如何为日耳曼语言调优多语言编码器模型:PEFT、全量微调和语言适配器研究
How to Tune a Multilingual Encoder Model for Germanic Languages: A Study of PEFT, Full Fine-Tuning, and Language Adapters
AI总结:
本文比较 mDeBERTa 在德语、瑞典语和冰岛语任务上的全量微调与 LoRA、Pfeiffer 适配器等 PEFT 方法,并评估文本训练适配器模块,揭示语言资源和任务类型对最优策略的影响。
摘要:
本文研究多语言编码器模型 mDeBERTa 在三种日耳曼语言——德语、瑞典语和冰岛语——任务中的最佳使用方式;这三种语言在 mDeBERTa 预训练数据中的覆盖程度和可能的数据质量各不相同。我们将全量微调与参数高效微调(PEFT)方法 LoRA 和 Pfeiffer 瓶颈适配器进行比较,发现 PEFT 对资源较丰富的德语更有效。然而,瑞典语和冰岛语的结果较不稳定。我们还观察到任务间差异:尽管 PEFT 往往更适合问答任务,但全量微调更适合命名实体识别。受以往结合任务适配器与语言适配器的模块化方法研究启发,我们评估了加入在非结构化文本上训练的 PEFT 模块的影响,发现该方法并无益处。
英文摘要:
This paper investigates the optimal use of the multilingual encoder model mDeBERTa for tasks in three Germanic languages -- German, Swedish, and Icelandic -- representing varying levels of presence and likely data quality in mDeBERTas pre-training data. We compare full fine-tuning with the parameter-efficient fine-tuning (PEFT) methods LoRA and Pfeiffer bottleneck adapters, finding that PEFT is more effective for the higher-resource language, German. However, results for Swedish and Icelandic are less consistent. We also observe differences between tasks: While PEFT tends to work better for question answering, full fine-tuning is preferable for named entity recognition. Inspired by previous research on modular approaches that combine task and language adapters, we evaluate the impact of adding PEFT modules trained on unstructured text, finding that this approach is not beneficial.
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。