Hierarchical Planning on BabyAI Combined Skills 2
2,528Token CostTheoryCoder-2
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TheoryCoder-2Ablation Configuration=Full2026.01 | 2,528 | — | |
| TheoryCoder-2Ablation Configuration=TC - C2026.01 | 45,175 | — | |
| LLM + πAgent Architecture=LLM + π2026.01 | 49,973 | — | |
| TheoryCoder-2Ablation Configuration=TC - P2026.01 | 53,376 | — | |
| LLM + PAgent Architecture=LLM + P2026.01 | 59,003 | — | |
| WorldCoderAgent Architecture=WorldCoder2026.01 | 120,200 | — |