Binary classification on HNC (test)
80.2Attribute AccuracyBLIP
Evaluation Results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| BLIPStream Type=Contrastive Dual-Encoder, Checkpoint=HNCfull2026.05 | 80.2 | — | 77 | 64.7 | 70.3 | 72 | 55 | 73.5 | 67 | 77.5 | 58.5 | 73 | 65 | 69 | |
| ViLBERTStream Type=Dual-Stream, Checkpoint=HNCfull2026.05 | 78.2 | — | 75 | 62.8 | 66.4 | 73 | 58.5 | 76 | 59 | 73.5 | 57 | 65.5 | 59 | 66.4 | |
| UNITERStream Type=Single-Stream, Checkpoint=HNCfull2026.05 | 77.7 | — | 75 | 63.2 | 67.8 | 68.5 | 54.5 | 73 | 63.5 | 71 | 59.5 | 65 | 68 | 67.1 | |
| VisualBERTStream Type=Single-Stream, Checkpoint=HNCfull2026.05 | 77.7 | — | 79 | 65.7 | 67 | 66 | 62 | 75 | 61.5 | 74 | 58.5 | 68 | 66.5 | 67.9 | |
| LXMERTStream Type=Dual-Stream, Checkpoint=HNCfull2026.05 | 76.2 | — | 74.5 | 64.7 | 67 | 71.5 | 54.5 | 65.5 | 64 | 70 | 56.5 | 67 | 64.5 | 66.2 | |
| VL-BERTStream Type=Single-Stream, Checkpoint=HNCfull2026.05 | 75.7 | — | 72 | 63.8 | 67.5 | 65 | 62.5 | 72 | 66 | 74.5 | 57.5 | 67 | 70.5 | 67.3 | |
| UNITERStream Type=Single-Stream, Checkpoint=HNCsubset2026.05 | 67.4 | — | 61.5 | 55.2 | 61.4 | 59.1 | 56.6 | 61.3 | 56.8 | 54.4 | 51.1 | 55.6 | 58.3 | 58.2 | |
| VisualBERTStream Type=Single-Stream, Checkpoint=HNCsubset2026.05 | 65.7 | — | 60.4 | 56.1 | 61 | 62.7 | 58.2 | 57.6 | 56.5 | 56.6 | 52.7 | 57.3 | 57.9 | 58.6 | |
| VL-BERTStream Type=Single-Stream, Checkpoint=HNCsubset2026.05 | 64.4 | — | 59.2 | 55.2 | 59.4 | 63 | 60.9 | 54.2 | 58.3 | 55.4 | 52.8 | 54.9 | 58.7 | 58 | |
| LXMERTStream Type=Dual-Stream, Checkpoint=HNCsubset2026.05 | 59.7 | — | 55.6 | 54.7 | 57.8 | 55.3 | 55.6 | 49.8 | 54.8 | 54.4 | 51 | 51 | 60.2 | 55 | |
| ViLBERTStream Type=Dual-Stream, Checkpoint=HNCsubset2026.05 | 57.5 | — | 54.3 | 55 | 55.4 | 55.9 | 57.2 | 52.1 | 56.3 | 52.2 | 52.7 | 51.7 | 57.6 | 54.1 | |
| BLIPStream Type=Contrastive Dual-Encoder, Checkpoint=ITM2026.05 | 55 | — | 57 | 56 | 56.3 | 54 | 49.5 | 51.5 | 49.5 | 55 | 54.5 | 43.5 | 49.5 | 53.5 | |
| LXMERTStream Type=Dual-Stream, Checkpoint=FOIL2026.05 | 54.5 | — | 49 | 53.8 | 56.1 | 51 | 51 | 49.5 | 49.5 | 48.5 | 50 | 48.5 | 49.5 | 50.9 | |
| ViLBERTStream Type=Dual-Stream, Checkpoint=FOIL2026.05 | 52.5 | — | 54 | 54.7 | 54.4 | 49.5 | 48.5 | 50.5 | 51 | 51.5 | 52 | 51 | 49.5 | 51.6 | |
| VL-BERTStream Type=Single-Stream, Checkpoint=FOIL2026.05 | 52 | — | 51.5 | 53.5 | 54.1 | 50 | 50.5 | 49.5 | 50 | 50.5 | 52.5 | 50.5 | 50.5 | 51.3 | |
| VisualBERTStream Type=Single-Stream, Checkpoint=FOIL2026.05 | 51 | — | 49.5 | 52.2 | 52.8 | 48 | 51 | 50 | 48.5 | 51 | 52 | 50 | 50.5 | 50.5 | |
| UNITERStream Type=Single-Stream, Checkpoint=FOIL2026.05 | 50.5 | — | 51.5 | 53.8 | 52.8 | 50 | 51 | 50 | 49.5 | 50.5 | 52.5 | 49 | 49 | 50.8 | |
| LXMERTStream Type=Dual-Stream, Checkpoint=VOLTA2026.05 | 45.5 | — | 50.5 | 47.7 | 48.6 | 49 | 49 | 48.5 | 51 | 50 | 51 | 49 | 51.5 | 49 | |
| VisualBERTStream Type=Single-Stream, Checkpoint=VOLTA2026.05 | 45 | — | 47.5 | 47.3 | 47 | 49 | 48.5 | 49 | 48.5 | 50 | 48.5 | 52.5 | 51.5 | 48.3 | |
| ViLBERTStream Type=Dual-Stream, Checkpoint=VOLTA2026.05 | 44.1 | — | 47.5 | 46.2 | 47 | 51 | 50 | 49 | 49.5 | 48.5 | 52.5 | 50 | 51 | 48.3 | |
| UNITERStream Type=Single-Stream, Checkpoint=VOLTA2026.05 | 42.6 | — | 47 | 45.8 | 46.2 | 46.5 | 49.5 | 50.5 | 50 | 49.5 | 49.5 | 51 | 51 | 47.7 | |
| VL-BERTStream Type=Single-Stream, Checkpoint=VOLTA2026.05 | 41.1 | — | 47.5 | 47 | 46.2 | 51.5 | 49 | 50 | 49.5 | 49 | 51.5 | 51.5 | 50 | 48.1 |