Performance of Large Language Models [93.13542053835542]
  Performance of large language models (LLMs) is acutely sensitive to the phrasing of prompts. We introduce RobustAlpacaEval, a new benchmark that consists of semantically equivalent case-level queries. Experiments on RobustAlpacaEval with ChatGPT and six open-source LLMs from the Llama, Mistral, and Gemma families uncover substantial variability in model performance.
  arXiv  Detail & Related papers  (2024-06-08T13:40:38Z)
• Evaluating Generative Language Models in Information Extraction as Subjective Question Correction [49.729908337372436]
  We propose a new evaluation method, SQC-Score. Inspired by the principles in subjective question correction, we propose a new evaluation method, SQC-Score. Results on three information extraction tasks show that SQC-Score is more preferred by human annotators than the baseline metrics.
  arXiv  Detail & Related papers  (2024-04-04T15:36:53Z)
• GenCeption: Evaluate Multimodal LLMs with Unlabeled Unimodal Data [3.08543976986593]
  Multimodal Large Language Models (MLLMs) are typically assessed using expensive annotated multimodal benchmarks. This paper outlines and validates GenCeption, a novel, annotation-free evaluation method. It requires only unimodal data to measure inter-modality semantic coherence and inversely assesses MLLMs' tendency to hallucinate.
  arXiv  Detail & Related papers  (2024-02-22T21:22:04Z)
• Are Large Language Models Really Robust to Word-Level Perturbations? [68.60618778027694]
  We propose a novel rational evaluation approach that leverages pre-trained reward models as diagnostic tools. Longer conversations manifest the comprehensive grasp of language models in terms of their proficiency in understanding questions. Our results demonstrate that LLMs frequently exhibit vulnerability to word-level perturbations that are commonplace in daily language usage.
  arXiv  Detail & Related papers  (2023-09-20T09:23:46Z)
• LLMs as Factual Reasoners: Insights from Existing Benchmarks and Beyond [135.8013388183257]
  We propose a new protocol for inconsistency detection benchmark creation and implement it in a 10-domain benchmark called SummEdits. Most LLMs struggle on SummEdits, with performance close to random chance. The best-performing model, GPT-4, is still 8% below estimated human performance.
  arXiv  Detail & Related papers  (2023-05-23T21:50:06Z)

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.