Instruction Following on AlpacaEval trained on UF-G
17.314LC Win RateRDRO
Evaluation results
| Method | Links | |
|---|---|---|
| RDROBase Model=Llama-8B2026.04 | 17.314 | |
| DDROBase Model=Llama-8B2026.04 | 15.66 | |
| RDROBase Model=Llama-3B2026.04 | 12.634 | |
| DDROBase Model=Llama-3B2026.04 | 12.069 | |
| KTOBase Model=Llama-3B2026.04 | 9.937 | |
| KTOBase Model=Llama-8B2026.04 | 9.344 | |
| RDROBase Model=Qwen-3B2026.04 | 8.933 | |
| KTOBase Model=Qwen-3B2026.04 | 7.92 | |
| DDROBase Model=Qwen-3B2026.04 | 7.888 | |
| RDROBase Model=Qwen-1.5B2026.04 | 5.599 | |
| DDROBase Model=Qwen-1.5B2026.04 | 5.364 | |
| KTOBase Model=Qwen-1.5B2026.04 | 5.22 |