TrICy：基于意图感知注意力-复制机制的触发引导数据到文本生成
TrICy: Trigger-guided Data-to-text Generation with Intent aware Attention-Copy
AI总结：
针对设备端部署中大型语言模型内存占用高的问题，提出轻量级框架TrICy，利用意图感知注意力-复制机制和可选触发词引导，在多个数据集上显著提升数据到文本生成质量，超越现有LLM。
摘要：
数据到文本（D2T）生成是许多自然语言理解（NLU）应用中的关键任务，并构成任务型对话系统的基础。在能够直接处理用户设备上本地数据的对话式AI解决方案背景下，由于高内存占用，使用大型预训练语言模型（PLMs）的架构不适用于设备端部署。为此，我们提出了TrICy，一种新颖的轻量级框架，用于增强的D2T任务，该框架基于上下文中的意图生成文本序列，并可进一步由用户提供的触发词引导。我们利用注意力-复制机制来准确预测词汇表外（OOV）单词。在E2E NLG数据集（BLEU：66.43%，ROUGE-L：70.14%）、WebNLG数据集（BLEU：已见64.08%，未见52.35%）以及我们与短信应用相关的自定义数据集上的性能分析，展示了我们架构的有效性。此外，我们表明，通过利用可选的触发输入，数据到文本生成质量显著提高，并在E2E NLG上实现了新的SOTA分数，BLEU达到69.29%。进一步的分析表明，与GPT-3、ChatGPT和Llama 2等LLM相比，TrICy在BLEU和METEOR上分别至少提高了24%和3%。我们还证明，在某些场景中，即使触发词在训练中不存在，也能观察到由触发词带来的性能提升。
英文摘要：
Data-to-text (D2T) generation is a crucial task in many natural language understanding (NLU) applications and forms the foundation of task-oriented dialog systems. In the context of conversational AI solutions that can work directly with local data on the user's device, architectures utilizing large pre-trained language models (PLMs) are impractical for on-device deployment due to a high memory footprint. To this end, we propose TrICy, a novel lightweight framework for an enhanced D2T task that generates text sequences based on the intent in context and may further be guided by user-provided triggers. We leverage an attention-copy mechanism to predict out-of-vocabulary (OOV) words accurately. Performance analyses on E2E NLG dataset (BLEU: 66.43%, ROUGE-L: 70.14%), WebNLG dataset (BLEU: Seen 64.08%, Unseen 52.35%), and our Custom dataset related to text messaging applications, showcase our architecture's effectiveness. Moreover, we show that by leveraging an optional trigger input, data-to-text generation quality increases significantly and achieves the new SOTA score of 69.29% BLEU for E2E NLG. Furthermore, our analyses show that TrICy achieves at least 24% and 3% improvement in BLEU and METEOR respectively over LLMs like GPT-3, ChatGPT, and Llama 2. We also demonstrate that in some scenarios, performance improvement due to triggers is observed even when they are absent in training.
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。
