Radiology Report Generation on Tstd Standard Benchmark (test)
42.2BLEU-1EKAGen
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| EKAGen2026.04 | 42.2 | 12 | 15.8 | 27.1 | 38.5 | 27.8 | 29.4 | 53.4 | 28.4 | |
| R2GenGPTdual-loop training strategy=true2026.04 | 41.8 | 13.1 | 16.7 | 29.1 | 33.4 | 25.9 | 27.7 | 54.2 | 28 | |
| Mamba_Xray2026.04 | 39.8 | 12.1 | 17.2 | 27 | 33.1 | 25.8 | 27.7 | 54 | 28 | |
| R2GenGPT2026.04 | 39 | 12.3 | 17.4 | 28.8 | 33.7 | 25.4 | 27.5 | 54.2 | 27.9 | |
| Qwen3VL4Bsupervised fine-tuning=true2026.04 | 24.3 | 6.3 | 14.1 | 21.2 | 33.2 | 24.7 | 25.2 | 53.8 | 27.1 | |
| Qwen3VL8Bsupervised fine-tuning=true2026.04 | 23.6 | 6.1 | 13.9 | 21.2 | 35.8 | 23.3 | 24.3 | 53.6 | 26.6 | |
| GPT5.22026.04 | 13.7 | 0.6 | 10.8 | 10.9 | 21.5 | 20.3 | 18.5 | 48.1 | 18 | |
| Qwen3VL8B2026.04 | 9.4 | 0.8 | 11.1 | 11.4 | 22.6 | 16.3 | 16.3 | 51.8 | 16.7 | |
| Qwen3VL4B2026.04 | 8.7 | 0.9 | 11.1 | 11.5 | 21.5 | 22.8 | 18.3 | 51.8 | 15.5 |