Gemma 2 27B (IT): benchmark results
Instruction-tuned Gemma 2 27B checkpoint. Provider: Google. Released 2024-06-27. Access: Open.
Aggregate Rating 1539 ± 6, rank #645 of 1656 rated models, from 1065 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EPPCMinerBen (0-Shot) - Code Classification | 64.62 | Micro-averaged F1 (%) of EPPC code classification (sentence- | 100 |
| EuroEval Norwegian NLU - Norec | 67.35 | Matthews correlation coefficient (%) | 100 |
| Open Persian LLM - ARC Easy | 95.98 | Accuracy (%) | 100 |
| PARSE (Persian Zero-Shot) - Factoid Multihop List | 0.47 | List match (0-1; answers containing a gold item over the lar | 100 |
| PARSE (Persian Zero-Shot) - Factoid Reasoning List | 0.28 | List match (0-1; answers containing a gold item over the lar | 100 |
| PEC-Home (ICL) - Multi-User Level 1 | 93.63 | Execution accuracy (%; 1,068 smart-home commands at ellipsis | 100 |
| Pinocchio Italian - Cultura | 76.3 | Accuracy (%) | 100 |
| Pinocchio Italian - Diritto | 66.23 | Accuracy (%) | 100 |
| Pinocchio Italian - Generale | 71.82 | Accuracy (%) | 100 |
| Pinocchio Italian - Lingua Straniera | 86.83 | Accuracy (%) | 100 |
| Pinocchio Italian - Logica | 56.99 | Accuracy (%) | 100 |
| Pinocchio Italian Leaderboard | 70.97 | Average Accuracy (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gemma-2-27b-it · How It Works · Data refreshed daily, snapshot 2026-10-20.