Image Captioning on COCO zero-shot DeCap
68.5GDOriginal CLIP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Original CLIPzero-shot=true, learned projection layer=false2026.03 | 68.5 | 21 | 2.9 | 25 | |
| Mean-Centeringzero-shot=true, learned projection layer=false2026.03 | 68.5 | 22.7 | 4 | 25 | |
| CLIP-Refinezero-shot=true, learned projection layer=false2026.03 | 64.7 | 23.6 | 4.5 | 26.5 | |
| M2-Mixzero-shot=true, learned projection layer=false2026.03 | 63.7 | 19.7 | 3.2 | 23.9 | |
| CS-Alignerzero-shot=true, learned projection layer=false2026.03 | 61.7 | 17.4 | 3.6 | 24.9 | |
| TPC-CMAalpha_target=0.01, zero-shot=true, learned projection layer=false2026.03 | 61 | 22.7 | 5 | 27.3 | |
| TPC-CMAalpha_target=0.05, zero-shot=true, learned projection layer=false2026.03 | 55.4 | 25.4 | 4.4 | 26.9 | |
| AlignCLIPzero-shot=true, learned projection layer=false2026.03 | 55 | 11.2 | 1.5 | 24.4 | |
| TPC-CMAalpha_target=0.3, zero-shot=true, learned projection layer=false2026.03 | 47.4 | 29.3 | 4.8 | 27.6 | |
| TPC-CMAalpha_target=0.5, zero-shot=true, learned projection layer=false2026.03 | 44.1 | 33 | 5.7 | 28.8 | |
| TPC-CMAalpha_target=0.9, zero-shot=true, learned projection layer=false2026.03 | 39.2 | 35.6 | 6.5 | 29.6 |