arXiv cs.CL· Carlos Carvalho, Francisco Teixeira, Thomas Rolland, Alberto Abad·· 11 小时前AI 评分31
模型合并在多域 ASR 适配中的潜力与局限：11 种算法在 10 个欧洲葡萄牙语域上的基准测试
Exploring the potential and limitations of Model Merging for Multi-Domain Adaptation in ASR
AI 导读
研究针对多域 ASR 基准测试了 11 种模型合并算法，覆盖 10 个欧洲葡萄牙语域，并评估域内准确率、分布偏移下的鲁棒性以及英语和多语言表现。作者提出基于 TSV-M 的新算法 BoostedTSV-M，通过奇异值增强缓解秩坍缩并提升数值稳定性。该方法在欧洲葡萄牙语上优于全量微调，同时以单一模型保留分布外泛化能力。
正文 · 原文
Abstract:Model merging is a scalable alternative to multi-task training that combines the capabilities of multiple specialised models into a single model. This is particularly attractive for large speech foundation models, which are typically adapted through domain-specific fine-tuning, resulting in multiple customised checkpoints, for which repeating full fine-tuning when new data becomes available is computationally prohibitive. In this work, we study model merging for multi-domain ASR and benchmark 11 merging algorithms for 10 European Portuguese domains, evaluating in-domain accuracy, robustness under distribution shift, as well as English and multilingual performance. We further propose BoostedTSV-M, a new merging algorithm based on TSV-M that mitigates rank collapse via singular-value boosting and improves numerical stability. Overall, our approach outperforms full fine-tuning on European Portuguese while preserving out-of-distribution generalisation in a single model.
| Comments: | Accepted at INTERSPEECH 2026 |
| Subjects: | Computation and Language (cs.CL); Audio and Speech Processing (eess.AS) |
| Cite as: | arXiv:2603.05354 [cs.CL] |
| (or arXiv:2603.05354v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2603.05354 arXiv-issued DOI via DataCite |
Submission history
From: Carlos Manuel Ferreira Carvalho [view email]
[v1]
Thu, 5 Mar 2026 16:34:24 UTC (65 KB)
[v2]
Fri, 30 Oct 2026 16:42:54 UTC (43 KB)
来源：arXiv cs.CL · arxiv.org