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

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

4.55Speedup

BASTION

2.3662.9333.54.067May 28, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.05
4.556.57
2026.05
4.536.58
2026.05
4.326.32
2026.05
4.316.28
2026.05
2.814.46
2026.05
2.744.63
2026.05
2.692.48
2026.05
2.682.24
2026.05
2.654.21
2026.05
2.64
2026.05
2.592.37
2026.05
2.452.04