Question Answering on Smart Expressway Knowledge QA
96.8MCQ AccuracyLlama-3.3-70B
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Llama-3.3-70BParameters=70B2026.03 | 96.8 | 95.9 | 96.3 | 95.8 | 95.4 | 96.1 | 95.7 | 88.5 | 84 | 86.9 | |
| ExpressMind-14BParameters=14B2026.03 | 96.7 | 96.2 | 96.4 | 97.5 | 97 | 97.6 | 96.4 | 89.2 | 85.4 | 87.8 | |
| Qwen-32BParameters=32B2026.03 | 96.5 | 95.2 | 95.8 | 95.5 | 94.8 | 95.7 | 83.1 | 78.5 | 76.4 | 79.5 | |
| DeepSeek-R1-Distill-Llama-70BParameters=70B, Backbone=Llama2026.03 | 95.9 | 96.1 | 95.5 | 94.8 | 96.2 | 95.7 | 94.1 | 88.4 | 84.2 | 86.1 | |
| DeepSeek-R1-Distill-Qwen-32BParameters=32B, Backbone=Qwen2026.03 | 95.4 | 96.1 | 94.8 | 96.7 | 95.5 | 96.2 | 95.6 | 88.9 | 83.1 | 86 | |
| GLM-4-32BParameters=32B2026.03 | 90.6 | 88.7 | 89.6 | 90.2 | 89.1 | 90.4 | 89.3 | 79.8 | 70.9 | 75.2 | |
| Baichuan-32BParameters=32B2026.03 | 89.7 | 87.5 | 88.6 | 88.2 | 87 | 88.1 | 87.7 | 77.5 | 68.4 | 73.8 |