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

PersonaMem

Benchmarks

Task NameDataset NameSOTA ResultTrend
Query-AnsweringPersonaMem 128K context length
Query-Answering Accuracy70
60
Query-AnsweringPersonaMem 32K context length
Query-Answering Accuracy90
60
Query-AnsweringPersonaMem 1M context length
Query-Answering Accuracy72
38
Personalized Dialogue Response GenerationPersonaMem 1.0
Overall Score76.06
33
Memory-Augmented DialoguePersonaMem v1.0 (test)
Overall Score74.53
28
Dialogue-Style Memory ReasoningPersonaMem
Exact Match (EM)37.19
24
Multiple-choice Query AnsweringPersonaMem (Average)
Accuracy72
22
Long-context Memory Retrieval and ReasoningPersonaMem 128K
F1 Score23.75
20
Long-context Memory Retrieval and ReasoningPersonaMem 32K
F1 Score26.45
20
Privacy ExtractionPersonaMem v2 (test)
F1 Score0.9448
18
Response SelectionPersonaMem
Accuracy64.36
16
Long-horizon implicit preference inferencePersonaMem 32k v2
Accuracy59.3
12
Long-horizon memory recallPersonaMem 32k
Accuracy77.59
12
Language Agent Memory ManagementPersonaMem
Recall Facts88.64
12
Question AnsweringPersonaMem
Accuracy77.03
12
Agentic Memory ManagementPersonaMem
Preference Recall69.7
11
Memory-intensive taskPersonaMem v2
Accuracy47.97
8
Preference evolution over long multi-session historiesPersonaMem 128K context scale
Accuracy47.24
8
Preference evolution over long multi-session historiesPersonaMem 32K context scale
Accuracy57.06
8
Personality-based MemoryPERSONAMEM 32k context
Accuracy57.93
8
Personalized Memory RetrievalPersonaMem
Precision58.9
8
Question AnsweringPersonaMem v1
R-Fact Score52.74
7
Memory-bank storage size measurementPersonaMem v2
Storage Size (GiB)0.34
6
Persona Memory Management for Language AgentsPersonaMem 128k
Accuracy63.87
5
Persona-based memory dialoguePersonaMem
Normalized Score65.2
5
Showing 25 of 32 rows