Unveiling the Achilles' Heel of NLG Evaluators: A Unified Adversarial Framework Driven by Large Language Models

• URL: http://arxiv.org/abs/2405.14646v2
• Date: Wed, 02 Oct 2024 01:47:34 GMT
• Title: Unveiling the Achilles' Heel of NLG Evaluators: A Unified Adversarial Framework Driven by Large Language Models
• Authors: Yiming Chen, Chen Zhang, Danqing Luo, Luis Fernando D'Haro, Robby T. Tan, Haizhou Li,
• Abstract summary: We introduce AdvEval, a novel black-box adversarial framework against NLG evaluators. AdvEval is specially tailored to generate data that yield strong disagreements between human and victim evaluators. We conduct experiments on 12 victim evaluators and 11 NLG datasets, spanning tasks including dialogue, summarization, and question evaluation.
• Score: 52.368110271614285
• License:
• Abstract: The automatic evaluation of natural language generation (NLG) systems presents a long-lasting challenge. Recent studies have highlighted various neural metrics that align well with human evaluations. Yet, the robustness of these evaluators against adversarial perturbations remains largely under-explored due to the unique challenges in obtaining adversarial data for different NLG evaluation tasks. To address the problem, we introduce AdvEval, a novel black-box adversarial framework against NLG evaluators. AdvEval is specially tailored to generate data that yield strong disagreements between human and victim evaluators. Specifically, inspired by the recent success of large language models (LLMs) in text generation and evaluation, we adopt strong LLMs as both the data generator and gold evaluator. Adversarial data are automatically optimized with feedback from the gold and victim evaluator. We conduct experiments on 12 victim evaluators and 11 NLG datasets, spanning tasks including dialogue, summarization, and question evaluation. The results show that AdvEval can lead to significant performance degradation of various victim metrics, thereby validating its efficacy.

Related papers

• Dialogue You Can Trust: Human and AI Perspectives on Generated Conversations [1.178527785547223]
  This study explores the comparative performance of human and AI assessments across a range of dialogue scenarios. In Experiment 1, we evaluated multi-party conversations on Coherence, Innovation, Concreteness, and Goal Contribution. In Experiment 2, we focused on dyadic dialogues and assessing Commonsense Contradiction, Incorrect Fact, and Redundancy.
  arXiv  Detail & Related papers  (2024-09-03T11:40:38Z)
• Themis: A Reference-free NLG Evaluation Language Model with Flexibility and Interpretability [39.12792986841385]
  In this paper, we construct a large-scale NLG evaluation corpus NLG-Eval with annotations from both human and GPT-4. We also propose an LLM dedicated to NLG evaluation, which has been trained with our designed multi-perspective consistency verification and rating-oriented preference alignment methods. Themis exhibits superior evaluation performance on various NLG tasks, simultaneously generalizing well to unseen tasks and surpassing other evaluation models, including GPT-4.
  arXiv  Detail & Related papers  (2024-06-26T14:04:29Z)
• DEBATE: Devil's Advocate-Based Assessment and Text Evaluation [6.2689399557794525]
  We propose DEBATE, an NLG evaluation framework based on multi-agent scoring system. Within the framework, one agent is instructed to criticize other agents' arguments. We show that the extensiveness of debates among agents and the persona of an agent can influence the performance of evaluators.
  arXiv  Detail & Related papers  (2024-05-16T09:41:12Z)
• Collaborative Evaluation: Exploring the Synergy of Large Language Models and Humans for Open-ended Generation Evaluation [71.76872586182981]
  Large language models (LLMs) have emerged as a scalable and cost-effective alternative to human evaluations. We propose a Collaborative Evaluation pipeline CoEval, involving the design of a checklist of task-specific criteria and the detailed evaluation of texts.
  arXiv  Detail & Related papers  (2023-10-30T17:04:35Z)
• ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate [57.71597869337909]
  We build a multi-agent referee team called ChatEval to autonomously discuss and evaluate the quality of generated responses from different models. Our analysis shows that ChatEval transcends mere textual scoring, offering a human-mimicking evaluation process for reliable assessments.
  arXiv  Detail & Related papers  (2023-08-14T15:13:04Z)
• Towards a Unified Multi-Dimensional Evaluator for Text Generation [101.47008809623202]
  We propose a unified multi-dimensional evaluator UniEval for Natural Language Generation (NLG) We re-frame NLG evaluation as a Boolean Question Answering (QA) task, and by guiding the model with different questions, we can use one evaluator to evaluate from multiple dimensions. Experiments on three typical NLG tasks show that UniEval correlates substantially better with human judgments than existing metrics.
  arXiv  Detail & Related papers  (2022-10-13T17:17:03Z)
• Towards Automatic Evaluation of Dialog Systems: A Model-Free Off-Policy Evaluation Approach [84.02388020258141]
  We propose a new framework named ENIGMA for estimating human evaluation scores based on off-policy evaluation in reinforcement learning. ENIGMA only requires a handful of pre-collected experience data, and therefore does not involve human interaction with the target policy during the evaluation. Our experiments show that ENIGMA significantly outperforms existing methods in terms of correlation with human evaluation scores.
  arXiv  Detail & Related papers  (2021-02-20T03:29:20Z)
• Robustness Gym: Unifying the NLP Evaluation Landscape [91.80175115162218]
  Deep neural networks are often brittle when deployed in real-world systems. Recent research has focused on testing the robustness of such models. We propose a solution in the form of Robustness Gym, a simple and evaluation toolkit.
  arXiv  Detail & Related papers  (2021-01-13T02:37:54Z)

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.