Code Generation on CodeContests (Signal, UT, Code, BoN)
38SignalBaseline
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| BaselineModel=DeepSeek-V3.2-685B, Model Category=Base Models2026.05 | 38 | 76.3 | 58.8 | 63.3 | |
| + ClusterModel=DeepSeek-V3.2-685B, Model Category=Base Models2026.05 | 36 | 76.9 | 60.2 | 66.7 | |
| CoSPlay (Ours)Model=DeepSeek-V3.2-685B, Model Category=Base Models2026.05 | 35.3 | 76.9 | 60.2 | 66 | |
| CoSPlay (Ours)Model=Seed-Coder-8B-Instruct, Model Category=Base Models2026.05 | 34 | 77.4 | 29.5 | 35.3 | |
| + ClusterModel=Seed-Coder-8B-Instruct, Model Category=Base Models2026.05 | 34 | 77.4 | 29.5 | 35.3 | |
| BaselineModel=DeepSeek-Coder-V2-Lite-Instruct-16B, Model Category=Base Models2026.05 | 34 | 51.9 | 22.8 | 34 | |
| + ClusterModel=AceCoder-Rule-7B, Model Category=RL-tuned Models2026.05 | 34 | 72.7 | 25.5 | 33.3 | |
| CoSPlay (Ours)Model=AceCoder-Rule-7B, Model Category=RL-tuned Models2026.05 | 33.3 | 72.7 | 25.5 | 34 | |
| CoSPlay (Ours)Model=Gemini-2.0-Flash, Model Category=Base Models2026.05 | 32.7 | 67 | 48.7 | 50.7 | |
| + ClusterModel=Gemini-2.0-Flash, Model Category=Base Models2026.05 | 32.7 | 67 | 48.7 | 51.3 | |
| + ClusterModel=AceCoder-RM-7B, Model Category=RL-tuned Models2026.05 | 32.6 | 73.5 | 25.8 | 34.7 | |
| CoSPlay (Ours)Model=DeepSeek-Coder-V2-Lite-Instruct-16B, Model Category=Base Models2026.05 | 32 | 82.8 | 35.8 | 40.7 | |
| + ClusterModel=DeepSeek-Coder-V2-Lite-Instruct-16B, Model Category=Base Models2026.05 | 31.3 | 82.8 | 35.8 | 41.3 | |
| CoSPlay (Ours)Model=AceCoder-RM-7B, Model Category=RL-tuned Models2026.05 | 31.3 | 73.5 | 25.8 | 34.7 | |
| BaselineModel=Gemini-2.0-Flash, Model Category=Base Models2026.05 | 30.7 | 36.2 | 44.9 | 48.7 | |
| CoSPlay (Ours)Model=Absolute-Zero-7B, Model Category=RL-tuned Models2026.05 | 30.7 | 88.9 | 21.7 | 23.3 | |
| + ClusterModel=Absolute-Zero-7B, Model Category=RL-tuned Models2026.05 | 29.3 | 88.9 | 21.7 | 26.7 | |
| BaselineModel=Seed-Coder-8B-Instruct, Model Category=Base Models2026.05 | 28.7 | 21.4 | 16.6 | 26.7 | |
| BaselineModel=Absolute-Zero-7B, Model Category=RL-tuned Models2026.05 | 28 | 28.5 | 8.9 | 23.3 | |
| BaselineModel=AceCoder-Rule-7B, Model Category=RL-tuned Models2026.05 | 27.3 | 5.3 | 23.3 | 25.3 | |
| BaselineModel=AceCoder-RM-7B, Model Category=RL-tuned Models2026.05 | 27.3 | 8.5 | 22.4 | 25.3 |