Task-free exploration on AMZ
70.5SR (%)WA → Target → AMZ
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WA → Target → AMZTraining Setting=Sequential Curriculum, Iterations=50 + 50 + 502025.10 | 70.5 | 43.2 | |
| AMZTraining Setting=Single-Domain Specialists, Iterations=502025.10 | 69.5 | 48.3 | |
| AMZ + Target + WATraining Setting=Self-guided Exploration, Iterations=1502025.10 | 66.7 | 36.4 | |
| Target → AMZ → WATraining Setting=Sequential Curriculum, Iterations=50 + 50 + 502025.10 | 66.1 | 40.8 | |
| AMZ → WATraining Setting=Sequential Curriculum, Iterations=75 + 752025.10 | 65.3 | 42.7 | |
| AMZ → Target → WATraining Setting=Sequential Curriculum, Iterations=50 + 50 + 502025.10 | 65.2 | 43.3 | |
| SkillWeaver*Training Setting=Sequential Curriculum, Iterations=1502025.10 | 64.4 | 25.2 | |
| WATraining Setting=Single-Domain Specialists, Iterations=502025.10 | 50.2 | 3.3 | |
| TargetTraining Setting=Single-Domain Specialists, Iterations=502025.10 | 48.5 | 3.5 | |
| BaselineIterations=–2025.10 | 47.3 | — |