Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Multi-turn Dialogue on MT-Bench (MAT and Speedup)

6.93MAT Score

Graft

1.33482.78744.245.6926May 19, 2026
Updated 14d ago

Evaluation Results

MethodLinks
2026.05
6.934.13
2026.05
6.813.85
2026.05
6.493.88
2026.05
6.423.61
2026.05
6.354.05
2026.05
6.233.77
2026.05
6.073.99
2026.05
4.542.93
2026.05
4.392.69
2026.05
3.792.36
2026.05
3.712.03
2026.05
3.682.18
2026.05
3.412.18
2026.05
3.341.89
2026.05
3.152.46
2026.05
3.022.03
2026.05
2.962.27
2026.05
2.612.13
2026.05
2.612.18
2026.05
2.561.81
2026.05
2.382.02
2026.05
2.371.82
2026.05
2.371.83
2026.05
2.361.86
2026.05
2.261.79
2026.05
2.151.74
2026.05
2.051.72
2026.05
1.871.76
2026.05
1.651.61
2026.05
1.551.6