General Text Question Answering on SQuAD
76.4AccuracyVimRAG
Evaluation results
| Method | Links | |
|---|---|---|
| VimRAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 76.4 | |
| MemAgentBackbone=Qwen3-VL-8B-Instruct2026.02 | 74.8 | |
| VimRAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 73.7 | |
| MemAgentBackbone=Qwen3-VL-4B-Instruct2026.02 | 70.4 | |
| Mem1Backbone=Qwen3-VL-8B-Instruct2026.02 | 68.4 | |
| Mem1Backbone=Qwen3-VL-4B-Instruct2026.02 | 67.5 | |
| ReActBackbone=Qwen3-VL-8B-Instruct2026.02 | 65.5 | |
| UniversalRAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 65.3 | |
| UniversalRAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 64.7 | |
| Vanilla RAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 64.2 | |
| ReActBackbone=Qwen3-VL-4B-Instruct2026.02 | 63.8 | |
| VideoRAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 63.2 | |
| Vanilla RAGBackbone=Qwen3-VL-4B-Instruct2026.02 | 63 | |
| VideoRAGBackbone=Qwen3-VL-8B-Instruct2026.02 | 62.2 |