Table 9 Test cases analysis.
From: Multi-modal transformer architecture for medical image analysis and automated report generation
Test case no. | Model | Skip thought CS | RAG answer similarity | RAG answer correctness |
|---|---|---|---|---|
Test case 1 | ViGPT2 | 0.976686 | 0.927576 | 0.721483 |
DEiTGPT2 | 0.993528 | 0.897711 | 0.621039 | |
BEiTGPT2 | 0.995373 | 0.875159 | 0.522198 | |
Test case 2 | ViGPT2 | 0.975778 | 0.893751 | 0.494292 |
DEiTGPT2 | 0.985377 | 0.942548 | 0.64892 | |
BEiTGPT2 | 0.973291 | 0.912141 | 0.531985 | |
Test case 3 | ViGPT2 | 0.982428 | 0.915123 | 0.628129 |
DEiTGPT2 | 0.941918 | 0.871831 | 0.533929 | |
BEiTGPT2 | 0.981342 | 0.921913 | 0.593288 | |
Test case 4 | ViGPT2 | 0.983156 | 0.912382 | 0.583743 |
DEiTGPT2 | 0.991838 | 0.931561 | 0.673849 | |
BEiTGPT2 | 0.983137 | 0.923817 | 0.712237 |
