Multi-step Reasoning on multi-step reasoning tasks
61.6Average ScoreEmbodiedAct
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EmbodiedActTotal Tokens=1,115,740, Ratio=10.6×2026.02 | 61.6 | 51.1 | |
| Baseline (Direct)Total Tokens=104,889, Ratio=1.0×2026.02 | 53.3 | 33.3 | |
| CodeActTotal Tokens=657,141, Ratio=6.3×2026.02 | 44 | 15.6 |