PersonaMem

Benchmarks

Task Name	Dataset Name	SOTA Result
Personalized Dialogue Response Generation	PersonaMem 1.0	Overall Score76.06	33
Memory-Augmented Dialogue	PersonaMem v1.0 (test)	Overall Score74.53	28
Long-context Memory Retrieval and Reasoning	PersonaMem 128K	F1 Score23.75	20
Long-context Memory Retrieval and Reasoning	PersonaMem 32K	F1 Score26.45	20
Query-Answering	PersonaMem 1M context length	Query-Answering Accuracy69	16
Query-Answering	PersonaMem 128K context length	Query-Answering Accuracy70	16
Query-Answering	PersonaMem 32K context length	Query-Answering Accuracy64	16
Response Selection	PersonaMem	Accuracy64.36	16
Agentic Memory Management	PersonaMem	Preference Recall69.7	11
Personality-based Memory	PERSONAMEM 32k context	Accuracy57.93	8
Personalized Memory Retrieval	PersonaMem	Precision58.9	8
Persona-based memory dialogue	PersonaMem	Normalized Score65.2	5
Personalized Generation	PersonaMem 128K memory corpus 1.0 (test)	Revisit Reasons81.41	5
Personalized Generation	PersonaMem 32K memory corpus 1.0 (test)	Revisit Reasons94.95	5
Memory Retrieval	PersonaMem 128K	NDCG@143.5	5
Personalized memory reasoning	PERSONAMEM	Recall User Shared Facts67.81	4
Personalized Generation	PersonaMem 1M memory corpus 1.0 (test)	Revisit Reasons77.87	4

Showing 17 of 17 rows