Offline Goal-Conditioned Reinforcement Learning on puzzle-4x6-1B
Evaluation results
| Method | Links | |
|---|
| NSBackup=n-step return backup2025.12 | | 9,100 |
|---|
| DQCMethod=Decoupled Q-chunking2025.12 | | 8,300 |
|---|
| SHARSA2025.12 | | 6,400 |
|---|
| DQC-naïveExecution=partial action chunk2025.12 | | 3,300 |
|---|
| QCStrategy=Q-chunking2025.12 | | 2,800 |
|---|
| OSBackup=1-step TD-backup2025.12 | | 1,900 |
|---|
| HIQL2025.12 | | 900 |
|---|
| IQL2025.12 | | 600 |
|---|
| HFBC2025.12 | | 400 |
|---|
| FBC2025.12 | | 100 |
|---|