Multi-turn agent interaction on ALFWorld OOD
40.8Success Rate (SR)Zero-shot
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Zero-shotParams=30B-A3B2026.06 | 40.8 | 22.78 | |
| Guided-OPDParams=4B2026.06 | 35.08 | 23.15 | |
| Vanilla OPDParams=4B2026.06 | 32.09 | 23.34 | |
| TCODParams=4B2026.06 | 31.34 | 24.04 | |
| Guided-OPDParams=1.7B2026.06 | 29.85 | 23.6 | |
| AdaSwitchParams=4B2026.06 | 29.1 | 23.96 | |
| Zero-shotParams=4B2026.06 | 27.86 | 24.73 | |
| Vanilla OPDParams=1.7B2026.06 | 26.87 | 24.54 | |
| TCODParams=1.7B2026.06 | 24.63 | 25.01 | |
| AdaSwitchParams=1.7B2026.06 | 23.13 | 25.26 | |
| Guided-OPDParams=0.6B2026.06 | 19.4 | 25.9 | |
| AdaSwitchParams=0.6B2026.06 | 17.91 | 26.27 | |
| Vanilla OPDParams=0.6B2026.06 | 16.42 | 26.67 | |
| TCODParams=0.6B2026.06 | 14.93 | 27.14 | |
| Zero-shotParams=1.7B2026.06 | 7.1 | 27.93 | |
| Zero-shotParams=0.6B2026.06 | 0.25 | 29.75 |