Radiology Report Generation on MianYang n=500 (external val)
0.446BLEU-1EXACT-CHAT
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| EXACT-CHATRefinement status=Refined2026.04 | 0.446 | 0.227 | 0.022 | 0.275 | 0.41 | 0.338 | 0.667 | — | |
| EXACT-CHATRefinement status=unrefined2026.04 | 0.402 | 0.214 | 0.012 | 0.266 | 0.29 | 0.326 | 0.367 | — | |
| CT-CHATBackbone=LLaMA 3.1 70B2026.04 | 0.259 | 0.184 | 0.003 | 0.259 | 0.073 | 0.119 | 0.088 | — | |
| Reg2RG2026.04 | 0.249 | 0.161 | 0.006 | 0.184 | 0.086 | 0.169 | 0.066 | — | |
| Hulu-Med2026.04 | 0.134 | 0.111 | 0.003 | 0.155 | 0.175 | 0.265 | 0.176 | — | |
| RadFM2026.04 | 0 | 0.008 | 0 | 0.019 | 0.023 | 0.143 | 0.046 | 0.001 | |
| M3D2026.04 | 0 | 0.024 | 0 | 0.051 | 0.068 | 0.162 | 0.055 | — | |
| Merlin2026.04 | 0 | 0.023 | 0 | 0.052 | 0.024 | 0.074 | 0.059 | — |