List your company

Multi-turn Dialogue Evaluation on MT-Bench OOD

8.64R1 ScorePrompt
3.92885.15196.3757.5981Jun 22, 2026

Evaluation results

MethodLinks
PromptModel=Deepseek-V4-Flash, Category=Vanilla2026.068.648.06
PromptModel=Qwen3-32B, Category=Vanilla2026.068.588.06
PromptModel=Qwen3-8B, Category=Vanilla2026.068.047.04
SFTModel=Qwen3-8B, Category=Supervised2026.067.967.43
PromptModel=Qwen3-4B, Category=Vanilla2026.067.826.82
EvoRubricsModel=Qwen3-4B, Category=Ours2026.067.747.11
RuscaRLModel=Qwen3-4B, Category=RL w/ Rubric2026.067.646.71
OnlineRubricsModel=Qwen3-8B, Category=RL w/ Rubric2026.067.66.71
SFTModel=Qwen3-4B, Category=Supervised2026.067.56.44
EvoRubricsModel=Qwen3-8B, Category=Ours2026.067.487.05
GoldenRubricsModel=Qwen3-8B, Category=RL w/ Rubric2026.067.256.63
OnlineRubricsModel=Qwen3-4B, Category=RL w/ Rubric2026.066.44.33
GoldenRubricsModel=Qwen3-4B, Category=RL w/ Rubric2026.066.044.01
RuscaRLModel=Qwen3-8B, Category=RL w/ Rubric2026.064.113.4