Indic-gemma-7B-finetuned-sft-Navarasa-2.0: benchmark results
Telugu-LLM-Labs' Navarasa 2.0, a LoRA fine-tune of Gemma 7B on instruction data across 15 Indian languages plus English, via unsloth. Provider: Telugu-LLM-Labs. Released 2024-03-17. Access: Open.
Aggregate Rating 1344 ± 26, rank #1372 of 1635 rated models, from 24 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BhashaBench - Ayur | 35.13 | Accuracy (%) | 63.3 |
| BhashaBench - Ayur - English | 37.12 | Accuracy (%) | 63.3 |
| BhashaBench - Finance | 34.99 | Accuracy (%) | 63.3 |
| BhashaBench - Finance - English | 37 | Accuracy (%) | 63.3 |
| BhashaBench - Legal | 41.08 | Accuracy (%) | 63.3 |
| BhashaBench - Legal - English | 44.08 | Accuracy (%) | 63.3 |
| BhashaBench - Krishi | 40.73 | Accuracy (%) | 60 |
| BhashaBench - Ayur - Hindi | 31.83 | Accuracy (%) | 56.7 |
| BhashaBench - Krishi - English | 42.31 | Accuracy (%) | 56.7 |
| BhashaBench - Krishi - Hindi | 33.44 | Accuracy (%) | 56.7 |
| BhashaBench - Legal - Hindi | 34.09 | Accuracy (%) | 56.7 |
| BhashaBench - Finance - Hindi | 30.47 | Accuracy (%) | 53.3 |
Interactive version: theaggregate.ai/model?slug=indic-gemma-7b-finetuned-sft-navarasa-2-0 · How It Works · Data refreshed daily, snapshot 2026-11-05.