Consistency Evaluation on Diagnostic (Avg. YouCook2, COIN, CrossTask) (test)
76.92State AccuracyCAST
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CASTBackbone=VideoPrism-B, Setting=+ CAST2026.03 | 76.92 | 77.66 | |
| CASTBackbone=InternVideo2-1B, Setting=+ CAST2026.03 | 75.43 | 77.77 | |
| VideoPrism-BBackbone=VideoPrism-B, Setting=Zero-Shot2026.03 | 68.38 | 33.68 | |
| CASTBackbone=GME-Qwen2-VL-2B, Setting=+ CAST2026.03 | 67.2 | 72.28 | |
| CASTBackbone=Qwen3-VL-Embedding-2B, Setting=+ CAST2026.03 | 66.18 | 69.11 | |
| InternVideo2-1BBackbone=InternVideo2-1B, Setting=Zero-Shot2026.03 | 65.7 | 30.85 | |
| Qwen3-VL-Embedding-2BBackbone=Qwen3-VL-Embedding-2B, Setting=Zero-Shot2026.03 | 58.44 | 29.79 | |
| GME-Qwen2-VL-2BBackbone=GME-Qwen2-VL-2B, Setting=Zero-Shot2026.03 | 56.31 | 29.73 |