Multimodal Visual Pattern Question Answering on MMVP
66.1AccuracyDLWM
Evaluation results
| Method | Links | |
|---|---|---|
| DLWMSteps=5.0, Worlds=2.2, Memory=14.5, Throughput=8.92026.06 | 66.1 | |
| LaReSteps=8.0, Worlds=1.0, Memory=19.5, Throughput=5.92026.06 | 61.7 | |
| LVRSteps=8.0, Worlds=1.0, Memory=19.0, Throughput=6.12026.06 | 60.9 | |
| Soft ThinkingSteps=6.9, Worlds=1.0, Memory=16.8, Throughput=7.42026.06 | 60.5 | |
| HeimaSteps=4.5, Worlds=1.0, Memory=13.8, Throughput=9.22026.06 | 59.8 | |
| CoLaRSteps=5.8, Worlds=1.0, Memory=15.5, Throughput=8.22026.06 | 57.9 | |
| CoconutSteps=8.0, Worlds=1.0, Memory=16.0, Throughput=7.02026.06 | 57.2 | |
| CoTSteps=8.0, Worlds=1.0, Memory=18.5, Throughput=6.22026.06 | 54.3 |