Multilingual State Space Models for Structured Question Answering in Indic Languages
Arpita Vats, Rahul Raja, Mrinal Mathur, Vinija Jain, Aman Chadha · 2025
Read Multilingual State Space Models for Structured Question Answering in Indic Languages by Arpita Vats, Rahul Raja, Mrinal Mathur, Vinija Jain, Aman Chadha (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-01673.txt.
Description
Arpita Vats, Rahul Raja, Mrinal Mathur, Vinija Jain, and Aman Chadha apply state space models to question answering in Indic languages, in a 2025 paper. India's languages pose hard problems for language tech: rich word forms, tangled syntax, and scarce training data. State space models track long and short links in sequences well, which suits these traits. The team tested several such designs on question answering data sets across Indic languages.
Their results show the models catch subtle language cues and align questions with context more soundly. Answer quality improved over earlier approaches in their comparisons. The authors call this the first use of these models for Indic question answering, setting a benchmark for later work. They also propose tweaks to fit low-resource and multilingual settings better.
Catalog record
| CREATOR | Arpita Vats, Rahul Raja, Mrinal Mathur, Vinija Jain, Aman Chadha |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-01673