Back to Arena

AutoGen Core Memory

by Microsoft

System Card

OrganizationMicrosoft
Released2024-01
Architectureagentic-workflow / BaseMemory + ListMemory abstractions
Details`autogen_core/memory` module provides `_base_memory.py` and `_list_memory.py` — a pluggable memory protocol for AgentChat agents. Now in maintenance mode, recommended successor is Microsoft Agent Framework.
Parameters—
Domainagent-memory
Open SourceYes
PaperView Paper
WebsiteVisit
CodeRepository
autogenbase-memorylist-memorymicrosoft

Capability Profile

Benchmark Scores

5 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
1/2
LoCoMo
77.469p
MemoryBank
no data
Cross-Session Memory
1/1
LongMemEval
77.863p
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
74.373p
HotpotQA
76.480p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:AutoGen Core Memory paper (arXiv:2308.08155); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)AutoGen Core Memory paper (arXiv:2308.08155); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)AutoGen Core Memory paper (arXiv:2308.08155); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)AutoGen Core Memory paper (arXiv:2308.08155); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)AutoGen Core Memory paper (arXiv:2308.08155); evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)