Language Understanding on MMLU SAGE-generated (test)
94.12Overall Accuracy (OA)GPT-4o
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-4oFamily=GPT2026.05 | 94.12 | 67.33 | 28.46 | 64.76 | |
| Gemma-3-27BFamily=Gemma-3, Parameters=27B2026.05 | 75.21 | 49.22 | 25.99 | 39.92 | |
| Gemma-3-12BFamily=Gemma-3, Parameters=12B2026.05 | 74.79 | 49.58 | 25.21 | 40.16 | |
| Llama-3.1-70BFamily=Llama, Parameters=70B2026.05 | 70.17 | 48.8 | 21.37 | 37.21 | |
| Llama-3.1-8BFamily=Llama, Parameters=8B2026.05 | 66.81 | 47.72 | 19.09 | 34.93 | |
| Gemma-3-4BFamily=Gemma-3, Parameters=4B2026.05 | 65.55 | 47.78 | 17.77 | 34.57 | |
| Llama-3.2-3BFamily=Llama, Parameters=3B2026.05 | 62.18 | 47.12 | 15.07 | 32.53 | |
| Qwen3.5-35B-A3BFamily=Qwen3.5, Parameters=35B-A3B2026.05 | 58.4 | 47.12 | 11.28 | 30.49 | |
| Qwen3.5-9BFamily=Qwen3.5, Parameters=9B2026.05 | 56.72 | 46.88 | 9.84 | 29.95 | |
| Qwen3.5-4BFamily=Qwen3.5, Parameters=4B2026.05 | 53.36 | 46.52 | 6.84 | 27.79 | |
| Llama-3.2-1BFamily=Llama, Parameters=1B2026.05 | 52.1 | 45.5 | 6.6 | 27.91 | |
| Qwen3.5-2BFamily=Qwen3.5, Parameters=2B2026.05 | 46.22 | 45.86 | 0.36 | 24.37 |