Multi-turn agent interaction on ALFWorld ScienceWorld WebShop Average
71.58Success Rate (SR)ATOD
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ATODStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-4B GRPO2026.06 | 71.58 | — | — | |
| ATODStudent Model=Qwen3-4B, Teacher Model=Qwen3-30B-A3B GRPO2026.06 | 71.06 | — | — | |
| ATODStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-4B GRPO2026.06 | 70.62 | — | — | |
| GRPOStudent Model=None, Teacher Model=Qwen3-4B GRPO2026.06 | 68.93 | — | — | |
| GRPOStudent Model=Qwen3-4B, Teacher Model=None2026.06 | 68.93 | — | — | |
| GRPOStudent Model=None, Teacher Model=Qwen3-30B-A3B GRPO2026.06 | 68.91 | — | — | |
| SODStudent Model=Qwen3-4B, Teacher Model=Qwen3-30B-A3B GRPO2026.06 | 68.28 | — | — | |
| OPDStudent Model=Qwen3-4B, Teacher Model=Qwen3-30B-A3B GRPO2026.06 | 68.13 | — | — | |
| OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-4B GRPO2026.06 | 67.73 | — | — | |
| TCODStudent Model=Qwen3-4B, Teacher Model=Qwen3-30B-A3B GRPO2026.06 | 66.91 | — | — | |
| TCODStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-4B GRPO2026.06 | 66.88 | — | — | |
| TCODStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-4B GRPO2026.06 | 66.05 | — | — | |
| SODStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-4B GRPO2026.06 | 65.74 | — | — | |
| SODStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-4B GRPO2026.06 | 65.59 | — | — | |
| OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-4B GRPO2026.06 | 64.91 | — | — | |
| SDARStudent Model=Qwen3-4B, Teacher Model=None2026.06 | 64.37 | — | — | |
| SDARStudent Model=Qwen3-1.7B, Teacher Model=None2026.06 | 41.94 | — | — | |
| GRPOStudent Model=Qwen3-1.7B, Teacher Model=None2026.06 | 40.3 | — | — | |
| GRPOStudent Model=Qwen3-0.6B, Teacher Model=None2026.06 | 33.17 | — | — | |
| Zero-shotParams=30B-A3B2026.06 | 27.72 | 41.4 | 17.96 | |
| SDARStudent Model=Qwen3-0.6B, Teacher Model=None2026.06 | 25.44 | — | — | |
| Guided-OPDParams=4B2026.06 | 25.37 | 42.64 | 18.26 | |
| TCODParams=4B2026.06 | 23.11 | 39.5 | 18.71 | |
| Vanilla OPDParams=4B2026.06 | 21.13 | 38.66 | 18.78 | |
| Guided-OPDParams=1.7B2026.06 | 19.93 | 37.94 | 18.73 | |
| AdaSwitchParams=4B2026.06 | 19.25 | 37.08 | 19.4 | |
| VanillaStudent Model=Qwen3-4B, Teacher Model=None2026.06 | 19.24 | — | — | |
| TCODParams=1.7B2026.06 | 17.5 | 34.28 | 19.63 | |
| Zero-shotParams=4B2026.06 | 17.11 | 18.4 | 19.3 | |
| Vanilla OPDParams=1.7B2026.06 | 16.11 | 32.27 | 20.2 | |
| VanillaStudent Model=Qwen3-1.7B, Teacher Model=None2026.06 | 15.59 | — | — | |
| AdaSwitchParams=1.7B2026.06 | 15.52 | 31.91 | 20.59 | |
| Guided-OPDParams=0.6B2026.06 | 15.01 | 34.04 | 19.96 | |
| AdaSwitchParams=0.6B2026.06 | 13.01 | 26.49 | 21.43 | |
| TCODParams=0.6B2026.06 | 12.62 | 31.04 | 20.94 | |
| Vanilla OPDParams=0.6B2026.06 | 11.3 | 25.16 | 21.68 | |
| VanillaStudent Model=Qwen3-0.6B, Teacher Model=None2026.06 | 6.66 | — | — | |
| Zero-shotParams=1.7B2026.06 | 6.55 | 17.6 | 22.37 | |
| Zero-shotParams=0.6B2026.06 | 1.62 | 7.78 | 23.51 |