Evaluating Robustness of Large Language Models Against Multilingual Typographical Errors
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914484285603840 |
|---|---|
| author | Zhao, Raoyuan Liu, Yihong Altinger, Lena Schütze, Hinrich Hedderich, Michael A. |
| author_facet | Zhao, Raoyuan Liu, Yihong Altinger, Lena Schütze, Hinrich Hedderich, Michael A. |
| contents | Large language models (LLMs) are increasingly deployed in multilingual, real-world applications with user inputs -- naturally introducing \emph{typographical errors} (typos). Yet most benchmarks assume clean input, leaving the robustness of LLMs to typos across languages largely underexplored. To address this gap, we introduce MulTypo, a multilingual typo generation algorithm that simulates human-like errors based on language-specific keyboard layouts and typing behavior. We evaluate 18 open-source LLMs across three model families and five downstream tasks spanning language inference, multi-choice question answering, mathematical reasoning, and machine translation tasks. Our results show that typos consistently degrade performance, particularly in generative tasks and those requiring reasoning -- while the natural language inference task is comparatively more robust. Instruction tuning improves clean-input performance but may increase brittleness under noise. We also observe language-dependent robustness: high-resource languages are generally more robust than low-resource ones, and translation from English is more robust than translation into English. Our findings underscore the need for noise-aware training and multilingual robustness evaluation. We release a Python package for MulTypo and make the source code publicly available at https://github.com/cisnlp/multypo. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_09536 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Evaluating Robustness of Large Language Models Against Multilingual Typographical Errors Zhao, Raoyuan Liu, Yihong Altinger, Lena Schütze, Hinrich Hedderich, Michael A. Computation and Language Large language models (LLMs) are increasingly deployed in multilingual, real-world applications with user inputs -- naturally introducing \emph{typographical errors} (typos). Yet most benchmarks assume clean input, leaving the robustness of LLMs to typos across languages largely underexplored. To address this gap, we introduce MulTypo, a multilingual typo generation algorithm that simulates human-like errors based on language-specific keyboard layouts and typing behavior. We evaluate 18 open-source LLMs across three model families and five downstream tasks spanning language inference, multi-choice question answering, mathematical reasoning, and machine translation tasks. Our results show that typos consistently degrade performance, particularly in generative tasks and those requiring reasoning -- while the natural language inference task is comparatively more robust. Instruction tuning improves clean-input performance but may increase brittleness under noise. We also observe language-dependent robustness: high-resource languages are generally more robust than low-resource ones, and translation from English is more robust than translation into English. Our findings underscore the need for noise-aware training and multilingual robustness evaluation. We release a Python package for MulTypo and make the source code publicly available at https://github.com/cisnlp/multypo. |
| title | Evaluating Robustness of Large Language Models Against Multilingual Typographical Errors |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2510.09536 |