MT-Bench Evaluation on UltraFeedback
8.1MT-Bench ScoreSPA
Evaluation Results
| Method | Links | |
|---|---|---|
| SPABackbone=Llama3 (8B), Size=10%2025.10 | 8.1 | |
| DPOBackbone=Llama3 (8B), Size=1%2025.10 | 8 | |
| DPOBackbone=Llama3 (8B), Size=10%2025.10 | 8 | |
| ORPOBackbone=Llama3 (8B), Size=1%2025.10 | 8 | |
| SimPOBackbone=Llama3 (8B), Size=1%2025.10 | 8 | |
| KTOBackbone=Llama3 (8B), Size=10%2025.10 | 8 | |
| SSPOBackbone=Llama3 (8B), Size=1%2025.10 | 8 | |
| ORPOBackbone=Llama3 (8B), Size=10%2025.10 | 7.9 | |
| SimPOBackbone=Llama3 (8B), Size=10%2025.10 | 7.9 | |
| KTOBackbone=Llama3 (8B), Size=1%2025.10 | 7.9 | |
| SPABackbone=Llama3 (8B), Size=1%2025.10 | 7.9 | |
| SSPOBackbone=Llama3 (8B), Size=10%2025.10 | 7.9 | |
| SPABackbone=Mistral (7B), Size=10%2025.10 | 7.8 | |
| SPABackbone=Mistral (7B), Size=1%2025.10 | 7.7 | |
| SSPOBackbone=Mistral (7B), Size=1%2025.10 | 7.7 | |
| SSPOBackbone=Mistral (7B), Size=10%2025.10 | 7.7 | |
| DPOBackbone=Mistral (7B), Size=1%2025.10 | 7.6 | |
| DPOBackbone=Mistral (7B), Size=10%2025.10 | 7.6 | |
| SimPOBackbone=Mistral (7B), Size=1%2025.10 | 7.6 | |
| KTOBackbone=Mistral (7B), Size=1%2025.10 | 7.6 | |
| KTOBackbone=Mistral (7B), Size=10%2025.10 | 7.6 | |
| ORPOBackbone=Mistral (7B), Size=1%2025.10 | 7.5 | |
| ORPOBackbone=Mistral (7B), Size=10%2025.10 | 7.5 | |
| SimPOBackbone=Mistral (7B), Size=10%2025.10 | 7.5 | |
| DPOBackbone=Phi-2 (2.7B), Size=1%2025.10 | 6.3 | |
| DPOBackbone=Phi-2 (2.7B), Size=10%2025.10 | 6.3 | |
| ORPOBackbone=Phi-2 (2.7B), Size=1%2025.10 | 6.3 | |
| ORPOBackbone=Phi-2 (2.7B), Size=10%2025.10 | 6.3 | |
| SimPOBackbone=Phi-2 (2.7B), Size=1%2025.10 | 6.3 | |
| SimPOBackbone=Phi-2 (2.7B), Size=10%2025.10 | 6.3 | |
| SSRMBackbone=Phi-2 (2.7B), Size=10%2025.10 | 6.3 | |
| KTOBackbone=Phi-2 (2.7B), Size=1%2025.10 | 6.3 | |
| KTOBackbone=Phi-2 (2.7B), Size=10%2025.10 | 6.3 | |
| SPABackbone=Phi-2 (2.7B), Size=1%2025.10 | 6.3 | |
| SPABackbone=Phi-2 (2.7B), Size=10%2025.10 | 6.3 | |
| SSPOBackbone=Phi-2 (2.7B), Size=1%2025.10 | 6.3 | |
| SSPOBackbone=Phi-2 (2.7B), Size=10%2025.10 | 6.3 | |
| SSRMBackbone=Phi-2 (2.7B), Size=1%2025.10 | 6.2 | |
| SSRMBackbone=Llama3 (8B), Size=1%2025.10 | 6.2 | |
| SSRMBackbone=Llama3 (8B), Size=10%2025.10 | 6.2 | |
| SSRMBackbone=Mistral (7B), Size=10%2025.10 | 5.5 | |
| SSRMBackbone=Mistral (7B), Size=1%2025.10 | 5.4 |