面向推理的后训练与推理时LoRA重缩放：针对音频相关问答任务
Reasoning-Oriented Post-Training and Inference-Time LoRA Rescaling for Audio-Dependent Question Answering
浏览论文内容
中文总结 AI 辅助
该研究针对音频相关问答任务，提出面向推理的LoRA后训练与推理时重缩放方法，在Qwen和MOSS-Audio模型上验证了有效性，提交系统在挑战赛中获总体第三、轻量级系统第二。
中文摘要 AI 辅助
音频相关问答（ADQA）要求大型音频语言模型（LALMs）回答正确答案依赖给定音频内容的问题。成功的ADQA需要准确的音频感知、问题相关证据的识别以及跨模态推理。利用DCASE 2026任务5的官方ADQA数据集，我们针对Qwen2.5-Omni和MOSS-Audio-8B-Thaming两种模型，研究了面向推理的低秩适配（LoRA）后训练与推理时LoRA重缩放方法。我们引入了结构化思维链（CoT）框架，将推理过程分解为问题分析、问题类型、音频证据和推理四个环节。随后，我们分析了任务特定的LoRA适配对两种主干模型的影响，并进一步探索了训练后LoRA适配器的推理时重缩放。在开发集上的实验显示出明显的主干依赖行为：在我们的监督微调配置下，后训练提升了基于Qwen的系统，但大幅降低了MOSS-Audio的性能；适度的LoRA重缩放进一步将最佳Qwen系统的Top-1准确率从58.93%提升至61.05%，并部分恢复了微调后MOSS-Audio模型的性能，而最佳MOSS-Audio系统达到了67.70%的Top-1准确率。我们提交的系统在挑战赛中总体排名第三，且在参数规模小于10B的轻量级系统中排名第二。
英文摘要
Audio-Dependent Question Answering (ADQA) requires Large Audio-Language Models (LALMs) to answer questions whose correct answers depend on the given audio content. Successful ADQA requires accurate audio perception, identification of question-relevant evidence, and cross-modal reasoning. Using the official ADQA dataset of DCASE 2026 Task 5, we investigate reasoning-oriented post-training with Low-Rank Adaptation (LoRA) and inference-time LoRA rescaling for both Qwen2.5-Omni and MOSS-Audio-8B-Thinking. We introduce a structured Chain-of-Thought (CoT) framework that decomposes the reasoning process into question analysis, question type, audio evidence, and reasoning. We then analyze how task-specific LoRA adaptation affects the two backbones and further explore inference-time rescaling of trained LoRA adapters. Experiments on the development set reveal markedly backbone-dependent behavior: post-training improves the Qwen-based systems but substantially degrades MOSS-Audio under our supervised fine-tuning configuration. Moderate LoRA rescaling further improves the best Qwen system's top-1 accuracy from 58.93% to 61.05% and partially restores the performance of the fine-tuned MOSS-Audio models, while the best MOSS-Audio system achieves 67.70% top-1 accuracy. Our submitted systems ranked third overall and second among lightweight systems under 10B parameters in the challenge.
