GPT-4.5 and fine-tuned GPT-4o outperform the specialized IndicXlit model on most word-level Roman-to-Indic transliteration benchmarks across ten Indian languages.
Aksharantar: Open Indic-language transliteration datasets and models for the next billion users,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Beyond Specialization: Benchmarking LLMs for Transliteration of Indian Languages
GPT-4.5 and fine-tuned GPT-4o outperform the specialized IndicXlit model on most word-level Roman-to-Indic transliteration benchmarks across ten Indian languages.