Video Corpus Understanding on WikiHowQA
29.7AccuracyVimRAG
Evaluation results
| Method | Links | |
|---|---|---|
| VimRAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 29.7 | |
| VideoRAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 25.7 | |
| MemAgentBackbone=Qwen3-VL-8B-Instruct2026.02 | 23.1 | |
| ReActBackbone=Qwen3-VL-8B-Instruct2026.02 | 23 | |
| VimRAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 21.8 | |
| VideoRAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 20.1 | |
| Mem1Backbone=Qwen3-VL-8B-Instruct2026.02 | 19.9 | |
| MemAgentBackbone=Qwen3-VL-4B-Instruct2026.02 | 19.2 | |
| Vanilla RAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 15.7 | |
| Mem1Backbone=Qwen3-VL-4B-Instruct2026.02 | 14.3 | |
| ReActBackbone=Qwen3-VL-4B-Instruct2026.02 | 14.1 | |
| UniversalRAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 12 | |
| Vanilla RAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 11 | |
| UniversalRAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 3.3 |