Question Answering on MMLongBench-Doc Multi-page (MP)
24.18EMDocArena
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DocArenaPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 24.18 | 19.43 | 33.12 | |
| IKEAPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 21.03 | 18.89 | 30.02 | |
| StepSearchPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 20.96 | 17.3 | 30.5 | |
| AutoRefinePolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 20.47 | 17.5 | 29.82 | |
| Search-R1Policy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 20.1 | 17.92 | 28.98 | |
| DeepResearcherPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 19.71 | 16.25 | 27.16 | |
| R-SearchPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 18.9 | 16.02 | 28.56 | |
| ZeroSearchPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 18.73 | 15.89 | 28.14 | |
| ReSearchPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 18.41 | 15.8 | 26.22 | |
| ASearcherPolicy Model=Qwen2.5-7B-Instruct, Max Search Turns=32026.05 | 14.46 | 12.18 | 21.3 |