Marathi
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| NER | Marathi NER (test) | Macro-F144 | 34 | |
| Sentiment Classification | Marathi (test) | Macro F187 | 34 | |
| Tokenization | Marathi (mr) | NSL Score2.35 | 10 | |
| Supervised Fine-Tuning | Marathi Multilingual SFT (test) | Best-k Accuracy34.76 | 9 | |
| Question Answering | Marathi (MR) | Accuracy (ACC)90.7 | 6 | |
| Tokenization Efficiency | Marathi mr | Bytes per Token6.49 | 6 | |
| Disfluency Correction | Marathi Real Data | Proposed Rate30 | 2 | |
| Disfluency Correction | Marathi Manually Edited | Proposed (%)14 | 2 |
Showing 8 of 8 rows