AGIEval English
General QA · 2023-04-13
AGIEval English is the English-language portion of AGIEval, a benchmark of standardized admission and qualification exams. It measures exam-style reasoning and knowledge in a multiple-choice and short-answer setting.
Top models (higher is better)
| Model | Score |
|---|---|
| Nemotron 3 Ultra 550B A55B Base | 78.7 |
| Nemotron 3 Super 120B A12B Base | 77.9 |
| Kimi K2 Base | 72.5 |
| Llama 3.1 405B | 71.6 |
| Nemotron Nano 12B V2 Base | 68.0 |
| Gemma 3 27B PT | 66.2 |
| Mistral Small 3 24B Base 2501 | 65.8 |
| Nemotron Nano 9B V2 Base | 65.3 |
| Ministral 3 14B Base 2512 | 64.8 |
| Llama 3.1 70B | 64.6 |
| Llama 3 70B | 63.0 |
| Ministral 3 8B Base 2512 | 59.1 |
| Gemma 2 27B | 55.1 |
| Ministral 3 3B Base 2512 | 51.1 |
| LLaMA 65B | 47.6 |
| Llama 2 34B | 43.4 |
| Gemma 3 4B PT | 42.1 |
| LLaMA 33B | 41.7 |
| Mistral 7B v0.1 Base | 41.2 |
| RecurrentGemma 9B | 39.3 |
| Llama 3.2 3B | 39.2 |
| Falcon 40B | 37.0 |
| LLaMA-13B | 33.9 |
| MPT-30B | 33.8 |
| Gemma 2B | 24.2 |
| LLaMA 7B | 23.9 |
| RecurrentGemma 2B | 23.8 |
| MPT-7B | 23.5 |
| Llama 3.2 1B | 23.3 |
| Falcon 7B | 21.2 |