arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新，休息一下吧，周末愉快~~
arXiv 2402.15938cs.CLcs.AIcs.CRcs.LGcs.SE
收藏

泛化还是记忆：大型语言模型的数据污染与可信评估

Generalization or Memorization: Data Contamination and Trustworthy Evaluation for Large Language Models

Key Laboratory of High Confidence Software Technologies (Peking University)（高可信软件技术重点实验室（北京大学））；School of Computer Science, Peking University（北京大学计算机学院）；Beijing Institute of Control Engineering（北京控制工程研究所）；China Academy of Space Technology（中国空间技术研究院）

Yihong Dong, Xue Jiang, Huanyu Liu, Zhi Jin, Bin Gu, Mengfei Yang, Ge Li

备注：
Accepted to ACL
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文针对大型语言模型的数据污染问题，提出了基于输出分布的污染检测方法CDD和可信评估方法TED，并引入DetCon和ComiEval基准，实验证明两者在检测和缓解污染上效果显著。

摘要：

近期关于大型语言模型（LLMs）令人印象深刻的能力的声明，通常是通过在开放访问的基准上进行评估来支持的。考虑到 LLMs 训练数据的庞大规模和广泛来源，这些数据可能会显式或隐式地包含测试数据，导致 LLMs 更容易受到数据污染的影响。然而，由于训练数据的不透明性、模型的黑盒访问权限以及合成训练数据的快速增长，检测和缓解 LLMs 的数据污染面临重大挑战。在本文中，我们提出了 CDD，代表通过输出分布进行 LLMs 污染检测（Contamination Detection via output Distribution）。CDD 仅需采样文本即可检测数据污染，通过识别 LLM 输出分布的峰值来实现。为了减轻数据污染在评估中的影响，我们还提出了 TED：基于输出分布的可信评估（Trustworthy Evaluation via output Distribution），该方法基于对 LLM 输出分布的校正。为了促进这项研究，我们引入了两个基准，即 DetCon 和 ComiEval，分别用于数据污染检测和污染缓解评估任务。大量实验结果表明，在 Accuracy、F1 Score 和 AUC 指标方面，CDD 相比其他污染检测方法实现了 21.8%-30.2% 的平均相对改进，并且能够有效检测隐式污染。TED 在各种污染设置下，大幅缓解了高达 66.9% 归因于数据污染的性能提升。在实际应用中，我们揭示了 ChatGPT 在 HumanEval 基准上表现出遭受数据污染的高可能性。

英文摘要：

Recent statements about the impressive capabilities of large language models (LLMs) are usually supported by evaluating on open-access benchmarks. Considering the vast size and wide-ranging sources of LLMs' training data, it could explicitly or implicitly include test data, leading to LLMs being more susceptible to data contamination. However, due to the opacity of training data, the black-box access of models, and the rapid growth of synthetic training data, detecting and mitigating data contamination for LLMs faces significant challenges. In this paper, we propose CDD, which stands for Contamination Detection via output Distribution for LLMs. CDD necessitates only the sampled texts to detect data contamination, by identifying the peakedness of LLM's output distribution. To mitigate the impact of data contamination in evaluation, we also present TED: Trustworthy Evaluation via output Distribution, based on the correction of LLM's output distribution. To facilitate this study, we introduce two benchmarks, i.e., DetCon and ComiEval, for data contamination detection and contamination mitigation evaluation tasks. Extensive experimental results show that CDD achieves the average relative improvements of 21.8\%-30.2\% over other contamination detection approaches in terms of Accuracy, F1 Score, and AUC metrics, and can effectively detect implicit contamination. TED substantially mitigates performance improvements up to 66.9\% attributed to data contamination across various contamination setups. In real-world applications, we reveal that ChatGPT exhibits a high potential to suffer from data contamination on HumanEval benchmark.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

↑