Evidence Grounding on Biomedical QA (test)
76.9Recall (e)Qwen3-14B
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Qwen3-14BGrounding Level=claim level, Time (s)=26.0212026.01 | 76.9 | 64.4 | 66 | 97.3 | 82 | 86 | 72.7 | 0.839 | |
| eTRACERtau=0.0, Time (s)=22.1882026.01 | 74.3 | 73.5 | 70.9 | 96.9 | 94.9 | 94.6 | 76.7 | 0.943 | |
| eTRACERtau=0.5, Time (s)=14.3452026.01 | 73.6 | 73 | 70.5 | 96.5 | 94 | 93.9 | 77 | 0.946 | |
| Qwen3-8BGrounding Level=claim level, Time (s)=17.8522026.01 | 72.8 | 62.3 | 63.5 | 97.1 | 85.8 | 88.5 | 74.2 | 0.821 | |
| Qwen3-4B-InstructGrounding Level=claim level, Time (s)=14.1802026.01 | 70 | 64.3 | 63.9 | 97.5 | 76.1 | 81.7 | 72 | 0.658 | |
| Ministral-3-3B-InstructGrounding Level=claim level, Time (s)=13.8062026.01 | 69.7 | 51.4 | 53.8 | 84.1 | 80.5 | 79.4 | 69.5 | 0.682 | |
| Llama-3.1-8B-InstructGrounding Level=claim level, Time (s)=14.7152026.01 | 69.3 | 54.7 | 57 | 50.5 | 63.1 | 53.5 | 72.3 | 0.11 | |
| DeBERTa-v3-large-mnliGrounding Level=claim level, Time (s)=4.9242026.01 | 68.5 | 61.5 | 61 | 92.1 | 67.4 | 73.5 | 69.3 | 0.668 | |
| DeBERTa-v3-base-mnliGrounding Level=claim level, Time (s)=2.4842026.01 | 63.1 | 55.9 | 55.6 | 97.4 | 68.2 | 75.1 | 67.7 | 0.609 | |
| Qwen3-14BGrounding Level=sentence level, Time (s)=8.7022026.01 | 62.2 | 64.6 | 59.2 | 91.3 | 79.2 | 81.1 | — | — | |
| Qwen3-8BGrounding Level=sentence level, Time (s)=5.9622026.01 | 59.7 | 61.7 | 57 | 81.5 | 83.8 | 78.6 | — | — | |
| Ministral-3-14B-InstructGrounding Level=claim level, Time (s)=32.9552026.01 | 58.7 | 63.8 | 58.3 | 95.8 | 92.7 | 92.6 | 77.4 | 0.868 | |
| Ministral-3-8B-InstructGrounding Level=sentence level, Time (s)=7.3262026.01 | 56.9 | 56.9 | 52.9 | 91.8 | 80.1 | 81.4 | — | — | |
| Qwen3-4B-InstructGrounding Level=sentence level, Time (s)=4.7102026.01 | 55.5 | 61.8 | 55.7 | 88.8 | 81.8 | 81.5 | — | — | |
| DeBERTa-v3-large-mnliGrounding Level=sentence level, Time (s)=1.6182026.01 | 51.7 | 58 | 51.6 | 88.2 | 76.1 | 77.5 | — | — | |
| DeBERTa-v3-base-mnliGrounding Level=sentence level, Time (s)=0.8152026.01 | 49.9 | 57.1 | 50.3 | 84.4 | 73 | 73.1 | — | — | |
| Llama-3.1-8B-InstructGrounding Level=sentence level, Time (s)=4.9122026.01 | 49.8 | 43.4 | 42.2 | 27.3 | 39.9 | 29.8 | — | — | |
| Ministral-3-14B-InstructGrounding Level=sentence level, Time (s)=10.7572026.01 | 45.8 | 54.8 | 47.5 | 58.9 | 76.4 | 62.8 | — | — |