code: <a href=\"https://github.com/LycheeMem/LycheeMem\" rel=\"nofollow\">https://github.com/LycheeMem/LycheeMem</a></p>\n","updatedAt":"2026-08-14T02:33:57.474Z","author":{"_id":"6a621d506d33c5f918d57f0b","avatarUrl":"/avatars/500c84a3d8e0041d973065aa42894d39.svg","fullname":"liu","name":"dsf26","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":1,"identifiedLanguage":{"language":"en","probability":0.18271753191947937},"editors":["dsf26"],"editorAvatarUrls":["/avatars/500c84a3d8e0041d973065aa42894d39.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.12990","authors":[{"_id":"6a7e7e6542823931a1f1764c","user":{"_id":"62e38a3ef3b208e2aecf2c84","avatarUrl":"/avatars/eee9d7d53f3f9b7a18a21d289abd3c64.svg","isPro":false,"fullname":"Dongfang Li","user":"crazyofapple","type":"user","name":"crazyofapple"},"name":"Dongfang Li","status":"claimed_verified","statusLastChangedAt":"2026-08-14T08:45:04.675Z","hidden":false},{"_id":"6a7e7e6542823931a1f1764d","name":"Zixuan Liu","hidden":false},{"_id":"6a7e7e6542823931a1f1764e","name":"Junmai Wang","hidden":false},{"_id":"6a7e7e6542823931a1f1764f","name":"Jiahe Huang","hidden":false},{"_id":"6a7e7e6542823931a1f17650","name":"Fuhao Li","hidden":false},{"_id":"6a7e7e6542823931a1f17651","name":"Bonian Jia","hidden":false},{"_id":"6a7e7e6542823931a1f17652","name":"Baotian Hu","hidden":false},{"_id":"6a7e7e6542823931a1f17653","name":"Min Zhang","hidden":false}],"publishedAt":"2026-08-13T00:00:00.000Z","submittedOnDailyAt":"2026-08-14T00:00:00.000Z","title":"LycheeMemory V2: Efficient Long-Term Memory for LLM Agents via Semantic Segment-Level Consolidation","submittedOnDailyBy":{"_id":"6a621d506d33c5f918d57f0b","avatarUrl":"/avatars/500c84a3d8e0041d973065aa42894d39.svg","isPro":false,"fullname":"liu","user":"dsf26","type":"user","name":"dsf26"},"summary":"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.","upvotes":2,"discussionId":"6a7e7e6542823931a1f17654","githubRepo":"https://github.com/LycheeMem/LycheeMem","githubRepoAddedBy":"user","ai_summary":"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.","ai_keywords":["long-horizon LLM agents","long-term memory","semantic segment-level consolidation","typed memory records","query-planned evidence retrieval","LoCoMo","LongMemEval-S"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":1159},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"62e38a3ef3b208e2aecf2c84","avatarUrl":"/avatars/eee9d7d53f3f9b7a18a21d289abd3c64.svg","isPro":false,"fullname":"Dongfang Li","user":"crazyofapple","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.12990.md","query":{}}">
LycheeMemory V2: Efficient Long-Term Memory for LLM Agents via Semantic Segment-Level Consolidation
Published on Aug 13
· Submitted by liu on 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 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.
Community
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2608.12990 in a model README.md to link it from this page.
Cite arxiv.org/abs/2608.12990 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2608.12990 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.