Conducting applied research on low-resource Indic language tokenization, Kannada sentiment and dialect parsing, cross-lingual pseudo-labeling, and compute-efficient parameter-efficient fine-tuning (PEFT/LoRA) on open weights. Our objective is establishing verified linguistic benchmarks that serve under-resourced Dravidian dialects with rigorous evaluation criteria.
Developing a standardized tokenization and morphological parsing pipeline for low-resource Kannada administrative documents with comparative subword vocabulary boundary metrics.
Quantifying label noise transfer when bootstrapping regional Kannada sentiment classifiers from English instruction-tuned models with calibrated temperature bounds.
Comprehensive empirical analysis comparing byte-pair encoding (BPE) vs sentencepiece subword segmentation on low-resource regional text with vocabulary boundary metrics. Evaluates agglutinative morphology preservation across 10,000 sentences from regional public administrative records.
Modular Python utility package providing standardized Unicode normalization, sentence splitting, and whitespace sanitization for regional text datasets. Published with complete unit tests and typing stubs.
