Affordance Prediction on Mobile manipulation environment
91F1 ScorePaLM-E-12B
Evaluation Results
| Method | Links | |
|---|---|---|
| PaLM-E-12Btrained on=Full mixture, from scratch=false, LLM+ViT pretrain=false, LLM frozen=false2023.03 | 91 | |
| PaLM-E-12Btrained on=Full mixture, from scratch=false, LLM+ViT pretrain=true, LLM frozen=true2023.03 | 87 | |
| PaLM-E-12Btrained on=Single robot, from scratch=false, LLM+ViT pretrain=true, LLM frozen=true2023.03 | 78 | |
| QT-OPT2023.03 | 63 | |
| PaLIzero-shot=true2023.03 | 62 | |
| PaLM-E-12Btrained on=Single robot, from scratch=true, LLM+ViT pretrain=false, LLM frozen=n/a2023.03 | 46 |