Embodied AI Task Execution on ALFWorld v1 (ID)
100Success Rate (%)AgentMark
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| AgentMarkTask=A2: Pick And Place Simple2026.01 | 100 | 11 | 1.06 | 14 | 0.29 | 0.25 | |
| BaseTask=A1: Look At Obj In Light2026.01 | 97.1 | 12.8 | — | — | — | — | |
| AgentMarkTask=A1: Look At Obj In Light2026.01 | 97.1 | 11.5 | 1.19 | 14.2 | 0.44 | -0.2 | |
| RGTask=A1: Look At Obj In Light2026.01 | 96.2 | 13.8 | — | — | — | — | |
| AgentMarkTask=A4: Pick Cool Then Place2026.01 | 91 | 17.2 | 1.17 | 26.4 | -0.14 | 0.42 | |
| AgentMarkTask=Avg.2026.01 | 89.3 | 19.4 | 1.19 | 25.5 | 0.1 | 0.2 | |
| AgentMarkTask=A5: Pick Heat Then Place2026.01 | 88.9 | 19.4 | 1.37 | 34.3 | 0.26 | -0.08 | |
| AgentMarkTask=A3: Pick Clean Then Place2026.01 | 87.3 | 18.6 | 1.23 | 30.7 | 0.05 | 0.12 | |
| AgentMarkTask=A6: Pick Two Obj And Place2026.01 | 77.6 | 31.3 | 1.28 | 37.1 | -0.13 | 0.19 |