Coding on CodeForces
25.51Pass RateMARFT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MARFTBackbone=LaMAS, Scale=Trio2025.04 | 25.51 | — | — | |
| MARFTBackbone=LaMAS, Scale=Quad2025.04 | 25.07 | — | — | |
| MARFTBackbone=LaMAS, Scale=Duo2025.04 | 24.85 | — | — | |
| IPPO (MAPoRL)Backbone=LaMAS, Scale=Duo2025.04 | 24.28 | — | — | |
| IPPO (MAPoRL)Backbone=LaMAS, Scale=Trio2025.04 | 24.06 | — | — | |
| VanillaBackbone=LaMAS, Scale=Duo2025.04 | 24.04 | — | — | |
| VanillaBackbone=LaMAS, Scale=Trio2025.04 | 24 | — | — | |
| IPPO (MAPoRL)Backbone=LaMAS, Scale=Quad2025.04 | 21.88 | — | — | |
| VanillaBackbone=LaMAS, Scale=Quad2025.04 | 20.5 | — | — | |
| DeepSeek-V3.2variant=Thinking, protocol=Rating2025.12 | — | 2,386 | 42,000 | |
| DeepSeek-V3.2variant=Speciale, protocol=Rating2025.12 | — | 2,701 | 77,000 | |
| Gemini-3.0variant=Pro, protocol=Rating2025.12 | — | 2,708 | 22,000 | |
| GPT-5variant=High, protocol=Rating2025.12 | — | 2,537 | 29,000 |