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

Persona

Benchmarks

Task NameDataset NameSOTA ResultTrend
Behavioral Alignment SteeringSix Behavioral Traits (Persona Dataset) (test)
Conscientious (Accuracy)99.6
19
LLM SteeringPersona
Average Trait Score81.75
18
Watch duration predictionPersona B
SMAPE0.86
4
Watch duration predictionPersona A
SMAPE0.617
4
Stylized Dialogue148-query style-1 persona (test)
Context Score4.257
3
Dialogue Quality EvaluationPersona High Info
BF1 (qt, at)0.62
1
Dialogue Quality EvaluationPersona Med. Info
BF1 (qt, at)61
1
Dialogue Quality EvaluationPersona Low Info
BF1 (qt, at)61
1
Dialogue Quality EvaluationPersona Sing. Inst.
BF1 (qt, at)0.58
1
Showing 9 of 9 rows