LycheeMemory V2: Efficient Long-Term Memory for LLM Agents via Semantic Segment-Level Consolidation

Hugging Face Daily Papers Papers
long-term-memory llm-agents memory-consolidation semantic-segments efficient-retrieval

Summary

LycheeMemory V2 introduces semantic segment-level consolidation for LLM agent memory, reducing construction costs while improving or maintaining accuracy on LoCoMo and LongMemEval benchmarks.

Long-horizon LLM agents must preserve information from past interactions to support future tasks. Existing memory systems typically rely on eager consolidation, invoking LLMs after each interaction to extract, summarize, or update memories. This design makes memory construction increasingly costly as conversations grow. Coarse summarization can reduce construction cost but risks discarding fine-grained contextual evidence, whereas larger retrieval contexts or multi-hop LLM reasoning shift the overhead to query time. We present LycheeMemory V2, an efficient long-term memory framework that replaces turn-level consolidation with semantic segment-level consolidation. Instead of consolidating every interaction, LycheeMemory batches multiple exchanges into segments and encodes each finalized segment into context-independent typed memory records. Segment-level batching lowers LLM encoding frequency, while semantic boundary detection helps preserve coherent event-level and temporal evidence compared with fixed-window batching. The resulting records are organized with lightweight structured indexes for query-planned evidence retrieval. Experiments using GPT-4.1-Mini show that LycheeMemory achieves state-of-the-art performance, reaching 89.22% on LoCoMo and 92.20% on LongMemEval-S. Compared with A-Mem, it reduces construction tokens by 86.0% on LoCoMo and 75.9% on LongMemEval-S without increasing query-time token usage. More broadly, our results suggest that the accuracy--cost trade-off of long-term agent memory depends not only on what information is retained, but also on the granularity at which it is consolidated.
Original Article
View Cached Full Text

Cached at: 08/14/26, 03:26 AM

Paper page - LycheeMemory V2: Efficient Long-Term Memory for LLM Agents via Semantic Segment-Level Consolidation

Source: https://huggingface.co/papers/2608.12990 Published on Aug 13

·

Submitted byhttps://huggingface.co/dsf26

liuon Aug 14

Abstract

LycheeMemory V2 improves long-term agent memory by batching interactions into semantic segments for efficient consolidation and structured retrieval, reducing construction costs while maintaining high accuracy.

Long-horizon LLM agentsmust preserve information from past interactions to support future tasks. Existing memory systems typically rely on eager consolidation, invoking LLMs after each interaction to extract, summarize, or update memories. This design makes memory construction increasingly costly as conversations grow. Coarse summarization can reduce construction cost but risks discarding fine-grained contextual evidence, whereas larger retrieval contexts or multi-hop LLM reasoning shift the overhead to query time. We present LycheeMemory V2, an efficientlong-term memoryframework that replaces turn-level consolidation withsemantic segment-level consolidation. Instead of consolidating every interaction, LycheeMemory batches multiple exchanges into segments and encodes each finalized segment into context-independenttyped memory records. Segment-level batching lowers LLM encoding frequency, while semantic boundary detection helps preserve coherent event-level and temporal evidence compared with fixed-window batching. The resulting records are organized with lightweight structured indexes forquery-planned evidence retrieval. Experiments using GPT-4.1-Mini show that LycheeMemory achieves state-of-the-art performance, reaching 89.22% onLoCoMoand 92.20% onLongMemEval-S. Compared with A-Mem, it reduces construction tokens by 86.0% onLoCoMoand 75.9% onLongMemEval-Swithout increasing query-time token usage. More broadly, our results suggest that the accuracy--cost trade-off of long-term agent memory depends not only on what information is retained, but also on the granularity at which it is consolidated.

View arXiv pageView PDFGitHub1.16kAdd to collection

Get this paper in your agent:

hf papers read 2608\.12990

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.12990 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.12990 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.12990 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

SimpleMem: Efficient Lifelong Memory for LLM Agents

Papers with Code Trending

Introduces SimpleMem, an efficient memory framework for LLM agents that uses semantic lossless compression to improve accuracy and reduce token consumption, achieving 26.4% F1 improvement and up to 30x reduction in inference-time token usage.

HeLa-Mem: Hebbian Learning and Associative Memory for LLM Agents

arXiv cs.CL

HeLa-Mem is a bio-inspired memory architecture for LLM agents that models memory as a dynamic graph using Hebbian learning dynamics, featuring episodic and semantic memory stores to improve long-term coherence. Experiments on LoCoMo show superior performance across question categories while using fewer context tokens.

MemRefine: LLM-Guided Compression for Long-Term Agent Memory

arXiv cs.CL

MemRefine is an LLM-guided framework for compressing long-term agent memory under fixed storage budgets, using similarity for candidate pairing and an LLM judge for factual deletion/merge decisions, outperforming rule-based baselines on benchmarks.