List your company

ResearchFormerly WizWand

Next Utterance Prediction

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

Persona-Chat (val)

77.01Accuracy

13

Feb 26, 2026

SayNext-PC2k (Subject-Independent Evaluation Protocol)

1.939BLEU-4

9

Feb 27, 2026

SayNext-PC2k (Subject-Dependent Evaluation Protocol)

2.307BLEU-4

9

Feb 27, 2026

IEMOCAP cross-scenario

5.44BLEU

7

Feb 27, 2026

adapted IEMOCAP Cross-Scenarios Evaluation Protocol (test)

5.439BLEU-4

7

Feb 27, 2026

SayNext-Bench (test)

29.93Top-1 Rate (GPT-4.1)

6

Feb 27, 2026

2.49BLEU

5

Feb 27, 2026

SayNext-PC19K (test)

2.493BLEU-4

5

Feb 27, 2026

SayNext-Bench 400-question (test)

17.5E1 Error Rate

4

Feb 27, 2026

SayNext-Bench 400-question 1.0 (test)

—

—E1 Error

0

Feb 26, 2026