約束反向翻譯提升大型語言模型對複雜指令的遵循

Constraint Back-translation Improves Complex Instruction Following of Large Language Models

October 31, 2024

作者: Yunjia Qi, Hao Peng, Xiaozhi Wang, Bin Xu, Lei Hou, Juanzi Li

cs.AI

摘要

大型語言模型（LLMs）在遵循具有複雜約束條件（格式、長度等）的指示時遇到困難。根據傳統的指示調整實踐，先前的研究對複雜指示-回應對進行後訓練，通過將複雜指示提供給先進的LLMs生成。然而，即使是先進的LLMs也無法很好地遵循複雜指示，從而限制了生成數據的質量。在這項工作中，我們發現現有數據集內在地包含隱含的複雜約束條件，並提出一種新穎的數據生成技術，即約束反向翻譯。具體而言，我們採用現有數據集中的高質量指示-回應對，僅採用先進的LLMs添加已被回應滿足的指示的複雜約束條件，這自然地降低了成本和數據噪音。在實驗中，我們採用Llama3-70B-Instruct來反向翻譯約束條件，創建了一個高質量的複雜指示-回應數據集，名為CRAB。我們展示了對CRAB進行後訓練可以提高多個主幹LLMs的複雜指示遵循能力，在廣泛的指示遵循基準測試中進行評估。我們進一步發現，約束反向翻譯也作為後訓練中一個有用的輔助訓練目標。我們的代碼、數據和模型將被釋出以促進未來研究。

English

Large language models (LLMs) struggle to follow instructions with complex constraints in format, length, etc. Following the conventional instruction-tuning practice, previous works conduct post-training on complex instruction-response pairs generated by feeding complex instructions to advanced LLMs. However, even advanced LLMs cannot follow complex instructions well, thus limiting the quality of generated data. In this work, we find that existing datasets inherently contain implicit complex constraints and propose a novel data generation technique, constraint back-translation. Specifically, we take the high-quality instruction-response pairs in existing datasets and only adopt advanced LLMs to add complex constraints already met by the responses to the instructions, which naturally reduces costs and data noise. In the experiments, we adopt Llama3-70B-Instruct to back-translate constraints and create a high-quality complex instruction-response dataset, named CRAB. We present that post-training on CRAB improves multiple backbone LLMs' complex instruction-following ability, evaluated on extensive instruction-following benchmarks. We further find that constraint back-translation also serves as a useful auxiliary training objective in post-training. Our code, data, and models will be released to facilitate future research.