Consistent Video Retrieval on YouCook2 (test)
75.59AccuracyCAST
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CASTBackbone=VideoPrism-B, Setting=+ CAST2026.03 | 75.59 | 1.38 | |
| CASTBackbone=InternVideo2-1B, Setting=+ CAST2026.03 | 71.68 | 1.48 | |
| CASTBackbone=Qwen3-VL-Embedding-2B, Setting=+ CAST2026.03 | 56.64 | 1.85 | |
| CASTBackbone=GME-Qwen2-VL-2B, Setting=+ CAST2026.03 | 54.39 | 1.95 | |
| VideoPrism-BBackbone=VideoPrism-B, Setting=Zero-Shot2026.03 | 47.45 | 2.13 | |
| InternVideo2-1BBackbone=InternVideo2-1B, Setting=Zero-Shot2026.03 | 36.75 | 2.59 | |
| Qwen3-VL-Embedding-2BBackbone=Qwen3-VL-Embedding-2B, Setting=Zero-Shot2026.03 | 33.45 | 2.89 | |
| GME-Qwen2-VL-2BBackbone=GME-Qwen2-VL-2B, Setting=Zero-Shot2026.03 | 29.62 | 3.1 |