L3Cube-MahaSTS: A Marathi Sentence Similarity Dataset and Models
This provides a resource for Marathi NLP, addressing a low-resource language need, but it is incremental as it applies existing methods to a new dataset.
The authors tackled the lack of Marathi sentence similarity datasets by creating MahaSTS, a human-annotated dataset with 16,860 sentence pairs, and developed MahaSBERT-STS-v2, a fine-tuned model that outperforms alternatives like MahaBERT and IndicBERT in benchmarks.
We present MahaSTS, a human-annotated Sentence Textual Similarity (STS) dataset for Marathi, along with MahaSBERT-STS-v2, a fine-tuned Sentence-BERT model optimized for regression-based similarity scoring. The MahaSTS dataset consists of 16,860 Marathi sentence pairs labeled with continuous similarity scores in the range of 0-5. To ensure balanced supervision, the dataset is uniformly distributed across six score-based buckets spanning the full 0-5 range, thus reducing label bias and enhancing model stability. We fine-tune the MahaSBERT model on this dataset and benchmark its performance against other alternatives like MahaBERT, MuRIL, IndicBERT, and IndicSBERT. Our experiments demonstrate that MahaSTS enables effective training for sentence similarity tasks in Marathi, highlighting the impact of human-curated annotations, targeted fine-tuning, and structured supervision in low-resource settings. The dataset and model are publicly shared at https://github.com/l3cube-pune/MarathiNLP