发表机构
Indian Institute of Information Technology (IIIT), Sri City; University of Technology Nuremberg (UTN)（印度信息技术学院斯里城分校; 纽伦堡工业大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出R-DEIM Net，一种7600万参数的双专家模型，通过交互与推理专家结合，在Quora数据集上以90.07%准确率实现高效释义检测并生成可读理由。
AI 中文摘要
近期在释义检测方面的进展揭示了一个基本权衡：大型语言模型实现了高准确率，但需要高计算量，而高效的Siamese-BERT变体提供了实用的可扩展性，但在理由生成方面的透明度有所降低。我们提出了R-DEIM Net，一个拥有7600万参数的双专家架构，旨在探索中等规模模型是否能在释义检测上实现具有竞争力的准确率，同时能够生成人类可读的理由。该架构结合了两个专门组件：一个交互专家，通过多尺度二维卷积和允许可变输入长度的注意力头来捕获词元级别的相似性模式；以及一个推理专家，使用Flan-T5-small解码器生成理由作为辅助监督。我们不重新编码生成的文本，而是提取并池化解码器的隐藏状态作为分类的补充特征。在Quora问题对数据集上，R-DEIM Net通过10折交叉验证达到了90.07%的准确率和90.16%的F1分数。这代表了与强大的基于Transformer的基线（例如，MFAE BERT：90.54%的准确率）以及最近基于大型语言模型的方法（LLaMA-70B）相比具有竞争力的性能，同时使用了显著更小的参数预算。该模型在生成预测的同时生成理由，为辅助性人类可读描述提供了潜力。
英文摘要
Recent advances in paraphrase detection reveal a fundamental trade-off: large language models achieve high accuracy but require high computation, while efficient Siamese-BERT variants offer practical scalability with reduced transparency in rationale generation. We present R-DEIM Net, a 76M-parameter dual-expert architecture exploring whether moderate-scale models can achieve competitive accuracy on paraphrase detection while enabling human-readable rationale generation. The architecture combines two specialized components: an Interaction Expert that captures token-level similarity patterns through multi-scale 2D convolutions and attention head allowing variable input length, and a Reasoning Expert that uses a Flan-T5-small decoder to generate rationales as auxiliary supervision. Rather than re-encoding generated text, we extract and pool decoder hidden states as complementary features for classification. On the Quora Question Pairs dataset, R-DEIM Net achieves 90.07\% accuracy and 90.16\% F1-score via 10-fold cross-validation. This represents competitive performance with strong transformer-based baselines (e.g., MFAE BERT: 90.54\% accuracy) and recent large language model based approaches (LLaMA-70B) while using a substantially smaller parameter budget. The model generates rationales alongside predictions, providing potential for auxiliary human-readable descriptions.
