融合模型提升生物声学模型零样本泛化能力
Model Merging Improves Zero-Shot Generalization in Bioacoustic Foundation Models
- 将领域模型与基础语言模型线性融合，恢复指令遵循能力
- 合并后在未见物种分类上零样本性能提升超200%
- 适合需跨物种通用推理的生态监测应用
能够跨物种和任务泛化的基础模型是生物声学领域的前沿方向，NatureLM 是其中最具代表性的工作之一。尽管其领域微调在生物声学基准上表现优异，但存在指令遵循灵活性下降的问题：当同时要求给出常见名和学名时，准确率显著降低。本文提出一种简单的模型融合策略，将 NatureLM 与其基础语言模型进行插值，仅以极小代价恢复了指令遵循能力。最终，融合模型展现出显著更强的零样本泛化性能，在闭集零样本物种分类任务中实现超过200%的相对提升，并达到新的最先进水平。
原文摘要 · Abstract (English)
Foundation models capable of generalizing across species and tasks represent a promising new frontier in bioacoustics, with NatureLM being one of the most prominent examples. While its domain-specific fine-tuning yields strong performance on bioacoustic benchmarks, we observe that it also introduces trade-offs in instruction-following flexibility. For instance, NatureLM achieves high accuracy when prompted for either the common or scientific name individually, but its accuracy drops significantly when both are requested in a single prompt. We address this by applying a simple model merging strategy that interpolates NatureLM with its base language model, recovering instruction-following capabilities with minimal loss of domain expertise. Finally, we show that the merged model exhibits markedly stronger zero-shot generalization, achieving over a 200% relative improvement and setting a new state-of-the-art in closed-set zero-shot classification of unseen species.
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。