sinhala

This paper introduces a reference-free hallucination detection framework for low-resource Sinhala-to-English neural machine translation, presenting a 45,000-sample synthetic corruption dataset and fine-tuning mDeBERTa-v3 to achieve a token-level F1 of 0.841, benchmarked across eight NMT systems.

This paper presents the first neural benchmark for Sinhala Sandhi splitting using character-level seq2seq models on the SandhiLex dataset, with a bidirectional LSTM encoder reaching 68.40% exact-match accuracy on hard lexicalized/derivational Sandhi versus 94% on regular affixational cases.

This research investigates cross-lingual transfer learning from Sinhala to Dhivehi for automatic speech recognition, achieving significant improvements in word error rate compared to Dhivehi-only baselines.

This paper introduces sinhala-ocr-lk-acts-1010, the first publicly available real-world page-level dataset for Sinhala OCR, and fine-tunes three vision language models (DeepSeek-OCR V1, DeepSeek-OCR V2, LightOnOCR-2-1B) using QLoRA. LightOnOCR-2-1B achieves a CER of 1.05%, outperforming both open-source and commercial OCR models, and maintains consistent performance across degraded documents from different time periods.