Mathematical Reasoning on Minerva MATH500
65.2ScoreTrinity Large Base
Evaluation results
| Method | Links | |
|---|---|---|
| Trinity Large Base2026.02 | 65.2 | |
| BaselineModel=Llama 3.1-8B2026.02 | 21.6 | |
| TurboQuantModel=Llama 3.1-8B2026.02 | 21.2 | |
| KIVISinkModel=Llama 3.1-8B2026.02 | 21 | |
| InnerQBaseModel=Llama 3.1-8B2026.02 | 20.8 | |
| InnerQHybridModel=Llama 3.1-8B2026.02 | 19.6 | |
| InnerQSmallModel=Llama 3.1-8B2026.02 | 19 | |
| KIVIModel=Llama 3.1-8B2026.02 | 17.2 | |
| TurboQuantModel=Mistral 7B v0.32026.02 | 12.8 | |
| InnerQBaseModel=Mistral 7B v0.32026.02 | 12.6 | |
| BaselineModel=Mistral 7B v0.32026.02 | 11.6 | |
| TurboQuantModel=Llama 3.2-3B2026.02 | 11.4 | |
| InnerQHybridModel=Mistral 7B v0.32026.02 | 11.4 | |
| InnerQBaseModel=Llama 3.2-3B2026.02 | 10.8 | |
| KIVIModel=Mistral 7B v0.32026.02 | 10.4 | |
| KIVISinkModel=Mistral 7B v0.32026.02 | 10.4 | |
| InnerQSmallModel=Mistral 7B v0.32026.02 | 9.8 | |
| BaselineModel=Llama 3.2-3B2026.02 | 9.2 | |
| KIVISinkModel=Llama 3.2-3B2026.02 | 8.6 | |
| InnerQHybridModel=Llama 3.2-3B2026.02 | 8.4 | |
| KIVIModel=Llama 3.2-3B2026.02 | 6.8 | |
| InnerQSmallModel=Llama 3.2-3B2026.02 | 6.6 | |
| KIVIModel=Llama 3.2-1B2026.02 | 5.2 | |
| TurboQuantModel=Llama 2-7B2026.02 | 5.2 | |
| BaselineModel=Llama 3.2-1B2026.02 | 4.4 | |
| KIVISinkModel=Llama 2-7B2026.02 | 4 | |
| KIVISinkModel=Llama 3.2-1B2026.02 | 3.8 | |
| TurboQuantModel=Llama 3.2-1B2026.02 | 3.6 | |
| KIVIModel=Llama 2-7B2026.02 | 3.6 | |
| InnerQBaseModel=Llama 3.2-1B2026.02 | 3.4 | |
| InnerQHybridModel=Llama 3.2-1B2026.02 | 3.4 | |
| InnerQHybridModel=Llama 2-7B2026.02 | 3.4 | |
| BaselineModel=Llama 2-7B2026.02 | 3.2 | |
| InnerQBaseModel=Llama 2-7B2026.02 | 3 | |
| InnerQSmallModel=Llama 3.2-1B2026.02 | 2.8 | |
| InnerQSmallModel=Llama 2-7B2026.02 | 0.4 |