Pith. sign in

REVIEW 3 major objections 5 minor 61 references

Towards a Formal Definition of Agent Memory: Basis, Span, Optimality, and the Sequential Memory Problem

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proposes a formal definition of agent memory: a memory is a basis of stored events, its knowledge is the span generated by a generation operator, and the optimal memory is the capacity-limited choice that maximizes expected…

desk verdict A clean, honest formalization of agent memory whose coverage-theoretic core is correct under its stated assumptions; the single-item-support assumption keeps it from being a practical yardstick yet, but it deserves serious engagement. read the letter →

arxiv 2608.11654 v1 pith:YFQBZVMY submitted 2026-08-12 cs.LG

Hongyao Tang This is my paper · ORCID
classification cs.LG
keywords agentmemoryformaldefinitionutility-capacityfrontiercoverageutilitygenerationoperatorsequentialmemorizationMDPprecisionbasisandspan
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper is an attempt to give agent memory a formal definition and an optimality theory. It proposes that a memory is a basis: a subset of the events extracted from a material, and that the knowledge an agent can actually use is the span of that basis under a generation operator that turns event sets into knowledge. Under a single-item support assumption, a query is answerable exactly when one generated knowledge item covers it, so the best memory of a given size is the one that covers the most query mass. That optimality condition traces a utility–capacity frontier, giving memory systems a common yardstick and making 'how good is a memory' a measurable quantity. The same framework is then extended to noisy extraction, where coverage and precision diverge, and to sequential writing, formalized as a Markov decision process with memory as state and writing as action.

What carries the argument

The machinery is the two-layer separation of events from knowledge together with the generation operator $\Phi:2^E\to 2^N$, which maps any set of events to the knowledge it entails. Memory is a basis $M_D\subseteq E_D$; its span is $\Phi(M_D)$; and, under Assumption 2.1's single-item support, a query is answerable exactly when it lies in the union of the answerable sets $Q(n)$ of the spanned items. This reduces memory construction to a coverage problem whose optimal value defines the utility–capacity frontier $U_D^*(S)$, the key yardstick; under decomposability it is weighted maximum coverage, with greedy attaining $1-1/e$. A second mechanism, the good span $\Phi(M_D\cap E_D)$, separates coverage from precision under noise, and the sequential memorization MDP with state $M_{m,t}$, write action, and query-time reward unifies writing as a learning problem.

What would settle it

A concrete test: find or construct a query that an agent answers correctly only by combining two separate stored items at answer time, with no single spanned knowledge item covering the query. If the end-to-end answer is correct while the coverage utility is 0, Assumption 2.1 fails for that setting and the coverage-optimality theory does not apply; likewise, observing a stored claim that lowers rather than raises end-to-end answer quality would violate the monotonicity assumption on which the frontier picture depends.

Watch

Extended reading notes

Core claim

In the paper's own terms, the central claim is that agent memory is well defined as a subset $M_D\subseteq E_D$ of a material's events; its span $R_{M_D}=\Phi(M_D)$ is the knowledge those events generate, and the optimal memory of capacity $S$ is the constrained maximizer of expected coverage utility $u(q,M_D)=\mathbf{1}[q\in\bigcup_{n\in\Phi(M_D)} Q(n)]$ with respect to the query distribution $p_D$. The frontier $U_D^*(S)$ attained by that maximizer is non-decreasing in $S$ and saturates at the full-context baseline, delimiting a compression zone in which memory matters. When the generation operator decomposes over events, the problem reduces to weighted maximum coverage, which is NP-hard with a greedy $1-1/e$ approximation; in general it is monotone set-function maximization where cross-event synergy defeats such guarantees. The paper further claims that under noisy extraction, coverage and precision part ways, with the water-inflation degree measuring how much apparent memory quality is bought with false claims, and that continual memorization is a sequential MDP in which delayed query-time settlement is the reward.

Load-bearing premise

The load-bearing premise is single-item support: every query must be answerable by one generated knowledge item, with all composition pushed into the generation operator; if real queries require combining stored items at answer time, the coverage utility is not the right objective and the optimality theory collapses.

Editorial extensions

If this is right

  • Memory quality becomes quantifiable: every write policy defines a point in the size–utility plane below the frontier, and its vertical gap to the frontier is the memory-efficiency loss.
  • The frontier saturates at the full-context baseline, so memory matters only in the compression zone $S<|E_D|$, where the vertical gap to that baseline is the cost of compression.
  • With a decomposable generation operator, optimal memory is weighted maximum coverage, so greedy selection attains the $1-1/e$ approximation; with interacting events, cross-event synergy defeats such guarantees.
  • Under noisy extraction, optimizing coverage can prefer false claims; precision utility and the water-inflation degree separate true quality from inflated quality, and the noisy optimal-memory problem should target precision.
  • Writing becomes a delayed-reward learning problem formalized as a sequential MDP, with credit assignment and trust estimation as the core difficulties of learning a write policy.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: Because the framework's yardstick only needs a material, a query distribution, and an operator $\Phi$, one could approximate the frontier $U_D^*(S)$ on real corpora using an LLM as $\Phi$, producing a common ranking of memory modules without new benchmarks.
  • Beyond the paper: The single-item support assumption implies a measurable diagnostic: if a substantial share of correct agent responses requires composing several retrieved knowledge items at answer time, the coverage objective understates memory value and a joint-answer model is needed.
  • Beyond the paper: The water-inflation degree could be used as a predictor for hallucination-driven memory failure: agents whose stored claims are mostly false should show a large gap between coverage and precision, and injecting a conflicting false claim into a locally monotone $\Phi$ should visibly reduce end-to-end correctness.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper develops a formal framework for defining and optimizing agent memory. It models a material D by its event set ED, a generation operator Φ that maps event sets to knowledge items, and a memory MD as a subset of ED whose 'span' is Φ(MD). Under the single-item support assumption, a query is answerable exactly when some spanned knowledge item covers it, so expected query coverage becomes the objective. The optimal memory is the capacity-constrained maximizer of this coverage utility, which defines a utility–capacity frontier; under a decomposable Φ the problem reduces to weighted maximum coverage with the classical (1−1/e) greedy guarantee, while non-decomposable interactions can break submodularity. The paper then separates coverage from precision under noisy extraction, defines the water-inflation degree, organizes memorization settings into a taxonomy, and unifies them in a sequential MDP in which memory is the state, writing is the action, and query-time utility is the reward. The framework is instantiated on a hand-constructed Odyssey example and used to position existing memory systems. The appendices contain the stated proofs.

Significance. If the framework is accepted, it supplies a formal object for memory, a well-posed optimality criterion, and a common utility–capacity yardstick, which would be a useful conceptual contribution to the agent-memory literature. The formal propositions are simple but correctly argued, and the reduction to weighted maximum coverage in Proposition 3.3 is standard and correctly applied. The coverage-versus-precision separation and the water-inflation degree are genuinely clarifying ideas. A particular strength is that the derivation is definitional and self-contained: no empirical fitting is used, no prediction is recycled as evidence, and the paper openly states its four main limitations. The main caveat is that the applicability of the whole optimality theory hinges on single-item support, which the paper itself acknowledges to be only an idealization for real LLM agents; the framework is best read as a conditional definitional contribution rather than a demonstrated empirical account of how LLM memory works.

major comments (3)
  1. [Section 6, 'Where events interact'] The example is introduced with the statement 'The extraction procedure Φ is decomposable,' yet the same section later states that 'the same events exhibit the alternative: Φ composes events that belong together' and that for the listed pairs Φ({e,e'}) strictly contains Φ({e}) ∪ Φ({e'}). Under the definition of decomposability used in Proposition 3.3(a), these two statements cannot both hold for the same operator: a decomposable Φ must satisfy Φ({e,e'}) = Φ({e}) ∪ Φ({e'}). The frontier in Table 4 and Figure 2 is computed using the decomposable operator, so the composite items introduced in this paragraph appear to belong to a different, non-decomposable variant that is never explicitly defined. Because this example is the paper's only concrete illustration and is advertised as exhibiting 'the synergy of interacting events,' the confusion is load-bearing and must be resolved before the example can support the intended reading.
  2. [Assumption 2.1 and Section 8.2] Single-item support is the hinge of the entire framework: Definition 2.5, Definition 3.1, Proposition 3.2, and Proposition 3.3 all rely on the equivalence between answerability and coverage by a single knowledge item. Section 8.2 correctly concedes that 'an agent that genuinely combines items to answer a query lies outside the account.' This is a scope condition rather than an internal inconsistency, but the abstract and Section 1 present the framework as 'a unified formal account' of large-model agent memory, and Section 7 maps representative LLM systems, many of which perform multi-hop or answer-time reasoning, onto the framework without repeating this caveat. The revision should state the condition as an explicit scope boundary early in the paper, and should indicate how one would check for a given LLM whether Φ actually materializes the composite items on which the validity of the coverage objective depends.
  3. [Section 8.2, coverage-is-not-correctness] The paper optimizes coverage and adopts a 'proxy stance' that sufficient information approximately equals answerability, but the only numerical illustration is a hand-built coverage/precision computation with no reasoning policy. The limitation section says the example 'supports the stance' without establishing it; in fact the example contains no end-to-end reasoner at all, so it cannot provide even suggestive evidence for the proxy stance. Since the optimality results are explicitly about coverage rather than end-to-end correctness, the paper should consistently describe M*_D(S) as a coverage-optimal memory and should not rely on the example to justify the proxy.
minor comments (5)
  1. [Definitions 3.1 and 4.3] The symbol M*_D(S) is reused in Definition 4.3 for a different objective (precision instead of coverage). Renaming the noisy version, for example M*_D,prec(S), would avoid confusion when both frontiers are discussed.
  2. [Table 1] In the Level 1a row, the 'Object of learning' column says 'optimal memory, one-shot,' but at Level 1a the query distribution is known and there is nothing to learn; 'object of computation' would be more accurate.
  3. [Section 6, opening] 'Homer'sOdyssey' is missing a space, and the section alternates between a decomposable Φ and a non-decomposable discussion without explicitly marking the switch; a clear statement of which operator is in force for which calculation is needed.
  4. [Assumption B.1] The notation E⊂N is used for the claim that every event is a degenerate knowledge item; if non-strict inclusion is intended, E⊆N would be clearer, especially since the Appendix discusses self-containment with {e}⊆Φ({e}).
  5. [Section 3.2 and Section 7] The paper says the frontier makes 'how good is a memory' measurable, but U*_D(S) is defined relative to a specific Φ, query distribution pD, and answerability relation ans; the text should explicitly say that this is measurability relative to a formal specification, since obtaining these objects for real systems is left open.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's formal derivation is self-contained, definitional, and makes no empirical prediction that reduces to a fitted input or a self-citation chain.

full rationale

The derivation chain in this paper is definitional rather than empirical. Memory is defined as an event subset (Definition 2.3), knowledge as the span of a generation operator (Definition 2.2), answerability as single-item coverage (Assumption 2.1 and Definition 2.5), and optimal memory as the capacity-constrained maximizer of expected coverage (Definition 3.1). These are explicit formal definitions, not derived predictions. The utility–capacity frontier is the value of that maximization, and its saturation property (Proposition 3.2) follows directly from the stated monotonicity assumption (Assumption B.2), with the paper honestly noting in Section 8.2 and Remark B.3 that real operators are only locally monotone. Proposition 3.3 reduces the optimization to weighted maximum coverage in the decomposable case; this is a genuine reduction to a classical known problem, not a renaming of a result, and it is supported by a proof sketch and an appendix proof. The Odyssey example is explicitly illustrative and hand-checked, computed from manually assigned query masses and spans rather than fitted to data; it is presented as a concrete instantiation of the framework, not as empirical evidence for it. No prediction is recycled as evidence, and no load-bearing argument relies on a self-citation: the paper cites external systems and benchmarks for positioning, but the formal claims stand on their own definitions and assumptions. The paper's own limitation statements (Section 8.2) openly identify the single-item support assumption and the approximation of monotonicity, which further indicates that the framework is not concealing a circular step. Under the requirement to flag circularity only when a specific reduction can be exhibited, no such step exists.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The framework introduces no new physical or empirical entities; its formal constructs (generation operator, knowledge items, span, frontier) are explicit definitions rather than postulated objects with independent evidence. The central claim depends on the assumptions listed above, especially single-item support and monotonicity.

free parameters (4)
  • Query distribution pD(q1..q8) = 0.20, 0.05, 0.10, 0.15, 0.05, 0.20, 0.05, 0.20
    Hand-chosen in Eq. (18) for the Odyssey illustration; the shape of the frontier and the example's conclusions depend on these masses.
  • False claim apparent mass for e-hat = 0.45
    Assigned in Section 6 so the false claim beats the true event e3 and produces a nonzero water-inflation degree.
  • Composite query mass in the variant = 0.05
    Introduced in Section 6 to illustrate cross-event synergy; the marginal calculations shift if this value changes.
  • Event span assignments in Table 3 = Hand-assigned
    The facts and query answer sets for the six Odyssey events are invented for the example; real extraction would produce different spans.
assumptions (5)
  • domain assumption Single-item support (Assumption 2.1)
    Every query must be answerable by a single knowledge item, with all composition pushed into Phi; if false, coverage utility is not an adequate objective.
  • domain assumption Monotonicity of Phi (Assumption B.2)
    More stored events never decrease knowledge or coverage; real LLM extraction is only locally monotone (Remark B.3), so the frontier's monotone shape is fragile.
  • domain assumption Self-containment of Phi (Assumption B.1)
    Storing an event never discards it; a consistency condition for treating events as degenerate knowledge items.
  • domain assumption Existence of query and material distributions (Definition 2.7)
    pD and p(D) are assumed to exist and be stable; in real settings they are unknown and changing, which the sequential MDP addresses only partially.
  • ad hoc to paper Decomposability in the example
    Section 6 assumes Phi is decomposable except for three constructed interacting pairs; this is a special case that real operators meet only roughly.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards a Formal Definition of Agent Memory: Basis, Span, Optimality, and the Sequential Memory Problem." pith.science (2026). https://pith.science/paper/YFQBZVMY

@misc{pith2026260811654,
  author       = {Pith},
  title        = {Pith review of: Towards a Formal Definition of Agent Memory: Basis, Span, Optimality, and the Sequential Memory Problem},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YFQBZVMY}},
  note         = {Machine review of arXiv:2608.11654}
}
read the original abstract

Despite the wide deployment of memory in large-model agents, there is no unified formal account of what a memory is or when it is optimal. This paper takes a first step toward this account. The central idea is that memory is a basis, knowledge is its span, and answerability is a coverage problem: an agent stores events extracted from a material; a generation operator turns any event set into the knowledge it entails; and a query is answerable exactly when some single item in the span covers it. The optimal memory is then the capacity-constrained maximizer of expected coverage, and its value traces a utility--capacity frontier, the common yardstick on which memory systems can be compared. Next, we consider noise in the memory and discuss coverage versus precision under it: a memory may store false claims, so the write policy must infer the truth of what it stores. Drawing an analogy with biological memory, which is formed continuously through ongoing experience, we formalize the continual agent-memory problem in a sequential MDP that covers multiple levels, where memory is the state, writing is the action, and the utility settled at query time is the delayed reward that drives learning. To make the framework concrete, we instantiate it on Homer's \emph{Odyssey}, turning the frontier, the compression zone, and the divergence of coverage from precision into concrete numbers. Finally, we position existing systems within the framework, making ``how good is a memory'' measurable and recasting the open problems of constructing and learning agent memory as concrete research questions.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 37 canonical work pages

  1. [1]

    LongBench : A bilingual, multitask benchmark for long context understanding

    Yushi Bai, Xin Lv, Jiajie Zhang, Hongchang Lyu, Jiankai Tang, Zhidian Huang, Zhengxiao Du, Xiao Liu, Aohan Zeng, Lei Hou, Yuxiao Dong, Jie Tang, and Juanzi Li. LongBench : A bilingual, multitask benchmark for long context understanding. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL), 2024. arXiv:2308.14508

    arXiv 2024

  2. [2]

    Mirrokni

    Ali Behrouz, Peilin Zhong, and Vahab S. Mirrokni. Titans: Learning to memorize at test time. arXiv preprint arXiv:2501.00663, 2025

    arXiv 2025

  3. [3]

    Mem-Gallery : Benchmarking multimodal long-term conversational memory for MLLM agents

    Yuanchen Bei, Tianxin Wei, Xuying Ning, Yanjun Zhao, Zhining Liu, Xiao Lin, Yada Zhu, Hendrik Hamann, Jingrui He, and Hanghang Tong. Mem-Gallery : Benchmarking multimodal long-term conversational memory for MLLM agents. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL), 2026

    work page 2026

  4. [4]

    Walking down the memory maze: Beyond context limit through interactive reading

    Howard Chen, Ramakanth Pasunuru, Jason Weston, and Asli Celikyilmaz. Walking down the memory maze: Beyond context limit through interactive reading. arXiv preprint arXiv:2310.05029, 2023

    arXiv 2023

  5. [5]

    Mem0: Building production-ready AI agents with scalable long-term memory

    Prateek Chhikara, Dev Khant, Saket Aryan, Taranjeet Singh, and Deshraj Yadav. Mem0: Building production-ready AI agents with scalable long-term memory. arXiv preprint arXiv:2504.19413, 2025

    arXiv 2025

  6. [6]

    Yiming Du, Wenyu Huang, Danna Zheng, Zhaowei Wang, Sebastien Montella, Mirella Lapata, Kam-Fai Wong, and Jeff Z. Pan. Rethinking memory in LLM based agents: Representations, operations, and emerging topics. arXiv preprint arXiv:2505.00675, 2025

    arXiv 2025

  7. [7]

    MemoryArena : Benchmarking agent memory in interdependent multi-session agentic tasks

    Zexue He, Yu Wang, Churan Zhi, Yuanzhe Hu, Tzu-Ping Chen, Lang Yin, Ze Chen, Tong Arthur Wu, Siru Ouyang, Zihan Wang, Jiaxin Pei, Julian McAuley, Yejin Choi, and Alex Pentland. MemoryArena : Benchmarking agent memory in interdependent multi-session agentic tasks. In International Conference on Machine Learning (ICML), 2026. arXiv:2602.16313

    arXiv 2026

  8. [8]

    Evaluating memory in LLM agents via incremental multi-turn interactions

    Yuanzhe Hu, Yu Wang, and Julian McAuley. Evaluating memory in LLM agents via incremental multi-turn interactions. In International Conference on Learning Representations (ICLR), 2026. arXiv:2507.05257

    arXiv 2026

Show all 61 references
  1. [9]

    HippoRAG : Neurobiologically inspired long-term memory for large language models

    Bernal Jim \'e nez Guti \'e rrez, Yiheng Shu, Yu Gu, Michihiro Yasunaga, and Yu Su. HippoRAG : Neurobiologically inspired long-term memory for large language models. In Advances in Neural Information Processing Systems (NeurIPS), 2024

    2024

  2. [10]

    From RAG to memory: Non-parametric continual learning for large language models

    Bernal Jim \'e nez Guti \'e rrez, Yiheng Shu, Weijian Qi, Sizhe Zhou, and Yu Su. From RAG to memory: Non-parametric continual learning for large language models. In International Conference on Machine Learning (ICML), 2025. arXiv:2502.14802

    2025 arXiv

  3. [11]

    Needle in a Haystack : Pressure Testing LLMs

    Gregory Kamradt. Needle in a Haystack : Pressure Testing LLMs . GitHub repository, 2023. https://github.com/gkamradt/needle-in-a-haystack

    2023

  4. [12]

    DialSim : A dialogue simulator for evaluating long-term multi-party dialogue understanding of conversational agents

    Jiho Kim, Woosog Chay, Hyeonji Hwang, Daeun Kyung, Hyunseung Chung, Eunbyeol Cho, Yeonsu Kwon, Yohan Jo, and Edward Choi. DialSim : A dialogue simulator for evaluating long-term multi-party dialogue understanding of conversational agents. arXiv preprint arXiv:2406.13144, 2024

    2024

  5. [13]

    McClelland

    Dharshan Kumaran, Demis Hassabis, and James L. McClelland. What learning systems do intelligent agents need? complementary learning systems theory updated. Trends in Cognitive Sciences, 20 0 (7): 0 512--534, 2016

    2016

  6. [14]

    MemOS : A memory OS for AI system

    Zhiyu Li, Chenyang Xi, Chunyu Li, Ding Chen, Boyu Chen, Shichao Song, Simin Niu, Hanyu Wang, Jiawei Yang, Chen Tang, Qingchen Yu, Jihao Zhao, Yezhaohui Wang, Peng Liu, Zehao Lin, Pengyuan Wang, Jiahao Huo, Tianyi Chen, Kai Chen, Kehang Li, Zhen Tao, Huayi Lai, Hao Wu, Bo Tang,...

    2025 arXiv

  7. [15]

    Evaluating very long-term conversational memory of LLM agents

    Adyasha Maharana, Dong-Ho Lee, Sergey Tulyakov, Mohit Bansal, Francesco Barbieri, and Yuwei Fang. Evaluating very long-term conversational memory of LLM agents. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL), 2024. arXiv:2402.17753

    2024 arXiv

  8. [17]

    O'Brien, Carrie J

    Joon Sung Park, Joseph C. O'Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, and Michael S. Bernstein. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology (UIST), 2023

    2023

  9. [18]

    Parth Sarthi, Salman Abdullah, Aditi Tuli, Shubh Khanna, Anna Goldie, and Christopher D. Manning. RAPTOR : Recursive abstractive processing for tree-organized retrieval. In International Conference on Learning Representations (ICLR), 2024. arXiv:2401.18059

    2024 arXiv

  10. [19]

    Mem2ActBench : A benchmark for evaluating long-term memory utilization in task-oriented autonomous agents

    Yiting Shen, Kun Li, Wei Zhou, and Songlin Hu. Mem2ActBench : A benchmark for evaluating long-term memory utilization in task-oriented autonomous agents. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL), 2026

    2026

  11. [20]

    Look back to reason forward: Revisitable memory for long-context LLM agents

    Yaorui Shi, Yuxin Chen, Siyuan Wang, Sihang Li, Hengxing Cai, Qi Gu, Xiang Wang, and An Zhang. Look back to reason forward: Revisitable memory for long-context LLM agents. arXiv preprint arXiv:2509.23040, 2025

    2025

  12. [21]

    Reflexion: Language agents with verbal reinforcement learning

    Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. Reflexion: Language agents with verbal reinforcement learning. In Advances in Neural Information Processing Systems (NeurIPS), 2023

    2023

  13. [22]

    Evaluating long-term memory for long-context question answering

    Alessandra Terranova, Bj \"o rn Ross, and Alexandra Birch. Evaluating long-term memory for long-context question answering. In NeurIPS 2025 Workshop on Metacognition in Generative AI, 2025

    2025

  14. [23]

    Episodic and semantic memory

    Endel Tulving. Episodic and semantic memory. In Endel Tulving and Wayne Donaldson, editors, Organization of Memory, pages 381--403. Academic Press, New York, 1972

    1972

  15. [24]

    MemoryLLM : Towards self-updatable large language models

    Yu Wang, Yifan Gao, Xiusi Chen, Haoming Jiang, Shiyang Li, Jingfeng Yang, Qingyu Yin, Zheng Li, Xian Li, Bing Yin, Jingbo Shang, and Julian McAuley. MemoryLLM : Towards self-updatable large language models. arXiv preprint arXiv:2402.04624, 2024

    2024 arXiv

  16. [25]

    LongMemEval : Benchmarking chat assistants on long-term interactive memory

    Di Wu, Hongwei Wang, Wenhao Yu, Yuwei Zhang, Kai-Wei Chang, and Dong Yu. LongMemEval : Benchmarking chat assistants on long-term interactive memory. In International Conference on Learning Representations (ICLR), 2025. arXiv:2410.10813

    2025 arXiv

  17. [26]

    A-MEM : Agentic memory for LLM agents

    Wujiang Xu, Zujie Liang, Kai Mei, Hang Gao, Juntao Tan, and Yongfeng Zhang. A-MEM : Agentic memory for LLM agents. In Advances in Neural Information Processing Systems (NeurIPS), 2025. arXiv:2502.12110

    2025 arXiv

  18. [27]

    Pan, Hinrich Sch \"u tze, Volker Tresp, and Yunpu Ma

    Sikuan Yan, Xiufeng Yang, Zuchao Huang, Ercong Nie, Zifeng Ding, Zonggen Li, Xiaowen Ma, Jinhe Bi, Kristian Kersting, Jeff Z. Pan, Hinrich Sch \"u tze, Volker Tresp, and Yunpu Ma. Memory-R1 : Enhancing large language model agents to manage and utilize memories via reinforcemen...

    2025 arXiv

  19. [28]

    Memory ^3 : Language modeling with explicit memory

    Hongkang Yang, Zehao Lin, Wenjin Wang, Hao Wu, Zhiyu Li, Bo Tang, Wenqiang Wei, Jinbo Wang, Zeyun Tang, Shichao Song, Chenyang Xi, Yu Yu, Kai Chen, Feiyu Xiong, Linpeng Tang, and Weinan E. Memory ^3 : Language modeling with explicit memory. Journal of Machine Learning, 3: 0 30...

    2024

  20. [29]

    Memagent: Reshaping long-context LLM with multi-conv RL -based memory agent

    Hongli Yu, Tinghong Chen, Jiangtao Feng, Jiangjie Chen, Weinan Dai, Qiying Yu, Ya-Qin Zhang, Wei-Ying Ma, Jingjing Liu, Mingxuan Wang, and Hao Zhou. Memagent: Reshaping long-context LLM with multi-conv RL -based memory agent. In International Conference on Learning Representat...

    2026 arXiv

  21. [30]

    AMA-Bench : Evaluating long-horizon memory for agentic applications

    Yujie Zhao, Boqin Yuan, Junbo Huang, Haocheng Yuan, Zhongming Yu, Haozhou Xu, Lanxiang Hu, Abhilash Shankarampeta, Zimeng Huang, Wentao Ni, Yuandong Tian, and Jishen Zhao. AMA-Bench : Evaluating long-horizon memory for agentic applications. arXiv preprint arXiv:2602.22769, 2026

    2026 arXiv

  22. [31]

    Memorybank: Enhancing large language models with long-term memory

    Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. Memorybank: Enhancing large language models with long-term memory. In Proceedings of the 38th AAAI Conference on Artificial Intelligence (AAAI), 2024. arXiv:2305.10250

    2024 arXiv

  23. [32]

    Zhao, Yujie and Yuan, Boqin and Huang, Junbo and Yuan, Haocheng and Yu, Zhongming and Xu, Haozhou and Hu, Lanxiang and Shankarampeta, Abhilash and Huang, Zimeng and Ni, Wentao and Tian, Yuandong and Zhao, Jishen , title =

  24. [33]

    Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL) , year =

    Bei, Yuanchen and Wei, Tianxin and Ning, Xuying and Zhao, Yanjun and Liu, Zhining and Lin, Xiao and Zhu, Yada and Hamann, Hendrik and He, Jingrui and Tong, Hanghang , title =. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL) , year =

  25. [34]

    and Stoica, Ion and Gonzalez, Joseph E

    Packer, Charles and Wooders, Sarah and Lin, Kevin and Fang, Vivian and Patil, Shishir G. and Stoica, Ion and Gonzalez, Joseph E. , title =. arXiv preprint arXiv:2310.08560 , year =

    arXiv

  26. [35]

    and Cai, Carrie J

    Park, Joon Sung and O'Brien, Joseph C. and Cai, Carrie J. and Morris, Meredith Ringel and Liang, Percy and Bernstein, Michael S. , title =. Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology (UIST) , year =

  27. [36]

    Advances in Neural Information Processing Systems (NeurIPS) , year =

    Jim. Advances in Neural Information Processing Systems (NeurIPS) , year =

  28. [37]

    Jim. From. International Conference on Machine Learning (ICML) , year =

  29. [38]

    Advances in Neural Information Processing Systems (NeurIPS) , year =

    Shinn, Noah and Cassano, Federico and Gopinath, Ashwin and Narasimhan, Karthik and Yao, Shunyu , title =. Advances in Neural Information Processing Systems (NeurIPS) , year =

  30. [39]

    Proceedings of the 38th AAAI Conference on Artificial Intelligence (AAAI) , year =

    Zhong, Wanjun and Guo, Lianghong and Gao, Qiqi and Ye, He and Wang, Yanlin , title =. Proceedings of the 38th AAAI Conference on Artificial Intelligence (AAAI) , year =

  31. [40]

    Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL) , year =

    Maharana, Adyasha and Lee, Dong-Ho and Tulyakov, Sergey and Bansal, Mohit and Barbieri, Francesco and Fang, Yuwei , title =. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL) , year =

  32. [41]

    Chhikara, Prateek and Khant, Dev and Aryan, Saket and Singh, Taranjeet and Yadav, Deshraj , title =

  33. [42]

    Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL) , year =

    Shen, Yiting and Li, Kun and Zhou, Wei and Hu, Songlin , title =. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL) , year =

  34. [43]

    Advances in Neural Information Processing Systems (NeurIPS) , year =

    Xu, Wujiang and Liang, Zujie and Mei, Kai and Gao, Hang and Tan, Juntao and Zhang, Yongfeng , title =. Advances in Neural Information Processing Systems (NeurIPS) , year =

  35. [44]

    Kim, Jiho and Chay, Woosog and Hwang, Hyeonji and Kyung, Daeun and Chung, Hyunseung and Cho, Eunbyeol and Kwon, Yeonsu and Jo, Yohan and Choi, Edward , title =

  36. [45]

    , title =

    Sarthi, Parth and Abdullah, Salman and Tuli, Aditi and Khanna, Shubh and Goldie, Anna and Manning, Christopher D. , title =. International Conference on Learning Representations (ICLR) , year =

  37. [46]

    Chen, Howard and Pasunuru, Ramakanth and Weston, Jason and Celikyilmaz, Asli , title =

  38. [47]

    Kamradt, Gregory , title =

  39. [48]

    Li, Zhiyu and Xi, Chenyang and Li, Chunyu and Chen, Ding and Chen, Boyu and Song, Shichao and Niu, Simin and Wang, Hanyu and Yang, Jiawei and Tang, Chen and Yu, Qingchen and Zhao, Jihao and Wang, Yezhaohui and Liu, Peng and Lin, Zehao and Wang, Pengyuan and Huo, Jiahao and Che...

  40. [49]

    International Conference on Learning Representations (ICLR) , year =

    Yu, Hongli and Chen, Tinghong and Feng, Jiangtao and Chen, Jiangjie and Dai, Weinan and Yu, Qiying and Zhang, Ya-Qin and Ma, Wei-Ying and Liu, Jingjing and Wang, Mingxuan and Zhou, Hao , title =. International Conference on Learning Representations (ICLR) , year =

  41. [50]

    Shi, Yaorui and Chen, Yuxin and Wang, Siyuan and Li, Sihang and Cai, Hengxing and Gu, Qi and Wang, Xiang and Zhang, An , title =

  42. [51]

    Wang, Yu and Gao, Yifan and Chen, Xiusi and Jiang, Haoming and Li, Shiyang and Yang, Jingfeng and Yin, Qingyu and Li, Zheng and Li, Xian and Yin, Bing and Shang, Jingbo and McAuley, Julian , title =

  43. [52]

    Journal of Machine Learning , volume =

    Yang, Hongkang and Lin, Zehao and Wang, Wenjin and Wu, Hao and Li, Zhiyu and Tang, Bo and Wei, Wenqiang and Wang, Jinbo and Tang, Zeyun and Song, Shichao and Xi, Chenyang and Yu, Yu and Chen, Kai and Xiong, Feiyu and Tang, Linpeng and E, Weinan , title =. Journal of Machine Le...

  44. [53]

    , title =

    Behrouz, Ali and Zhong, Peilin and Mirrokni, Vahab S. , title =

  45. [54]

    Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL) , year =

    Bai, Yushi and Lv, Xin and Zhang, Jiajie and Lyu, Hongchang and Tang, Jiankai and Huang, Zhidian and Du, Zhengxiao and Liu, Xiao and Zeng, Aohan and Hou, Lei and Dong, Yuxiao and Tang, Jie and Li, Juanzi , title =. Proceedings of the 62nd Annual Meeting of the Association for ...

  46. [55]

    International Conference on Learning Representations (ICLR) , year =

    Wu, Di and Wang, Hongwei and Yu, Wenhao and Zhang, Yuwei and Chang, Kai-Wei and Yu, Dong , title =. International Conference on Learning Representations (ICLR) , year =

  47. [56]

    Evaluating Long-Term Memory for Long-Context Question Answering , booktitle =

    Terranova, Alessandra and Ross, Bj. Evaluating Long-Term Memory for Long-Context Question Answering , booktitle =

  48. [57]

    International Conference on Learning Representations (ICLR) , year =

    Hu, Yuanzhe and Wang, Yu and McAuley, Julian , title =. International Conference on Learning Representations (ICLR) , year =

  49. [58]

    International Conference on Machine Learning (ICML) , year =

    He, Zexue and Wang, Yu and Zhi, Churan and Hu, Yuanzhe and Chen, Tzu-Ping and Yin, Lang and Chen, Ze and Wu, Tong Arthur and Ouyang, Siru and Wang, Zihan and Pei, Jiaxin and McAuley, Julian and Choi, Yejin and Pentland, Alex , title =. International Conference on Machine Learn...

  50. [59]

    Yan, Sikuan and Yang, Xiufeng and Huang, Zuchao and Nie, Ercong and Ding, Zifeng and Li, Zonggen and Ma, Xiaowen and Bi, Jinhe and Kersting, Kristian and Pan, Jeff Z. and Sch

  51. [60]

    , title =

    Du, Yiming and Huang, Wenyu and Zheng, Danna and Wang, Zhaowei and Montella, Sebastien and Lapata, Mirella and Wong, Kam-Fai and Pan, Jeff Z. , title =

  52. [61]

    Organization of Memory , editor =

    Tulving, Endel , title =. Organization of Memory , editor =

  53. [62]

    , title =

    Kumaran, Dharshan and Hassabis, Demis and McClelland, James L. , title =. Trends in Cognitive Sciences , volume =

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.