Pith. sign in

REVIEW 3 major objections 4 minor 7 cited by

Position: Episodic Memory is the Missing Piece for Long-Term LLM Agents

T0 review · 3 major / 4 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read This paper argues that LLM agents need episodic memory—long-term, explicit, single-shot, instance-specific, contextual—and that current memory methods should be unified under that goal.

desk verdict A clean synthesis and roadmap for episodic memory in LLM agents, but the 'missing piece' claim outruns the evidence. read the letter →

arxiv 2502.06975 v1 pith:2FJ3VH5C submitted 2025-02-10 cs.AI

Mathis Pink , Qinyuan Wu , Vy Ai Vo , Javier Turek , Jianing Mu , Alexander Huth , Mariya Toneva This is my paper · ORCID
classification cs.AI
keywords episodicmemoryLLMagentslong-termcomplementarylearningsystemsin-contextexternalparametricconsolidation
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

This position paper claims that episodic memory—defined by five properties: long-term storage, explicit reasoning, single-shot learning, instance-specific memories, and contextual memories—is the missing piece for building LLM agents that operate effectively over long timescales. The authors argue that current approaches to LLM memory fall into three categories: in-context, external, and parametric, and each covers only a subset of these five properties, so no existing method can learn from a single event, retain it long-term, reason about it explicitly, and place it in context. They propose that treating episodic memory as an explicit, integrated design goal would unify these research directions and catalyze progress toward agents that keep constant computational cost per token while improving with experience. The paper lays out a roadmap with research questions for encoding, retrieval, consolidation, and benchmarks rather than presenting new experimental results.

What carries the argument

The load-bearing object is the five-property operationalization of episodic memory, used as a checklist to show that in-context, external, and parametric memory each satisfy only some properties. The second load-bearing mechanism is the complementary learning systems loop of encoding episodes from in-context memory into an external store, retrieving them back for explicit reasoning, and consolidating them into parametric memory over time. This loop is what would keep computational cost per token constant while allowing performance to improve.

What would settle it

A controlled study showing that an agent without any episodic-memory module, relying only on a very long context window or a standard RAG store, maintains constant per-token cost and improves performance over months of interaction on tasks requiring instance-specific, context-rich recall would directly contradict the claim that episodic memory is the missing piece.

Watch

Extended reading notes

Core claim

The central claim is that LLM agents need an episodic memory system with five specific properties, and that no current approach delivers all of them together. In-context memory provides single-shot, instance-specific, contextual learning but lacks long-term storage; external memory (RAG, graphs, slot-based modules) provides long-term explicit storage but often lacks rich context and single-shot instance learning; parametric updates provide long-term storage and explicit reasoning but are not single-shot or instance-specific. The paper argues that these approaches should be unified under a common episodic-memory framework, grounded in complementary learning systems theory, with a fast external store for episodes and slow consolidation into parameters.

Load-bearing premise

The five properties of episodic memory are asserted as a complete checklist of what long-term LLM agents need, without a derivation or empirical test showing that no other property—such as forgetting, consolidation timing, or motivational salience—is essential.

Editorial extensions

If this is right

  • Any credible long-term agent architecture will need a fast external episodic store plus a slow consolidation path into parameters, not just longer contexts or bigger retrieval databases.
  • Progress in KV-cache compression, RAG, fine-tuning, and knowledge editing should be evaluated against all five episodic properties; current evaluations miss instance-specific and contextual recall.
  • Consolidation becomes a first-class research problem: without moving episodes into parameters, external memory grows without bound and cost cannot stay constant.
  • Benchmarks for long-term agents should test recall of contextualized events after long delays, including temporal order and instance specificity, rather than only fact retrieval.

Reading between the lines

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

  • If the five properties are indeed sufficient, the framework predicts that hybrid architectures combining a fast, content-addressable episodic store with periodic parametric consolidation will outperform any single-mechanism approach on long-horizon agent tasks; this is a direct but untested extension.
  • A testable corollary the paper does not pursue: agents that segment experience into discrete episodes at encoding time should show better single-shot instance recall than agents that store uniform chunks.
  • The framework implicitly suggests that the right unit of memory for LLM agents is the episode, not the token or the fact; that choice, if adopted widely, would change how continual-learning and model-editing benchmarks are designed.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. This position paper argues that episodic memory is the missing piece for building efficient long-term LLM agents. It operationalizes episodic memory for LLM agents through five properties—long-term storage, explicit reasoning, single-shot learning, instance-specific memories, and contextual memories—and contrasts this combination with procedural, semantic, and working memory (Section 2). The paper then classifies current memory research into in-context, external, and parametric memory, using Table 2 to show that no existing family satisfies all five properties (Section 3). It proposes a framework in which experiences are encoded into external memory and periodically consolidated into parametric memory (Figure 1), and it defines six research questions on encoding, retrieval, consolidation, and benchmarks (Section 4). Section 5 defends the framework against two alternative views: indefinitely scaled in-context memory and contextualized external memory.

Significance. If the central claim holds, the paper offers a concrete target that could unify fragmented research on LLM memory, and its six research questions give the community an actionable agenda. The survey of current methods is broad and up-to-date, and the proposed architecture in Figure 1 is a useful synthesis of in-context, external, and parametric memory. The paper's strengths are its explicit property list, its clear enumeration of open questions, especially RQ5 on consolidation, and its willingness to discuss alternative views. The central claim is, however, normative and rests on two unproven premises: that the five properties are necessary and sufficient for long-term agents, and that the proposed consolidation path is cheaper than bounded external memory. These premises materially affect the force of the paper's headline claim, so the contribution is a promising position rather than an established result.

major comments (3)
  1. [Section 5, Figure 1(a), RQ5] The paper's efficiency argument is load-bearing but not quantified. The abstract and Section 1 promise 'efficient' long-term agents, and Section 5 claims that exclusive reliance on external memory 'will still incur high storage costs' while an episodic memory framework 'addresses these constraints' by consolidating into parametric memory. No cost model, asymptotic analysis, or experiment is provided to compare a bounded external vector store with compact embeddings and cheap retrieval against periodic consolidation via fine-tuning, distillation, or knowledge editing, whose expense and interference risks are acknowledged in Section 3.3 and RQ5. If amortized consolidation cost exceeds retrieval from a bounded store at relevant horizons, the efficiency premise fails. The authors should either supply such a comparison or weaken the efficiency claim to a hypothesis.
  2. [Section 2.2, Section 5] The five-property set is presented as the operationalization of episodic memory and as the core of the missing-piece claim, but the paper gives no completeness argument. Section 2.2 motivates each property as 'useful', and Section 5 only rebuts two alternative views (unbounded in-context memory; external memory). Directed forgetting, consolidation timing, and motivational salience are touched on or implicit in the text but never integrated into the necessity/sufficiency claim. The paper should either argue that the five properties are jointly sufficient for effective long-term adaptive behavior or explicitly recast the claim as 'five useful desiderata' rather than 'the missing piece'.
  3. [Table 2, Section 3] The survey conclusion that existing methods 'fall short' is presented through Table 2, but several cells are coded with tildes and the coding rules are never stated. For example, the RAG row marks long-term and explicit with checks but single-shot and instance-specific with tildes, and the text does not define when a method achieves a property versus only possibly achieves it. Without an explicit rubric, the table cannot be independently checked and the falsifiability of the paper's central claim is weakened. Please provide per-cell criteria or move uncertain cells into a more detailed discussion in Section 3.
minor comments (4)
  1. [Section 3.2] The phrase 'a database of pasts LLM inputs' contains a typo; it should read 'past LLM inputs'.
  2. [Section 3.1] The sentence 'In essence, humans avoid SSMs' trade-offs by storing compressed representations and retrieving knowledge adaptively' is confusing because the preceding text attributes compression and approximation to SSMs, not to humans; please rewrite for clarity.
  3. [Section 3.1] The sentence 'a fundamental challenge remain' should be 'a fundamental challenge remains'.
  4. [Table 2] The entries 'In-Context KV-Compression' and 'State-space-model' use inconsistent capitalization and hyphenation relative to the surrounding text; please align the nomenclature.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the episodic-memory position rests on external cognitive-science premises and a survey of methods; two author-overlapping citations are illustrative, not load-bearing.

full rationale

This is a position paper, so there is no fitted parameter, equation, or empirical derivation whose output equals an input. The five-property operationalization of episodic memory is introduced as a useful framing drawn from cognitive science, not as a prediction derived from LLM data. The biological premises are supported by external literature (Tulving; McClelland; Kumaran; Squire and Zola; O'Reilly), and the survey of in-context, external, and parametric memory methods is independent of the paper's own claims. The two author-overlapping citations — Michelmann et al. (2023) for LLM event segmentation and Pink et al. (2024) for an episodic-memory benchmark — are used only as illustrative examples of relevant ongoing work, not as load-bearing support for the central thesis that episodic memory should be a unifying goal. No equation is reused under a new name, no fitted quantity is relabeled as a prediction, and no uniqueness theorem from the authors is invoked to force the framework's choice. The paper explicitly discusses alternative views in Section 5, which would be unnecessary if the conclusion were definitionally forced. The strongest vulnerability is empirical: the claim that consolidating into parametric memory is cheaper or better than bounded external memory is asserted rather than measured, and the sufficiency of the five-property set is not proven. That is a correctness and support concern, not circularity. Accordingly, the paper is self-contained against external benchmarks and receives a low score.

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

The central claim rests on domain assumptions about the relevance of biological memory concepts to LLMs and the sufficiency of the five properties. No free parameters or invented entities are involved.

assumptions (4)
  • ad hoc to paper The five properties of biological episodic memory are necessary and sufficient for LLM-agent long-term memory.
    Section 2.2 lists the properties as 'useful' without derivation or empirical validation, yet the framework treats them as the defining checklist.
  • domain assumption Complementary learning systems theory transfers to artificial LLM agents.
    Section 4 bases the consolidation roadmap on CLS theory (O'Reilly et al., Kumaran et al.) without evidence that LLM parameter updates behave like slow-learning neocortical consolidation.
  • domain assumption LLM agents should have constant computational cost per new token and stable or improving performance over time.
    Section 1 states these as core necessities, but they are design goals, not proven requirements of all long-term agent applications.
  • domain assumption Existing memory methods can be cleanly divided into in-context, external, and parametric categories.
    Section 3 uses this trichotomy; some methods, such as Titans or Larimar, span multiple categories, making the classification a modeling choice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Position: Episodic Memory is the Missing Piece for Long-Term LLM Agents." pith.science (2026). https://pith.science/paper/2FJ3VH5C

@misc{pith2026250206975,
  author       = {Pith},
  title        = {Pith review of: Position: Episodic Memory is the Missing Piece for Long-Term LLM Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2FJ3VH5C}},
  note         = {Machine review of arXiv:2502.06975}
}
read the original abstract

As Large Language Models (LLMs) evolve from text-completion tools into fully fledged agents operating in dynamic environments, they must address the challenge of continually learning and retaining long-term knowledge. Many biological systems solve these challenges with episodic memory, which supports single-shot learning of instance-specific contexts. Inspired by this, we present an episodic memory framework for LLM agents, centered around five key properties of episodic memory that underlie adaptive and context-sensitive behavior. With various research efforts already partially covering these properties, this position paper argues that now is the right time for an explicit, integrated focus on episodic memory to catalyze the development of long-term agents. To this end, we outline a roadmap that unites several research directions under the goal to support all five properties of episodic memory for more efficient long-term LLM agents.

Figures

Figures reproduced from arXiv: 2502.06975 by the authors.

Figure 1
Figure 1. LLM-Agents with an Episodic Memory system. The LLM agent acts on and gets feedback from an environment. Feedback can come in the form of outputs from programs (E1), from other agents (E2), humans (E3), as well as external real-world data (E4). Actions can modify parts of the environment, and provide feedback for humans or other agents in the environment. Within the agent, an external memory system acts as a bridge b… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. LMEB: Long-horizon Memory Embedding Benchmark

    cs.CL 2026-03 unverdicted novelty 7.0 of 10

    LMEB is a new benchmark that evaluates embedding models on long-horizon memory retrieval and shows this skill is largely orthogonal to traditional passage-retrieval performance.

  2. MemCollab: Cross-Model Memory Collaboration via Contrastive Trajectory Distillation

    cs.AI 2026-03 conditional novelty 6.5 of 10

    Contrasting trajectories from heterogeneous LLM agents yields shared abstract reasoning constraints that transfer better than single-model or naively transferred memory.

  3. Try Once, Then Optimal: De-Redundified Procedure Memory for Cross-Episode Exploration Amortization

    cs.RO 2026-07 conditional novelty 6.0 of 10

    Object-centric procedure memory amortizes hidden-state exploration across encounters, cutting robot manipulation operations 16–30% at non-regressing success.

  4. AgentKVShift: Efficient KV Cache Reuse for Agentic Memory Systems

    cs.AI 2026-05 conditional novelty 6.0 of 10

    AgentKVShift corrects all reused KV cache entries in agentic memory by estimating a chunk-level residual offset from a small probe set, recovering near-fresh accuracy at 10–30% recompute ratios.

  5. The Future is Agentic: Definitions, Perspectives, and Open Challenges of Multi-Agent Recommender Systems

    cs.IR 2025-07 conditional novelty 6.0 of 10

    A framework for agentic recommender systems plus a pilot study showing multi-agent pipelines beat a single-shot LLM only on high-diversity user histories.

  6. Embodied AI Agents: Modeling the World

    cs.AI 2025-06 conditional novelty 4.0 of 10

    Embodied AI agents should be built around physical world models plus a mental world model of the user, with virtual, wearable, and robotic agents sharing this core.

  7. Memory-Augmented Transformers: A Systematic Review from Neuroscience Principles to Enhanced Model Architectures

    cs.LG 2025-08 unverdicted novelty 3.0 of 10

    Memory-augmented Transformer research is organized into a three-axis taxonomy bridging neuroscience memory concepts to network designs, but no new result is produced.

Reference graph

Works this paper leans on

114 extracted references · 5 canonical work pages · cited by 7 Pith papers

  1. [1]

    Mnemosyne: Parallelization strategies for efficiently serving multi-million context length llm inference requests without approximations, 2024

    Agrawal, A., Chen, J., Íñigo Goiri, Ramjee, R., Zhang, C., Tumanov, A., and Choukse, E. Mnemosyne: Parallelization strategies for efficiently serving multi-million context length llm inference requests without approximations, 2024. URL https://arxiv.org/abs/2409.17264

    arXiv 2024

  2. [2]

    Forgetting transformer: Softmax attention with a forget gate

    Anonymous. Forgetting transformer: Softmax attention with a forget gate. In Submitted to The Thirteenth International Conference on Learning Representations, 2024. URL https://openreview.net/forum?id=q2Lnyegkr8. under review

    2024

  3. [3]

    Learning fast, learning slow: A general continual learning method based on complementary learning system, 2022

    Arani, E., Sarfraz, F., and Zonooz, B. Learning fast, learning slow: A general continual learning method based on complementary learning system, 2022. URL https://arxiv.org/abs/2201.12604

    arXiv 2022

  4. [4]

    Baddeley, A. D. Working Memory. Clarendon Press, Oxford, UK, 1986

    1986

  5. [5]

    Baddeley, A. D. and Hitch, G. J. Working memory. In Bower, G. H. (ed.), The Psychology of Learning and Motivation: Advances in Research and Theory, volume 8, pp.\ 47--89. Academic Press, New York, 1974

    1974

  6. [6]

    Titans: Learning to memorize at test time, 2024

    Behrouz, A., Zhong, P., and Mirrokni, V. Titans: Learning to memorize at test time, 2024. URL https://arxiv.org/abs/2501.00663

    arXiv 2024

  7. [7]

    Memory layers at scale, 2024

    Berges, V.-P., Oğuz, B., Haziza, D., tau Yih, W., Zettlemoyer, L., and Ghosh, G. Memory layers at scale, 2024. URL https://arxiv.org/abs/2412.09764

    arXiv 2024

  8. [8]

    a is b" fail to learn

    Berglund, L., Tong, M., Kaufmann, M., Balesni, M., Stickland, A. C., Korbak, T., and Evans, O. The reversal curse: Llms trained on" a is b" fail to learn" b is a". arXiv preprint arXiv:2309.12288, 2023

    arXiv 2023

Show all 114 references
  1. [9]

    Large-scale simple question answering with memory networks, 2015

    Bordes, A., Usunier, N., Chopra, S., and Weston, J. Large-scale simple question answering with memory networks, 2015. URL https://arxiv.org/abs/1506.02075

    2015 arXiv

  2. [10]

    Brandon, W., Mishra, M., Nrusimha, A., Panda, R., and Kelly, J. R. Reducing transformer key-value cache size with cross-layer attention. arXiv preprint arXiv:2405.12981, 2024

    2024 arXiv

  3. [11]

    Bulatov, A., Kuratov, Y., and Burtsev, M. S. Recurrent memory transformer, 2022. URL https://arxiv.org/abs/2207.06881

    2022 arXiv

  4. [12]

    D., Aziz, W., and Titov, I

    Cao, N. D., Aziz, W., and Titov, I. Editing factual knowledge in language models, 2021. URL https://arxiv.org/abs/2104.08164

    2021 arXiv

  5. [13]

    Lift yourself up: Retrieval-augmented text generation with self-memory

    Cheng, X., Luo, D., Chen, X., Liu, L., Zhao, D., and Yan, R. Lift yourself up: Retrieval-augmented text generation with self-memory. In Oh, A., Naumann, T., Globerson, A., Saenko, K., Hardt, M., and Levine, S. (eds.), Advances in Neural Information Processing Systems, volume 3...

    2023

  6. [14]

    knowing how

    Cohen, N. J. and Squire, L. R. Preserved learning and retention of pattern-analyzing skill in amnesia: Dissociation of “knowing how” and “knowing that”. Science, 210 0 (4466): 0 207--210, 1980

    1980

  7. [15]

    Understanding memory through hippocampal remapping

    Colgin, L., Moser, E., and Moser, M. Understanding memory through hippocampal remapping. Trends in Neurosciences, 2008

    2008

  8. [16]

    Collins, A. M. and Quillian, M. R. Retrieval time from semantic memory. Journal of Verbal Learning and Verbal Behavior, 8 0 (2): 0 240--247, 1969

    1969

  9. [17]

    Sensory–perceptual episodic memory and its context: Autobiographical memory

    Conway, M. Sensory–perceptual episodic memory and its context: Autobiographical memory. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 2001

    2001

  10. [18]

    Transformer- XL : Attentive language models beyond a fixed-length context

    Dai, Z., Yang, Z., Yang, Y., Carbonell, J., Le, Q., and Salakhutdinov, R. Transformer- XL : Attentive language models beyond a fixed-length context. In Korhonen, A., Traum, D., and M \`a rquez, L. (eds.), Proceedings of the 57th Annual Meeting of the Association for Computatio...

    2019 doi

  11. [19]

    Larimar: Large language models with episodic memory control, 2024 a

    Das, P., Chaudhury, S., Nelson, E., Melnyk, I., Swaminathan, S., Dai, S., Lozano, A., Kollias, G., Chenthamarakshan, V., Jiří, Navrátil, Dan, S., and Chen, P.-Y. Larimar: Large language models with episodic memory control, 2024 a . URL https://arxiv.org/abs/2403.11901

    2024 arXiv

  12. [20]

    Larimar: Large language models with episodic memory control

    Das, P., Chaudhury, S., Nelson, E., et al. Larimar: Large language models with episodic memory control. In Proceedings of the 41st International Conference on Machine Learning (ICML), 2024 b

    2024

  13. [21]

    Skvq: Sliding-window key and value cache quantization for large language models, 2024

    Duanmu, H., Yuan, Z., Li, X., Duan, J., Zhang, X., and Lin, D. Skvq: Sliding-window key and value cache quantization for large language models, 2024. URL https://arxiv.org/abs/2405.06219

    2024 arXiv

  14. [22]

    From local to global: A graph rag approach to query-focused summarization, 2024

    Edge, D., Trinh, H., Cheng, N., Bradley, J., Chao, A., Mody, A., Truitt, S., and Larson, J. From local to global: A graph rag approach to query-focused summarization, 2024. URL https://arxiv.org/abs/2404.16130

    2024 arXiv

  15. [23]

    The hippocampus as a cognitive map

    Eichenbaum, H. The hippocampus as a cognitive map . . . of social space. Neuron, 2015

    2015

  16. [24]

    and Cohen, N

    Eichenbaum, H. and Cohen, N. Can we reconcile the declarative memory and spatial navigation views on hippocampal function? Neuron, 2014

    2014

  17. [25]

    A., Oomerjee, A., Christopoulou, F., Lampouras, G., Bou-Ammar, H., and Wang, J

    Fountas, Z., Benfeghoul, M. A., Oomerjee, A., Christopoulou, F., Lampouras, G., Bou-Ammar, H., and Wang, J. Human-like episodic memory for infinite context llms, 2024. URL https://arxiv.org/abs/2407.09450

    2024

  18. [26]

    Enabling large language models to generate text with citations, 2023

    Gao, T., Yen, H., Yu, J., and Chen, D. Enabling large language models to generate text with citations, 2023. URL https://arxiv.org/abs/2305.14627

    2023 arXiv

  19. [27]

    Model tells you what to discard: Adaptive kv cache compression for llms, 2024

    Ge, S., Zhang, Y., Liu, L., Zhang, M., Han, J., and Gao, J. Model tells you what to discard: Adaptive kv cache compression for llms, 2024. URL https://arxiv.org/abs/2310.01801

    2024 arXiv

  20. [28]

    Goldfinch: High performance rwkv/transformer hybrid with linear pre-fill and extreme kv-cache compression, 2024

    Goldstein, D., Obeid, F., Alcaide, E., Song, G., and Cheah, E. Goldfinch: High performance rwkv/transformer hybrid with linear pre-fill and extreme kv-cache compression, 2024. URL https://arxiv.org/abs/2407.12077

    2024 arXiv

  21. [29]

    Neural turing machines, 2014

    Graves, A., Wayne, G., and Danihelka, I. Neural turing machines, 2014. URL https://arxiv.org/abs/1410.5401

    2014 arXiv

  22. [30]

    G., Grefenstette, E., Ramalho, T., Agapiou, J., et al

    Graves, A., Wayne, G., Reynolds, M., Harley, T., Danihelka, I., Grabska-Barwi \'n ska, A., Colmenarejo, S. G., Grefenstette, E., Ramalho, T., Agapiou, J., et al. Hybrid computing using a neural network with dynamic external memory. Nature, 538 0 (7626): 0 471--476, 2016

    2016

  23. [31]

    and Dao, T

    Gu, A. and Dao, T. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752, 2023

    2023 arXiv

  24. [32]

    Model editing at scale leads to gradual and catastrophic forgetting, 2024

    Gupta, A., Rao, A., and Anumanchipalli, G. Model editing at scale leads to gradual and catastrophic forgetting, 2024. URL https://arxiv.org/abs/2401.07453

    2024 arXiv

  25. [33]

    J., Shu, Y., Gu, Y., Yasunaga, M., and Su, Y

    Gutiérrez, B. J., Shu, Y., Gu, Y., Yasunaga, M., and Su, Y. Hipporag: Neurobiologically inspired long-term memory for large language models, 2025. URL https://arxiv.org/abs/2405.14831

    2025 arXiv

  26. [34]

    and Schwartz, B

    Hampton, R. and Schwartz, B. Episodic memory in nonhumans: what, and where, is when? Current Opinion in Neurobiology, 2004

    2004

  27. [35]

    LM -infinite: Zero-shot extreme length generalization for large language models

    Han, C., Wang, Q., Peng, H., Xiong, W., Chen, Y., Ji, H., and Wang, S. LM -infinite: Zero-shot extreme length generalization for large language models. In Duh, K., Gomez, H., and Bethard, S. (eds.), Proceedings of the 2024 Conference of the North American Chapter of the Associ...

    2024 doi

  28. [36]

    W., Shao, Y

    Hooper, C., Kim, S., Mohammadzadeh, H., Mahoney, M. W., Shao, Y. S., Keutzer, K., and Gholami, A. Kvquant: Towards 10 million context length llm inference with kv cache quantization, 2024. URL https://arxiv.org/abs/2401.18079

    2024 arXiv

  29. [37]

    J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., and Chen, W

    Hu, E. J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., and Chen, W. Lo RA : Low-rank adaptation of large language models. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=nZeVKeeFYf9

    2022

  30. [38]

    B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., and Amodei, D

    Kaplan, J., McCandlish, S., Henighan, T., Brown, T. B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., and Amodei, D. Scaling laws for neural language models, 2020. URL https://arxiv.org/abs/2001.08361

    2020 arXiv

  31. [39]

    Transformers are rnns: Fast autoregressive transformers with linear attention

    Katharopoulos, A., Vyas, A., Pappas, N., and Fleuret, F. Transformers are rnns: Fast autoregressive transformers with linear attention. In International conference on machine learning, pp.\ 5156--5165. PMLR, 2020

    2020

  32. [40]

    Generalization through memorization: Nearest neighbor language models

    Khandelwal, U., Levy, O., Jurafsky, D., Zettlemoyer, L., and Lewis, M. Generalization through memorization: Nearest neighbor language models. In International Conference on Learning Representations, 2020. URL https://openreview.net/forum?id=HklBjCEKvH

    2020

  33. [41]

    Kirsten, E., Habernal, I., Nanda, V., and Zafar, M. B. The impact of inference acceleration strategies on bias of llms. arXiv preprint arXiv:2410.22118, 2024

    2024 arXiv

  34. [42]

    Kumaran, D., Hassabis, D., and McClelland, J. L. What learning systems do intelligent agents need? complementary learning systems theory updated. Trends in Cognitive Sciences, 20 0 (7): 0 512–534, July 2016. ISSN 1364-6613. doi:10.1016/j.tics.2016.05.004. URL http://dx.doi.org...

    2016 doi

  35. [43]

    H., Gonzalez, J., Zhang, H., and Stoica, I

    Kwon, W., Li, Z., Zhuang, S., Sheng, Y., Zheng, L., Yu, C. H., Gonzalez, J., Zhang, H., and Stoica, I. Efficient memory management for large language model serving with pagedattention. In Proceedings of the 29th Symposium on Operating Systems Principles, SOSP ’23, pp.\ 611–626...

    2023

  36. [44]

    Prompted llms as chatbot modules for long open-domain conversation

    Lee, G., Hartmann, V., Park, J., Papailiopoulos, D., and Lee, K. Prompted llms as chatbot modules for long open-domain conversation. In Findings of the Association for Computational Linguistics: ACL 2023. Association for Computational Linguistics, 2023. doi:10.18653/v1/2023.fi...

    2023 doi

  37. [45]

    Infinigen: Efficient generative inference of large language models with dynamic kv cache management, 2024

    Lee, W., Lee, J., Seo, J., and Sim, J. Infinigen: Efficient generative inference of large language models with dynamic kv cache management, 2024. URL https://arxiv.org/abs/2406.19707

    2024 arXiv

  38. [46]

    Linear attention mechanism: An efficient attention for semantic segmentation

    Li, R., Su, J., Duan, C., and Zheng, S. Linear attention mechanism: An efficient attention for semantic segmentation. arXiv preprint arXiv:2007.14902, 2020

    2007 arXiv

  39. [47]

    Graphreader: Building graph-based agent to enhance long-context abilities of large language models, 2024 a

    Li, S., He, Y., Guo, H., Bu, X., Bai, G., Liu, J., Liu, J., Qu, X., Li, Y., Ouyang, W., Su, W., and Zheng, B. Graphreader: Building graph-based agent to enhance long-context abilities of large language models, 2024 a . URL https://arxiv.org/abs/2406.14550

    2024 arXiv

  40. [48]

    Personal llm agents: Insights and survey about the capability, efficiency and security, 2024 b

    Li, Y., Wen, H., Wang, W., Li, X., Yuan, Y., Liu, G., Liu, J., Xu, W., Wang, X., Sun, Y., Kong, R., Wang, Y., Geng, H., Luan, J., Jin, X., Ye, Z., Xiong, G., Zhang, F., Li, X., Xu, M., Li, Z., Li, P., Liu, Y., Zhang, Y.-Q., and Liu, Y. Personal llm agents: Insights and survey ...

    2024 arXiv

  41. [49]

    and Losonczy, A

    Liao, Z. and Losonczy, A. Learning, fast and slow: Single- and many-shot learning in the hippocampus. Annual Review of Neuroscience, 2024

    2024

  42. [50]

    Infinite-llm: Efficient llm service for long context with distattention and distributed kvcache, 2024

    Lin, B., Zhang, C., Peng, T., Zhao, H., Xiao, W., Sun, M., Liu, A., Zhang, Z., Li, L., Qiu, X., Li, S., Ji, Z., Xie, T., Li, Y., and Lin, W. Infinite-llm: Efficient llm service for long context with distattention and distributed kvcache, 2024. URL https://arxiv.org/abs/2401.02669

    2024 arXiv

  43. [51]

    Y., and Lu, O

    Lin, C.-C., Huang, A. Y., and Lu, O. H. Artificial intelligence in intelligent tutoring systems toward sustainable education: a systematic review. Smart Learning Environments, 10 0 (1): 0 41, 2023

    2023

  44. [52]

    Minicache: Kv cache compression in depth dimension for large language models, 2024 a

    Liu, A., Liu, J., Pan, Z., He, Y., Haffari, G., and Zhuang, B. Minicache: Kv cache compression in depth dimension for large language models, 2024 a . URL https://arxiv.org/abs/2405.14366

    2024 arXiv

  45. [53]

    Think-in-memory: Recalling and post-thinking enable llms with long-term memory, 2023 a

    Liu, L., Yang, X., Shen, Y., Hu, B., Zhang, Z., Gu, J., and Zhang, G. Think-in-memory: Recalling and post-thinking enable llms with long-term memory, 2023 a . URL https://arxiv.org/abs/2311.08719

    2023 arXiv

  46. [54]

    F., Lin, K., Hewitt, J., Paranjape, A., Bevilacqua, M., Petroni, F., and Liang, P

    Liu, N. F., Lin, K., Hewitt, J., Paranjape, A., Bevilacqua, M., Petroni, F., and Liang, P. Lost in the middle: How language models use long contexts. Transactions of the Association for Computational Linguistics, 12: 0 157--173, 2024 b

    2024

  47. [55]

    Scissorhands: Exploiting the persistence of importance hypothesis for llm kv cache compression at test time

    Liu, Z., Desai, A., Liao, F., Wang, W., Xie, V., Xu, Z., Kyrillidis, A., and Shrivastava, A. Scissorhands: Exploiting the persistence of importance hypothesis for llm kv cache compression at test time. In Oh, A., Naumann, T., Globerson, A., Saenko, K., Hardt, M., and Levine, S...

    2023

  48. [56]

    Kivi: A tuning-free asymmetric 2bit quantization for kv cache

    Liu, Z., Yuan, J., Jin, H., Zhong, S., Xu, Z., Braverman, V., Chen, B., and Hu, X. Kivi: A tuning-free asymmetric 2bit quantization for kv cache. arXiv preprint arXiv:2402.02750, 2024 c

    2024 arXiv

  49. [57]

    Sparser is faster and less is more: Efficient sparse attention for long-range transformers

    Lou, C., Jia, Z., Zheng, Z., and Tu, K. Sparser is faster and less is more: Efficient sparse attention for long-range transformers. arXiv preprint arXiv:2406.16747, 2024

    2024 arXiv

  50. [58]

    Memochat: Tuning llms to use memos for consistent long-range open-domain conversation, 2023

    Lu, J., An, S., Lin, M., Pergola, G., He, Y., Yin, D., Sun, X., and Wu, Y. Memochat: Tuning llms to use memos for consistent long-range open-domain conversation, 2023. URL https://arxiv.org/abs/2308.08239

    2023 arXiv

  51. [59]

    and Roberts, N

    Mayes, A. and Roberts, N. Theories of episodic memory. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 2001

    2001

  52. [60]

    Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory

    McClelland, J., McNaughton, B., and O'Reilly, R. Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory. Psychological Review, 1995

    1995

  53. [61]

    Locating and editing factual associations in gpt, 2023 a

    Meng, K., Bau, D., Andonian, A., and Belinkov, Y. Locating and editing factual associations in gpt, 2023 a . URL https://arxiv.org/abs/2202.05262

    2023 arXiv

  54. [62]

    S., Andonian, A., Belinkov, Y., and Bau, D

    Meng, K., Sharma, A. S., Andonian, A., Belinkov, Y., and Bau, D. Mass-editing memory in a transformer, 2023 b . URL https://arxiv.org/abs/2210.07229

    2023 arXiv

  55. [63]

    A., and Toneva, M

    Michelmann, S., Kumar, M., Norman, K. A., and Toneva, M. Large language models can segment narrative events similarly to humans, 2023. URL https://arxiv.org/abs/2301.10297

    2023 arXiv

  56. [64]

    Les troubles de la mémoire accompagnant des lésions hippocampiques bilatérales

    Milner, B. Les troubles de la mémoire accompagnant des lésions hippocampiques bilatérales. Psychologie Médicale, 51: 0 39--52, 1962

    1962

  57. [65]

    D., and Finn, C

    Mitchell, E., Lin, C., Bosselut, A., Manning, C. D., and Finn, C. Memory-based model editing at scale, 2022. URL https://arxiv.org/abs/2206.06520

    2022 arXiv

  58. [66]

    Memllm: Finetuning llms to use an explicit read-write memory, 2025

    Modarressi, A., Köksal, A., Imani, A., Fayyaz, M., and Schütze, H. Memllm: Finetuning llms to use an explicit read-write memory, 2025. URL https://arxiv.org/abs/2404.11672

    2025 arXiv

  59. [67]

    Meta knowledge for retrieval augmented large language models, 2024

    Mombaerts, L., Ding, T., Banerjee, A., Felice, F., Taws, J., and Borogovac, T. Meta knowledge for retrieval augmented large language models, 2024. URL https://arxiv.org/abs/2408.09017

    2024 arXiv

  60. [68]

    Nawrot, P., Łańcucki, A., Chochowski, M., Tarjan, D., and Ponti, E. M. Dynamic memory compression: Retrofitting llms for accelerated inference, 2024. URL https://arxiv.org/abs/2403.09636

    2024 arXiv

  61. [69]

    and Norman, K

    O'Reilly, R. and Norman, K. Hippocampal and neocortical contributions to memory: Advances in the complementary learning systems framework. Trends in Cognitive Sciences, 2002

    2002

  62. [70]

    Complementary learning systems

    O'Reilly, R., Bhattacharyya, R., Howard, M., and Ketz, N. Complementary learning systems. Cognitive Science, 2014

    2014

  63. [71]

    and Nadel, L

    O’Keefe, J. and Nadel, L. The Hippocampus as a Cognitive Map. Oxford: Clarendon Press, 1978

    1978

  64. [72]

    G., Stoica, I., and Gonzalez, J

    Packer, C., Wooders, S., Lin, K., Fang, V., Patil, S. G., Stoica, I., and Gonzalez, J. E. Memgpt: Towards llms as operating systems, 2024. URL https://arxiv.org/abs/2310.08560

    2024 arXiv

  65. [73]

    Propagating knowledge updates to lms through distillation

    Padmanabhan, S., Onoe, Y., Zhang, M., Durrett, G., and Choi, E. Propagating knowledge updates to lms through distillation. In Oh, A., Naumann, T., Globerson, A., Saenko, K., Hardt, M., and Levine, S. (eds.), Advances in Neural Information Processing Systems, volume 36, pp.\ 47...

    2023

  66. [74]

    F., He, X., Chen, W., and Wang, L

    Pang, J., Ye, F., Wong, D. F., He, X., Chen, W., and Wang, L. Anchor-based large language models, 2024. URL https://arxiv.org/abs/2402.07616

    2024 arXiv

  67. [75]

    K., He, X., Hou, H., Lin, J., Kazienko, P., Kocon, J., Kong, J., Koptyra, B., Lau, H., Mantri, K

    Peng, B., Alcaide, E., Anthony, Q., Albalak, A., Arcadinho, S., Biderman, S., Cao, H., Cheng, X., Chung, M., Grella, M., GV, K. K., He, X., Hou, H., Lin, J., Kazienko, P., Kocon, J., Kong, J., Koptyra, B., Lau, H., Mantri, K. S. I., Mom, F., Saito, A., Song, G., Tang, X., Wang...

    2023 arXiv

  68. [76]

    Graph retrieval-augmented generation: A survey, 2024

    Peng, B., Zhu, Y., Liu, Y., Bo, X., Shi, H., Hong, C., Zhang, Y., and Tang, S. Graph retrieval-augmented generation: A survey, 2024. URL https://arxiv.org/abs/2408.08921

    2024 arXiv

  69. [77]

    A., Wu, Q., Mu, J., Turek, J

    Pink, M., Vo, V. A., Wu, Q., Mu, J., Turek, J. S., Hasson, U., Norman, K. A., Michelmann, S., Huth, A., and Toneva, M. Assessing episodic memory in llms with sequence order recall tasks, 2024. URL https://arxiv.org/abs/2410.08133

    2024 arXiv

  70. [78]

    Agent laboratory: Using llm agents as research assistants

    Schmidgall, S., Su, Y., Wang, Z., Sun, X., Wu, J., Yu, X., Liu, J., Liu, Z., and Barsoum, E. Agent laboratory: Using llm agents as research assistants. arXiv preprint arXiv:2501.04227, 2025

    2025 arXiv

  71. [79]

    and Evans, S

    Schwartz, B. and Evans, S. Episodic memory in primates. American Journal of Primatology: Official Journal of the American Society of Primatologists, 2001

    2001

  72. [80]

    Learning by distilling context, 2022

    Snell, C., Klein, D., and Zhong, R. Learning by distilling context, 2022. URL https://arxiv.org/abs/2209.15189

    2022 arXiv

  73. [81]

    and Zola, S

    Squire, L. and Zola, S. Structure and function of declarative and nondeclarative memory systems. PNAS, 1996

    1996

  74. [82]

    and Moser, M

    Sugar, J. and Moser, M. Episodic memory: Neuronal codes for what, where, and when. Hippocampus, 2019

    2019

  75. [83]

    End-to-end memory networks

    Sukhbaatar, S., Szlam, A., Weston, J., and Fergus, R. End-to-end memory networks. In Cortes, C., Lawrence, N., Lee, D., Sugiyama, M., and Garnett, R. (eds.), Advances in Neural Information Processing Systems, volume 28. Curran Associates, Inc., 2015. URL https://arxiv.org/abs/...

    2015 arXiv

  76. [84]

    You only cache once: Decoder-decoder architectures for language models, 2024 a

    Sun, Y., Dong, L., Zhu, Y., Huang, S., Wang, W., Ma, S., Zhang, Q., Wang, J., and Wei, F. You only cache once: Decoder-decoder architectures for language models, 2024 a . URL https://arxiv.org/abs/2405.05254

    2024 arXiv

  77. [85]

    Learning to (learn at test time): Rnns with expressive hidden states, 2024 b

    Sun, Y., Li, X., Dalal, K., Xu, J., Vikram, A., Zhang, G., Dubois, Y., Chen, X., Wang, X., Koyejo, S., Hashimoto, T., and Guestrin, C. Learning to (learn at test time): Rnns with expressive hidden states, 2024 b . URL https://arxiv.org/abs/2407.04620

    2024 arXiv

  78. [86]

    Massive editing for large language models via meta learning, 2024

    Tan, C., Zhang, G., and Fu, J. Massive editing for large language models via meta learning, 2024. URL https://arxiv.org/abs/2311.04661

    2024 arXiv

  79. [87]

    Razorattention: Efficient kv cache compression through retrieval heads, 2024

    Tang, H., Lin, Y., Lin, J., Han, Q., Hong, S., Yao, Y., and Wang, G. Razorattention: Efficient kv cache compression through retrieval heads, 2024. URL https://arxiv.org/abs/2407.15891

    2024 arXiv

  80. [88]

    Episodic and semantic memory

    Tulving, E. Episodic and semantic memory. In Tulving, E. and Donaldson, W. (eds.), Organization of Memory, pp.\ 381--403. Academic Press, New York, 1972

    1972

  81. [89]

    Dylora: Parameter efficient tuning of pre-trained models using dynamic search-free low rank adaptation, 2022

    Valipour, M., Rezagholizadeh, M., Kobyzev, I., and Ghodsi, A. Dylora: Parameter efficient tuning of pre-trained models using dynamic search-free low rank adaptation, 2022

    2022

  82. [90]

    Enhancing large language model with self-controlled memory framework, 2024 a

    Wang, B., Liang, X., Yang, J., Huang, H., Wu, S., Wu, P., Lu, L., Ma, Z., and Li, Z. Enhancing large language model with self-controlled memory framework, 2024 a . URL https://arxiv.org/abs/2304.13343

    2024 arXiv

  83. [91]

    Roselora: Row and column-wise sparse low-rank adaptation of pre-trained language model for knowledge editing and fine-tuning, 2024 b

    Wang, H., Liu, T., Li, R., Cheng, M., Zhao, T., and Gao, J. Roselora: Row and column-wise sparse low-rank adaptation of pre-trained language model for knowledge editing and fine-tuning, 2024 b . URL https://arxiv.org/abs/2406.10777

    2024 arXiv

  84. [92]

    M., and Dao, T

    Wang, J., Paliotta, D., May, A., Rush, A. M., and Dao, T. The mamba in the llama: Distilling and accelerating hybrid models, 2025. URL https://arxiv.org/abs/2408.15237

    2025 arXiv

  85. [93]

    Wise: Rethinking the knowledge memory for lifelong model editing of large language models, 2024 c

    Wang, P., Li, Z., Zhang, N., Xu, Z., Yao, Y., Jiang, Y., Xie, P., Huang, F., and Chen, H. Wise: Rethinking the knowledge memory for lifelong model editing of large language models, 2024 c . URL https://arxiv.org/abs/2405.14768

    2024 arXiv

  86. [94]

    Memformer: A memory-augmented transformer for sequence modeling

    Wu, Q., Lan, Z., Qian, K., Gu, J., Geramifard, A., and Yu, Z. Memformer: A memory-augmented transformer for sequence modeling. In Findings of the Association for Computational Linguistics: AACL-IJCNLP 2022, pp.\ 308--318. Association for Computational Linguistics, November 202...

    2022 doi

  87. [95]

    The kanerva machine: A generative distributed memory

    Wu, Y., Wayne, G., Graves, A., and Lillicrap, T. The kanerva machine: A generative distributed memory. In International Conference on Learning Representations, 2018. URL https://openreview.net/forum?id=S1HlA-ZAZ

    2018

  88. [96]

    N., Hutchins, D., and Szegedy, C

    Wu, Y., Rabe, M. N., Hutchins, D., and Szegedy, C. Memorizing transformers. In International Conference on Learning Representations, 2022 b . URL https://openreview.net/forum?id=TrjbxzRcnf-

    2022

  89. [97]

    D., and Potts, C

    Wu, Z., Arora, A., Wang, Z., Geiger, A., Jurafsky, D., Manning, C. D., and Potts, C. Reft: Representation finetuning for language models, 2024. URL https://arxiv.org/abs/2404.03592

    2024 arXiv

  90. [98]

    The rise and potential of large language model based agents: A survey, 2023

    Xi, Z., Chen, W., Guo, X., He, W., Ding, Y., Hong, B., Zhang, M., Wang, J., Jin, S., Zhou, E., Zheng, R., Fan, X., Wang, X., Xiong, L., Zhou, Y., Wang, W., Jiang, C., Zou, Y., Liu, X., Yin, Z., Dou, S., Weng, R., Cheng, W., Zhang, Q., Qin, W., Zheng, Y., Qiu, X., Huang, X., an...

    2023 arXiv

  91. [99]

    Infllm: Training-free long-context extrapolation for llms with an efficient context memory, 2024

    Xiao, C., Zhang, P., Han, X., Xiao, G., Lin, Y., Zhang, Z., Liu, Z., and Sun, M. Infllm: Training-free long-context extrapolation for llms with an efficient context memory, 2024. URL https://arxiv.org/abs/2402.04617

    2024 arXiv

  92. [100]

    Raise a child in large language model: Towards effective and generalizable fine-tuning

    Xu, R., Luo, F., Zhang, Z., Tan, C., Chang, B., Huang, S., and Huang, F. Raise a child in large language model: Towards effective and generalizable fine-tuning. In Moens, M.-F., Huang, X., Specia, L., and Yih, S. W.-t. (eds.), Proceedings of the 2021 Conference on Empirical Me...

    2021 doi

  93. [101]

    TIES -merging: Resolving interference when merging models

    Yadav, P., Tam, D., Choshen, L., Raffel, C., and Bansal, M. TIES -merging: Resolving interference when merging models. In Thirty-seventh Conference on Neural Information Processing Systems, 2023. URL https://openreview.net/forum?id=xtaX3WyCj1

    2023

  94. [102]

    Pyramidinfer: Pyramid kv cache compression for high-throughput llm inference, 2024 a

    Yang, D., Han, X., Gao, Y., Hu, Y., Zhang, S., and Zhao, H. Pyramidinfer: Pyramid kv cache compression for high-throughput llm inference, 2024 a . URL https://arxiv.org/abs/2405.12532

    2024 arXiv

  95. [103]

    Do large language models latently perform multi-hop reasoning? arXiv preprint arXiv:2402.16837, 2024 b

    Yang, S., Gribovskaya, E., Kassner, N., Geva, M., and Riedel, S. Do large language models latently perform multi-hop reasoning? arXiv preprint arXiv:2402.16837, 2024 b

    2024 arXiv

  96. [104]

    Chunkattention: Efficient self-attention with prefix-aware kv cache and two-phase partition, 2024 a

    Ye, L., Tao, Z., Huang, Y., and Li, Y. Chunkattention: Efficient self-attention with prefix-aware kv cache and two-phase partition, 2024 a . URL https://arxiv.org/abs/2402.15220

    2024 arXiv

  97. [105]

    Differential transformer, 2024 b

    Ye, T., Dong, L., Xia, Y., Sun, Y., Zhu, Y., Huang, G., and Wei, F. Differential transformer, 2024 b . URL https://arxiv.org/abs/2410.05258

    2024 arXiv

  98. [106]

    Long-context language modeling with parallel context encoding, 2024

    Yen, H., Gao, T., and Chen, D. Long-context language modeling with parallel context encoding, 2024. URL https://arxiv.org/abs/2402.16617

    2024 arXiv

  99. [107]

    Lofit: Localized fine-tuning on llm representations, 2024

    Yin, F., Ye, X., and Durrett, G. Lofit: Localized fine-tuning on llm representations, 2024. URL https://arxiv.org/abs/2406.01563

    2024 arXiv

  100. [108]

    Adaptive semiparametric language models

    Yogatama, D., de Masson d’Autume, C., and Kong, L. Adaptive semiparametric language models. Transactions of the Association for Computational Linguistics, 9: 0 362--373, 04 2021. ISSN 2307-387X. doi:10.1162/tacl_a_00371. URL https://doi.org/10.1162/tacl\_a\_00371

    2021 doi

  101. [109]

    Melo: Enhancing model editing with neuron-indexed dynamic lora, 2023

    Yu, L., Chen, Q., Zhou, J., and He, L. Melo: Enhancing model editing with neuron-indexed dynamic lora, 2023. URL https://arxiv.org/abs/2312.11795

    2023 arXiv

  102. [110]

    In defense of rag in the era of long-context language models, 2024

    Yu, T., Xu, A., and Akkiraju, R. In defense of rag in the era of long-context language models, 2024. URL https://arxiv.org/abs/2409.01666

    2024 arXiv

  103. [111]

    Wkvquant: Quantizing weight and key/value cache for large language models gains more, 2024

    Yue, Y., Yuan, Z., Duanmu, H., Zhou, S., Wu, J., and Nie, L. Wkvquant: Quantizing weight and key/value cache for large language models gains more, 2024. URL https://arxiv.org/abs/2402.12065

    2024 arXiv

  104. [112]

    H., Cao, S., Kozyrakis, C., Stoica, I., Gonzalez, J

    Zheng, L., Yin, L., Xie, Z., Sun, C., Huang, J., Yu, C. H., Cao, S., Kozyrakis, C., Stoica, I., Gonzalez, J. E., et al. Sglang: Efficient execution of structured language model programs. arXiv preprint arXiv:2312.07104, 2024

    2024 arXiv

  105. [113]

    Memorybank: Enhancing large language models with long-term memory

    Zhong, W., Guo, L., Gao, Q., Ye, H., and Wang, Y. Memorybank: Enhancing large language models with long-term memory. Proceedings of the AAAI Conference on Artificial Intelligence, 38 0 (17), Mar. 2024. doi:10.1609/aaai.v38i17.29946. URL https://ojs.aaai.org/index.php/AAAI/arti...

    2024 doi

  106. [114]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.