Long Video Understanding and Reasoning on VideoSIAH Eval
42AccuracyLongVT-7B-RFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LongVT-7B-RFTReasoning Prompt=true, Tool Calling=true, Frame Sampling Regime=Dense2025.11 | 42 | 47.7 | |
| LongVT-7B-RLReasoning Prompt=true, Tool Calling=true, Frame Sampling Regime=Dense2025.11 | 35.9 | 46.6 | |
| LongVT-7B-SFTReasoning Prompt=true, Tool Calling=true, Frame Sampling Regime=Dense2025.11 | 34.8 | 44.1 | |
| Qwen2.5-VL-7BReasoning Prompt=false, Tool Calling=false, Frame Sampling Regime=Dense2025.11 | 33.8 | 46 | |
| Video-R1-7BReasoning Prompt=true, Tool Calling=false, Frame Sampling Regime=Dense2025.11 | 33.1 | 42.7 | |
| LongVT-7B-RLReasoning Prompt=true, Tool Calling=true, Frame Sampling Regime=Sparse2025.11 | 31 | 43.7 | |
| Qwen2.5-VL-7BReasoning Prompt=false, Tool Calling=false, Frame Sampling Regime=Sparse2025.11 | 28.1 | 37.2 | |
| Video-R1-7BReasoning Prompt=true, Tool Calling=false, Frame Sampling Regime=Sparse2025.11 | 27.9 | 42.6 | |
| VideoRFT-7BReasoning Prompt=true, Tool Calling=false, Frame Sampling Regime=Dense2025.11 | 26.9 | 37 | |
| LongVT-7B-SFTReasoning Prompt=true, Tool Calling=true, Frame Sampling Regime=Sparse2025.11 | 26.8 | 36.2 | |
| VideoRFT-7BReasoning Prompt=true, Tool Calling=false, Frame Sampling Regime=Sparse2025.11 | 26.5 | 42.3 | |
| GPT-4oReasoning Prompt=false, Tool Calling=false2025.11 | 17.4 | 51.5 | |
| Video-Thinker-7BReasoning Prompt=true, Tool Calling=false, Frame Sampling Regime=Sparse2025.11 | 10.4 | 42.6 | |
| Video-Thinker-7BReasoning Prompt=true, Tool Calling=false, Frame Sampling Regime=Dense2025.11 | 6.6 | 42.9 |