Multi-Task Learning on Real-world 1.0 (test)
87.5Success RateLifeLong-RFT
Evaluation results
| Method | Links | |
|---|---|---|
| LifeLong-RFTTraining Strategy=RFT2026.02 | 87.5 | |
| π0Training Strategy=SFT2026.02 | 81.3 | |
| NORA-LongTraining Strategy=SFT2026.02 | 78.8 | |
| OpenVLATraining Strategy=SFT2026.02 | 71.3 |