Back to Arena

Memp

by Zhejiang University (Fang et al.)

System Card

OrganizationZhejiang University (Fang et al.)
Released2025-08
Architectureepisodic-buffer / Procedural memory of distilled step-instructions and scripts
DetailsDistills past trajectories into fine-grained step-by-step instructions and higher-level script-like abstractions. Dynamic build/retrieve/update cycle continuously revises, corrects, and deprecates procedures as agents gain experience.
Parameters—
Domainagent-memorylifelong-learning
Open SourceYes
PaperView Paper
CodeRepository
proceduralalfworldtravelplannerscriptcross-model

Capability Profile

Benchmark Scores

6 of 14 benchmarks
Long-Context Retrieval
0/5
RULER
no data
NIAH
no data
LooGLE
no data
LongBench
no data
∞Bench
no data
Multi-Turn Recall
2/2
LoCoMo
76.664p
MemoryBank
80.378p
Cross-Session Memory
1/1
LongMemEval
82.190p
Multi-Hop QA
2/3
BABILong
74.345p
MultiHop-RAG
no data
HotpotQA
72.455p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:Memp paper (arXiv:2508.06433); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)Memp paper (arXiv:2508.06433); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)Memp paper (arXiv:2508.06433); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)Memp paper (arXiv:2508.06433); evaluated on BABILong: Testing the Limits of LLMs with Long-Context Reasoning-in-a-Haystack (AIRI, 2406)Memp paper (arXiv:2508.06433); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Memp paper (arXiv:2508.06433); evaluated on MemoryBank: Enhancing LLMs with Long-Term Memory (Sun Yat-sen University, 2305)