Two-agent cooperation on C-WAH
42.76Total StepsPCE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PCEBackbone=GPT-4o mini2026.02 | 42.76 | 1.7 | 44,353.56 | |
| REVECABackbone=GPT-4o mini2026.02 | 46.8 | 6 | 46,312.16 | |
| PCEBackbone=GPT-OSS:20B2026.02 | 49.6 | 2.11 | 73,535.24 | |
| REVECABackbone=GPT-OSS:20B2026.02 | 53.86 | 6.49 | 74,764.24 | |
| PCEBackbone=Gemma3:4B2026.02 | 59.2 | 3.02 | 50,984.7 | |
| CoELABackbone=GPT-4o mini2026.02 | 60.4 | 9.88 | 55,467.12 | |
| CaPoBackbone=GPT-4o mini2026.02 | 60.82 | 8.72 | 41,702 | |
| REVECABackbone=Gemma3:4B2026.02 | 62.56 | 9.14 | 44,637.58 | |
| CoTSBackbone=GPT-4o mini2026.02 | 64 | 10.24 | 44,628.12 | |
| CoTSBackbone=GPT-OSS:20B2026.02 | 65.26 | 8.32 | 95,428.84 | |
| CaPoBackbone=GPT-OSS:20B2026.02 | 68.34 | 8.2 | 99,810.44 | |
| CoTSBackbone=Gemma3:4B2026.02 | 72.32 | 4.04 | 51,966.64 | |
| CoELABackbone=GPT-OSS:20B2026.02 | 72.72 | 9.22 | 77,727.2 | |
| CaPoBackbone=Gemma3:4B2026.02 | 75.88 | 7.92 | 64,015.24 | |
| CoELABackbone=Gemma3:4B2026.02 | 77.2 | 9.48 | 49,271.24 |