Code reasoning on LCB v5
33.72AccuracyMCPO-DAPO
Evaluation Results
| Method | Links | |
|---|---|---|
| MCPO-DAPOBackbone=Qwen3-8B2026.05 | 33.72 | |
| MCPO-GRPOBackbone=Qwen3-8B2026.05 | 32.54 | |
| MT-GRPOBackbone=Qwen3-8B2026.05 | 30.33 | |
| MGSBackbone=Qwen3-8B2026.05 | 29.08 | |
| MCPO-DAPOBackbone=Qwen3-4B2026.05 | 28.93 | |
| MCPO-GRPOBackbone=Qwen3-4B2026.05 | 28.46 | |
| CLIPOBackbone=Qwen3-8B2026.05 | 28.46 | |
| DAPOBackbone=Qwen3-8B2026.05 | 28.04 | |
| GRPOBackbone=Qwen3-8B2026.05 | 27.73 | |
| MT-GRPOBackbone=Qwen3-4B2026.05 | 27.41 | |
| MGSBackbone=Qwen3-4B2026.05 | 27.04 | |
| DAPOBackbone=Qwen3-4B2026.05 | 25.88 | |
| GRPOBackbone=Qwen3-4B2026.05 | 25.79 | |
| CLIPOBackbone=Qwen3-4B2026.05 | 24.03 | |
| Base ModelBackbone=Qwen3-8B2026.05 | 17.29 | |
| Base ModelBackbone=Qwen3-4B2026.05 | 14.43 |