Overall Performance Evaluation on Consolidated Evaluation Benchmark
49.49Overall Average ScoreAlphaToken
Evaluation Results
| Method | Links | |
|---|---|---|
| AlphaTokenBackbone=Qwen-3.5-9B, Warm-start Reference=UltraChat-200k2026.06 | 49.49 | |
| ConfPOBackbone=Qwen-3.5-9B, Warm-start Reference=UltraChat-200k2026.06 | 47.66 | |
| SePOBackbone=Qwen-3.5-9B, Warm-start Reference=UltraChat-200k2026.06 | 47.12 | |
| TI-DPOBackbone=Qwen-3.5-9B, Warm-start Reference=UltraChat-200k2026.06 | 46.97 | |
| DPOBackbone=Qwen-3.5-9B, Warm-start Reference=UltraChat-200k2026.06 | 46.46 | |
| BaseBackbone=Qwen-3.5-9B, Warm-start Reference=UltraChat-200k2026.06 | 38.32 | |
| AlphaTokenBackbone=Gemma-3-4B, Warm-start Reference=UltraChat-200k2026.06 | 37.03 | |
| ConfPOBackbone=Gemma-3-4B, Warm-start Reference=UltraChat-200k2026.06 | 35.49 | |
| SePOBackbone=Gemma-3-4B, Warm-start Reference=UltraChat-200k2026.06 | 34.66 | |
| TI-DPOBackbone=Gemma-3-4B, Warm-start Reference=UltraChat-200k2026.06 | 34.64 | |
| DPOBackbone=Gemma-3-4B, Warm-start Reference=UltraChat-200k2026.06 | 32.25 | |
| AlphaTokenBackbone=Llama-3.2-3B, Warm-start Reference=UltraChat-200k2026.06 | 29.5 | |
| ConfPOBackbone=Llama-3.2-3B, Warm-start Reference=UltraChat-200k2026.06 | 27.77 | |
| TI-DPOBackbone=Llama-3.2-3B, Warm-start Reference=UltraChat-200k2026.06 | 27.42 | |
| SePOBackbone=Llama-3.2-3B, Warm-start Reference=UltraChat-200k2026.06 | 27.26 | |
| BaseBackbone=Gemma-3-4B, Warm-start Reference=UltraChat-200k2026.06 | 26.87 | |
| DPOBackbone=Llama-3.2-3B, Warm-start Reference=UltraChat-200k2026.06 | 25.66 | |
| BaseBackbone=Llama-3.2-3B, Warm-start Reference=UltraChat-200k2026.06 | 22.51 |