Accident Reason Answering on MM-AU (test)
93.97AccuracyLegal Multi-Agent Reasoning framework
Evaluation results
| Method | Links | |
|---|---|---|
| Legal Multi-Agent Reasoning framework2026.03 | 93.97 | |
| SeViLAYear=2023, Vision Representation=object-centric, Text Representation=Flan T5, Parameters (M)=1082024.03 | 89.02 | |
| SeViLA2026.03 | 89.02 | |
| CoVGTYear=2023, Vision Representation=object-centric, Text Representation=RoBERTa, Parameters (M)=1592024.03 | 79.97 | |
| CoVGT2026.03 | 79.97 | |
| FrozenGQAYear=2023, Vision Representation=frame-centric, Text Representation=DeBERTa, Parameters (M)=302024.03 | 77.01 | |
| FrozenGQA2026.03 | 77.01 | |
| ClipBERTYear=2021, Vision Representation=frame-centric, Text Representation=BERT, Parameters (M)=1372024.03 | 72.71 | |
| VGTYear=2022, Vision Representation=object-centric, Text Representation=BERT, Parameters (M)=1432024.03 | 68.66 | |
| VGT2026.03 | 68.66 | |
| HCRNYear=2020, Vision Representation=frame-centric, Text Representation=GloVe, Parameters (M)=422024.03 | 64.65 |