用合成数据提升少样本自然语言推理性能，准确率突破94%。
First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI
- 先生成后训练：用T5模型合成句子对并清洗注入训练
- 在SNLI上达94.7%准确率，刷新少样本推理新纪录
- 适合研究数据增强与少样本学习的NLP工程师
自然语言推理（NLI）任务需判断句对间的关系，通常分为蕴含、矛盾或中立。当前最先进模型EFL在斯坦福自然语言推理（SNLI）数据集上达到93.1%准确率，但受数据集限制，进一步提升受限。为此，我们提出一种新方法，通过合成数据增强提升数据多样性与复杂性。本文提出UnitedSynT5，作为EFL的改进版本，利用基于T5的生成器合成额外前提-假设对，经严格清洗后融入训练数据。这些增强样本在EFL框架内处理，将标签直接嵌入假设以保证一致性。我们使用GTR-T5-XL模型在扩展数据集上训练，实现SNLI上94.7%、E-SNLI上94.0%、MultiNLI上92.6%的准确率，超越此前最先进模型。该研究证明了合成数据增强在提升NLI模型方面的潜力，为自然语言理解任务的进一步发展提供了新路径。
原文摘要 · Abstract (English)
Natural Language Inference (NLI) tasks require identifying the relationship between sentence pairs, typically classified as entailment, contradiction, or neutrality. While the current state-of-the-art (SOTA) model, Entailment Few-Shot Learning (EFL), achieves a 93.1% accuracy on the Stanford Natural Language Inference (SNLI) dataset, further advancements are constrained by the dataset's limitations. To address this, we propose a novel approach leveraging synthetic data augmentation to enhance dataset diversity and complexity. We present UnitedSynT5, an advanced extension of EFL that leverages a T5-based generator to synthesize additional premise-hypothesis pairs, which are rigorously cleaned and integrated into the training data. These augmented examples are processed within the EFL framework, embedding labels directly into hypotheses for consistency. We train a GTR-T5-XL model on this expanded dataset, achieving a new benchmark of 94.7% accuracy on the SNLI dataset, 94.0% accuracy on the E-SNLI dataset, and 92.6% accuracy on the MultiNLI dataset, surpassing the previous SOTA models. This research demonstrates the potential of synthetic data augmentation in improving NLI models, offering a path forward for further advancements in natural language understanding tasks.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。