Radiology Report Generation on Taux Auxiliary Benchmark (test)
0.39BLEU-1EKAGen
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| EKAGen2026.04 | 0.39 | 0.103 | 0.154 | 0.259 | 43.7 | 27.3 | 29.9 | 0.503 | 0.266 | |
| R2GenGPTdual-loop training strategy=true2026.04 | 0.388 | 0.104 | 0.159 | 0.269 | 38.8 | 29.3 | 30.7 | 0.523 | 0.267 | |
| R2GenGPT2026.04 | 0.383 | 0.107 | 0.157 | 0.271 | 32.9 | 28 | 28.6 | 0.519 | 0.263 | |
| Mamba_Xray2026.04 | 0.361 | 0.091 | 0.16 | 0.247 | 38.7 | 27.2 | 27.8 | 0.514 | 0.254 | |
| Qwen3VL4Bsupervised fine-tuning=true2026.04 | 0.237 | 0.054 | 0.131 | 0.198 | 33.8 | 28.4 | 28.6 | 0.516 | 0.255 | |
| Qwen3VL8Bsupervised fine-tuning=true2026.04 | 0.225 | 0.054 | 0.131 | 0.199 | 33.5 | 23.8 | 25.5 | 0.505 | 0.243 | |
| GPT5.22026.04 | 0.137 | 0.005 | 0.099 | 0.107 | 21.8 | 21.8 | 18.6 | 0.468 | 0.153 | |
| Qwen3VL8B2026.04 | 0.081 | 0.006 | 0.106 | 0.101 | 19.8 | 20.1 | 18.2 | 0.493 | 0.14 | |
| Qwen3VL4B2026.04 | 0.077 | 0.007 | 0.104 | 0.104 | 22.3 | 27.5 | 20.8 | 0.498 | 0.137 |