Robot Manipulation on LIBERO Goal (Succ, Cos, LLM)
97Success Rateπ0
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| π02026.04 | 97 | — | — | |
| π0+ERTMethod Variant=ERT2026.04 | 29.2 | 12.1 | 6.6 | |
| π0+GRPOMethod Variant=GRPO2026.04 | 6.6 | 6.1 | 3.7 | |
| π0+GRPO w/o KLMethod Variant=GRPO without KL2026.04 | 3.4 | 6.5 | 4.8 | |
| π0+DAERTMethod Variant=DAERT2026.04 | 3 | 11.3 | 7.8 |