Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Dialogue Evaluation research benchmarks – Page 2 · SOTA2 Research

Dialogue Evaluation

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Wizard of Wikipedia (WW)

R2C2 BlenderBot

12.4Perplexity

4

Feb 26, 2026

10.2Perplexity

4

Feb 26, 2026

Fixed Response strategy

97.7USR RET

4

Feb 26, 2026

0.998USR RET

4

Feb 26, 2026

ACUTE-Eval Human-Chat (test)

75Engagingness

4

Feb 26, 2026

0.173Spearman's rho

3

Apr 13, 2026

Soccer dialogue

0.04SentBLEU

3

Feb 26, 2026

In-car dialogue

0.17SentBLEU

3

Feb 26, 2026

DSTC9 Interactive Conversation (test)

0.91Pearson

3

Feb 26, 2026

Elec2Deb20 Normal students

83Divergence

2

Jun 2, 2026

Elec2Deb20 Normal Students 1.0 (test)

86Divergence

2

Jun 2, 2026

Amazon Topical-Chat

MILE-RefHumEval

0.806Naturalness (Pearson r)

2

Feb 26, 2026

Page 2 of 2

Popular datasets

DailyDialog Empathetic Dialogues TopicalChat Elec2Deb MT-Bench

Follow for updates

Join SOTA2 Contact