Generalized Planning on 17 PDDL Domains
98Average CoverageSymbolic ff 30m
Evaluation Results
| Method | Links | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Symbolic ff 30mMethod=Symbolic Planner (ff), Time Limit=30m2025.08 | 98 | — | 100 | 100 | 100 | 100 | 100 | 100 | 59 | 100 | 100 | 100 | 100 | 100 | 100 | 100 | 100 | 100 | 100 | |
| DeepSeek (F5-3) BestFoundation Model=DeepSeek, Framework=F5-3, Evaluation Protocol=Coverage best run2025.08 | 93 | — | 100 | 100 | 100 | 100 | 100 | 100 | 100 | 43 | 100 | 100 | 100 | 100 | 78 | 60 | 100 | 100 | 100 | |
| DeepSeek (F3-6)Foundation Model=DeepSeek, Framework=F3-6, Evaluation Protocol=Avg coverage three runs2025.08 | 82 | — | 100 | 100 | 100 | 100 | 86 | 83 | 100 | 30 | 100 | 85 | 100 | 88 | 64 | 60 | 45 | 67 | 82 | |
| Qwen3 Thinking (F3-6) BestFoundation Model=Qwen3 Thinking, Framework=F3-6, Evaluation Protocol=Coverage best run2025.08 | 82 | — | 100 | 100 | 64 | 100 | 100 | 100 | 100 | 10 | 100 | 65 | 100 | 100 | 65 | 48 | 44 | 100 | 100 | |
| DeepSeek (F5-3)Foundation Model=DeepSeek, Framework=F5-3, Evaluation Protocol=Avg coverage three runs2025.08 | 79 | — | 67 | 100 | 100 | 100 | 100 | 100 | 100 | 24 | 100 | 67 | 80 | 88 | 72 | 35 | 63 | 67 | 82 | |
| DeepSeek (Bas)Foundation Model=DeepSeek, Framework=Bas, Evaluation Protocol=Avg coverage three runs2025.08 | 68 | — | 100 | 100 | 88 | 100 | 86 | 71 | 100 | 2 | 42 | 73 | 100 | 88 | 62 | 20 | 52 | 0 | 81 | |
| Qwen3 Thinking (F3-6-MC)Foundation Model=Qwen3 Thinking, Framework=F3-6, Ablation=-MC, Evaluation Protocol=Avg coverage three runs2025.08 | 65 | — | 100 | 100 | 64 | 100 | 33 | 89 | 100 | 2 | 85 | 34 | 98 | 67 | 34 | 15 | 44 | 67 | 74 | |
| Qwen3 Thinking (F3-6)Foundation Model=Qwen3 Thinking, Framework=F3-6, Evaluation Protocol=Avg coverage three runs2025.08 | 64 | — | 70 | 100 | 64 | 100 | 33 | 41 | 100 | 7 | 78 | 37 | 100 | 85 | 53 | 20 | 43 | 91 | 65 | |
| Qwen3 Thinking (F5-3)Foundation Model=Qwen3 Thinking, Framework=F5-3, Evaluation Protocol=Avg coverage three runs2025.08 | 60 | — | 100 | 100 | 43 | 100 | 44 | 41 | 100 | 0 | 42 | 28 | 100 | 94 | 55 | 13 | 47 | 33 | 77 | |
| Qwen3 Thinking (F3-6-SD)Foundation Model=Qwen3 Thinking, Framework=F3-6, Ablation=-SD, Evaluation Protocol=Avg coverage three runs2025.08 | 59 | — | 100 | 100 | 76 | 100 | 0 | 11 | 100 | 3 | 59 | 27 | 100 | 94 | 61 | 11 | 44 | 41 | 70 | |
| DeepSeek (Sil)Foundation Model=DeepSeek, Framework=Sil, Evaluation Protocol=Avg coverage three runs2025.08 | 56 | — | 100 | 100 | 67 | 67 | 33 | 41 | 33 | 0 | 67 | 66 | 91 | 0 | 77 | 16 | 15 | 100 | 83 | |
| Qwen3 Thinking (F3-6-CR)Foundation Model=Qwen3 Thinking, Framework=F3-6, Ablation=-CR, Evaluation Protocol=Avg coverage three runs2025.08 | 54 | — | 100 | 100 | 60 | 100 | 33 | 9 | 100 | 0 | 44 | 8 | 91 | 94 | 37 | 5 | 44 | 33 | 52 | |
| Symbolic lm 45sMethod=Symbolic Planner (lm), Time Limit=45s2025.08 | 53 | — | 0 | 31 | 15 | 100 | 100 | 56 | 15 | 0 | 79 | 89 | 22 | 38 | 99 | 88 | 76 | 15 | 82 | |
| Qwen3 Thinking (Bas)Foundation Model=Qwen3 Thinking, Framework=Bas, Evaluation Protocol=Avg coverage three runs2025.08 | 44 | — | 100 | 100 | 64 | 100 | 27 | 21 | 67 | 0 | 71 | 3 | 98 | 15 | 18 | 3 | 35 | 0 | 30 | |
| Qwen3 Thinking (Sil)Foundation Model=Qwen3 Thinking, Framework=Sil, Evaluation Protocol=Avg coverage three runs2025.08 | 39 | — | 70 | 67 | 43 | 100 | 76 | 46 | 15 | 1 | 34 | 1 | 98 | 4 | 14 | 0 | 33 | 33 | 32 |