Planning on Sokoban Grid
63Validity RateL-ICL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| L-ICLNumber of training examples (m)=60, Input representation (ASCII grid)=true2026.01 | 63 | 49 | |
| L-ICLNumber of training examples (m)=60, Input representation (ASCII grid)=false2026.01 | 62 | 44 | |
| RAG-ICLContext size (characters)=20k, Input representation (ASCII grid)=true2026.01 | 25 | 10 | |
| ReActInference feedback=Oracle feedback, Input representation (ASCII grid)=true2026.01 | 21 | 13 | |
| L-ICLNumber of training examples (m)=0, Input representation (ASCII grid)=true2026.01 | 21 | 17 | |
| ReActPrompting strategy=ReAct (Prompt-Only), Input representation (ASCII grid)=true2026.01 | 19 | 12 | |
| RAG-ICLContext size (characters)=10k, Input representation (ASCII grid)=true2026.01 | 17 | 4 | |
| Zero-ShotInput representation (ASCII grid)=true2026.01 | 15 | 0 | |
| Self-RefineReasoning samples (k)=5, Input representation (ASCII grid)=true2026.01 | 13 | 8 | |
| Self-ConsistencyReasoning samples (k)=5, Input representation (ASCII grid)=true2026.01 | 10 | 5 | |
| L-ICLNumber of training examples (m)=0, Input representation (ASCII grid)=false2026.01 | 10 | 8 | |
| ToTPrompting strategy=ToT (Prompt-Only), Input representation (ASCII grid)=true2026.01 | 3 | 2 |