General Instruction Following Accuracy on IFBench
84.1AccuracyMulti-DPOP
Evaluation Results
| Method | Links | |
|---|---|---|
| Multi-DPOPBackbone=Qwen3-30B2025.10 | 84.1 | |
| Base ModelBackbone=Qwen3-30B2025.10 | 83.9 | |
| A-DPOBackbone=Qwen3-30B2025.10 | 83.5 | |
| Multi-DPOPBackbone=Tulu-3.1-8B2025.10 | 82.6 | |
| Base ModelBackbone=Tulu-3.1-8B2025.10 | 82.4 | |
| A-DPOBackbone=Tulu-3.1-8B2025.10 | 82.1 | |
| GRPOBackbone=Qwen3-30B2025.10 | 81.3 | |
| Multi-DPOPBackbone=Qwen3-4B2025.10 | 80.7 | |
| Base ModelBackbone=Qwen3-4B2025.10 | 80.5 | |
| A-DPOBackbone=Qwen3-4B2025.10 | 80.2 | |
| SFTBackbone=Qwen3-30B2025.10 | 80.1 | |
| GRPOBackbone=Tulu-3.1-8B2025.10 | 79.8 | |
| SFTBackbone=Tulu-3.1-8B2025.10 | 78.6 | |
| GRPOBackbone=Qwen3-4B2025.10 | 77.9 | |
| SFTBackbone=Qwen3-4B2025.10 | 76.8 | |
| Multi-DPOPBackbone=Gemma-3-4B2025.10 | 76.5 | |
| Base ModelBackbone=Gemma-3-4B2025.10 | 76.3 | |
| A-DPOBackbone=Gemma-3-4B2025.10 | 76 | |
| GRPOBackbone=Gemma-3-4B2025.10 | 73.6 | |
| DPOBackbone=Qwen3-30B2025.10 | 73.6 | |
| SFTBackbone=Gemma-3-4B2025.10 | 72.5 | |
| DPOBackbone=Tulu-3.1-8B2025.10 | 71.2 | |
| DPOBackbone=Qwen3-4B2025.10 | 69.3 | |
| Multi-DPOPBackbone=GLM-4-9B2025.10 | 69.2 | |
| Base ModelBackbone=GLM-4-9B2025.10 | 69 | |
| A-DPOBackbone=GLM-4-9B2025.10 | 68.7 | |
| GRPOBackbone=GLM-4-9B2025.10 | 66.1 | |
| SFTBackbone=GLM-4-9B2025.10 | 65.3 | |
| DPOBackbone=Gemma-3-4B2025.10 | 64.8 | |
| DPOBackbone=GLM-4-9B2025.10 | 57.4 |