pith. machine review for the scientific record. sign in

arxiv: 2605.13872 · v1 · submitted 2026-05-05 · 💻 cs.NE · cs.AI

Recognition: 2 theorem links

· Lean Theorem

S-AI-Recursive: A Bio-Inspired and Temporal Sparse AI Architecture for Iterative, Introspective, and Energy-Frugal Reasoning

Said Slaoui
Authors on Pith no claims yet

Pith reviewed 2026-05-15 06:38 UTC · model grok-4.3

classification 💻 cs.NE cs.AI
keywords sparse AIrecursive reasoningbio-inspired architecturehormonal controliterative refinementtemporal parsimonyLyapunov stability
0
0 comments X

The pith

Reasoning emerges from iterative hormonal feedback in small AI models rather than from wide feed-forward layers.

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

The paper presents S-AI-Recursive as a system that turns reasoning into a closed hormonal loop instead of a single pass through a large network. Two new signals, Clarifine for convergence and Confusionin for uncertainty, push the internal state through repeated cycles until it settles at a stable point. A reader would care because the approach claims that adding time for refinement can replace the need for millions more parameters. Experiments on abstract and symbolic tasks show that models under ten million parameters reach competitive accuracy when allowed to iterate.

Core claim

S-AI-Recursive formalizes reasoning as a Recursive Reasoning Cycle governed by Clarifine and Confusionin hormones whose antagonistic action drives state refinement, supported by Lyapunov stability, entropic contraction, and a finite-time hormonal stopping rule; the resulting system matches larger models on symbolic benchmarks while using fewer than ten million parameters, establishing that iterative cognitive depth can stand in for architectural width.

What carries the argument

The Recursive Reasoning Cycle, a dynamical system in which Clarifine and Confusionin hormones regulate iterative state updates toward cognitive equilibrium.

Load-bearing premise

The hormones and recursive dynamics produce genuine iterative refinement and stable convergence rather than benchmark-specific behavior.

What would settle it

A new abstract reasoning benchmark on which increasing the number of iterations produces no accuracy gain or on which the state fails to converge within the budgeted steps.

read the original abstract

This article introduces S-AI-Recursive, a bio-inspired Sparse Artificial Intelligence architecture in which reasoning is operationalized as a hormonal closed-loop iteration rather than a single feed-forward pass. Building upon the S-AI foundational framework [1], the hormonal-probabilistic unification doctrine [2], and the formal mathematical methodology established in S-AI-IoT [3], the present work formalizes the Recursive Reasoning Cycle (RRC) as a dynamical system governed by two novel hormones: Clarifine, a convergence signal, and Confusionin, an uncertainty detector, whose antagonistic regulation drives iterative state refinement toward a stable cognitive equilibrium. The complete mathematical framework is developed, including recursive state dynamics, Lyapunov stability proof, entropic contraction theorem, hormonal stopping criterion with finite-time termination guarantee, Euler-Maruyama discretization with projection, primal-dual agent selection under iteration budget, and recursive engram memory with warm-start acceleration. Experimental validation on the SAI-UT+ testbench demonstrates that S-AI-Recursive achieves competitive reasoning performance on abstract and symbolic benchmarks with fewer than ten million parameters, confirming the central principle of temporal parsimony: iterative cognitive depth substitutes for architectural width.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The paper introduces S-AI-Recursive, a bio-inspired sparse AI architecture that operationalizes reasoning as a hormonal closed-loop iteration via the Recursive Reasoning Cycle (RRC) driven by two novel hormones, Clarifine (convergence signal) and Confusionin (uncertainty detector). Building on prior S-AI works, it claims to develop a full mathematical framework including recursive state dynamics, Lyapunov stability proof, entropic contraction theorem, hormonal stopping criterion with finite-time termination, Euler-Maruyama discretization, primal-dual selection, and recursive engram memory, plus experimental validation on the SAI-UT+ testbench showing competitive performance on abstract/symbolic benchmarks with under 10 million parameters, supporting temporal parsimony where iterative depth substitutes for width.

Significance. If the claimed stability proofs and benchmark results hold under independent verification, the work could contribute to energy-frugal AI by formalizing iterative refinement as a substitute for model scale. The bio-inspired hormonal mechanism and finite-time termination guarantee are potentially interesting extensions of dynamical systems ideas to reasoning architectures, though the heavy reliance on self-cited prior S-AI frameworks limits claims of broad novelty.

major comments (3)
  1. [Abstract] Abstract: The manuscript asserts a Lyapunov stability proof, entropic contraction theorem, and finite-time termination guarantee for the RRC with Clarifine/Confusionin dynamics, yet supplies no equations, derivations, or proof sketches. Without these, the central claim that the hormonal closed loop produces genuine stable convergence cannot be evaluated.
  2. [Abstract] Abstract (experimental validation paragraph): Claims of competitive reasoning performance on SAI-UT+ with <10M parameters and confirmation of temporal parsimony are stated without any data tables, error bars, baseline comparisons, exclusion criteria, or statistical details. This absence makes it impossible to assess whether the results support the substitution of depth for width or merely reflect benchmark-specific behavior.
  3. [Abstract] Abstract: The framework is defined in terms of prior self-cited S-AI works [1-3] and introduces Clarifine/Confusionin to regulate the iteration whose stability is then proved, creating a closed conceptual loop. No external benchmark or independent derivation is shown to break this circularity.
minor comments (1)
  1. [Abstract] The abstract mentions 'primal-dual agent selection under iteration budget' and 'recursive engram memory with warm-start acceleration' without defining the terms or their relation to the RRC dynamics.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their detailed review and constructive feedback on our manuscript. We address each major comment point by point below, clarifying that the abstract serves as a high-level summary while the full derivations, proofs, and experimental details are provided in the body of the paper.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The manuscript asserts a Lyapunov stability proof, entropic contraction theorem, and finite-time termination guarantee for the RRC with Clarifine/Confusionin dynamics, yet supplies no equations, derivations, or proof sketches. Without these, the central claim that the hormonal closed loop produces genuine stable convergence cannot be evaluated.

    Authors: The abstract provides a concise overview of the contributions as is conventional due to length constraints. The complete mathematical framework—including recursive state dynamics, the Lyapunov stability proof, the entropic contraction theorem, the hormonal stopping criterion, and the finite-time termination guarantee—is fully developed with equations, derivations, and proof sketches in Sections 3 and 4 of the manuscript. revision: no

  2. Referee: [Abstract] Abstract (experimental validation paragraph): Claims of competitive reasoning performance on SAI-UT+ with <10M parameters and confirmation of temporal parsimony are stated without any data tables, error bars, baseline comparisons, exclusion criteria, or statistical details. This absence makes it impossible to assess whether the results support the substitution of depth for width or merely reflect benchmark-specific behavior.

    Authors: The abstract summarizes the key experimental findings. Detailed results, including performance tables, baseline comparisons, error bars, exclusion criteria, and statistical analysis supporting competitive performance with under 10 million parameters and the temporal parsimony principle, are presented in the Experiments section of the full manuscript. revision: no

  3. Referee: [Abstract] Abstract: The framework is defined in terms of prior self-cited S-AI works [1-3] and introduces Clarifine/Confusionin to regulate the iteration whose stability is then proved, creating a closed conceptual loop. No external benchmark or independent derivation is shown to break this circularity.

    Authors: We build upon the foundational S-AI framework, hormonal-probabilistic unification, and mathematical methodology from our prior works [1-3], which is standard practice for incremental research. However, this manuscript introduces independent novel contributions: the Recursive Reasoning Cycle (RRC) operationalized via the specific Clarifine and Confusionin hormones, the full stability analysis with Lyapunov proof and entropic contraction theorem, the finite-time termination guarantee, Euler-Maruyama discretization, primal-dual selection, recursive engram memory, and the SAI-UT+ experimental validation. These elements extend the prior foundations with new derivations and results. revision: no

Circularity Check

1 steps flagged

RRC framework and Lyapunov claims reduce to self-cited prior S-AI works plus definitional hormones

specific steps
  1. self citation load bearing [Abstract]
    "Building upon the S-AI foundational framework [1], the hormonal-probabilistic unification doctrine [2], and the formal mathematical methodology established in S-AI-IoT [3], the present work formalizes the Recursive Reasoning Cycle (RRC) as a dynamical system governed by two novel hormones: Clarifine, a convergence signal, and Confusionin, an uncertainty detector, whose antagonistic regulation drives iterative state refinement toward a stable cognitive equilibrium."

    The central RRC dynamical system, its stability proof, and the substitution of depth for width are justified solely by citation to the author's own prior works [1-3]; the new hormones are defined to generate exactly the convergence whose Lyapunov stability is then claimed, reducing the derivation to the inputs of the self-citation chain.

full rationale

The abstract explicitly grounds the entire Recursive Reasoning Cycle, hormonal dynamics, Lyapunov proof, and temporal parsimony principle in three prior self-authored S-AI papers. Clarifine and Confusionin are introduced precisely to produce the iterative refinement and stable equilibrium whose convergence is then asserted, with no independent external benchmark or non-self derivation exhibited for the core dynamical system. This satisfies the self-citation load-bearing pattern and borders on self-definitional construction of the claimed result.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 2 invented entities

The central claim rests on newly postulated hormonal signals and stability theorems that receive no independent evidence or external grounding in the abstract; the architecture adds conceptual entities without deriving them from prior literature or data.

free parameters (1)
  • iteration budget
    Chosen to trade off performance against compute; no value or fitting procedure stated.
axioms (1)
  • domain assumption The recursive state dynamics governed by Clarifine and Confusionin admit a Lyapunov function that guarantees stability.
    Invoked to support the finite-time termination guarantee.
invented entities (2)
  • Clarifine no independent evidence
    purpose: Convergence signal that drives state refinement
    Newly introduced hormone with no external validation or derivation from prior work.
  • Confusionin no independent evidence
    purpose: Uncertainty detector that antagonizes Clarifine
    Newly introduced hormone with no external validation or derivation from prior work.

pith-pipeline@v0.9.0 · 5509 in / 1545 out tokens · 54015 ms · 2026-05-15T06:38:32.006566+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

49 extracted references · 49 canonical work pages · 3 internal anchors

  1. [1]

    S-AI: A sparse artificial intelligence system orchestrated by a hormonal MetaAgent and context-aware specialized agents,

    S. Slaoui, "S-AI: A sparse artificial intelligence system orchestrated by a hormonal MetaAgent and context-aware specialized agents," International Journal for Multidisciplinary Research (IJFMR), vol. 1, no. 2, 2025

    work page 2025

  2. [2]

    From Hormones to Probabilities: A Unified Doctrine of Cognitive Homeostasis in Sparse Artificial Intelligence,

    S. Slaoui, "From Hormones to Probabilities: A Unified Doctrine of Cognitive Homeostasis in Sparse Artificial Intelligence," SSRN Preprint, Nov. 2025, doi: 10.2139/ssrn.5735582

    work page doi:10.2139/ssrn.5735582 2025

  3. [3]

    S-AI-IoT: Formal Agent Specification, Mathematical Modeling, and Stability Analysis of the Hormonal Orchestration Framework,

    S. Slaoui, "S-AI-IoT: Formal Agent Specification, Mathematical Modeling, and Stability Analysis of the Hormonal Orchestration Framework," IJAIA, forthcoming

    work page

  4. [4]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, "Attention is all you need," in Advances in Neural Information Processing Systems (NeurIPS), vol. 30, 2017

    work page 2017

  5. [5]

    Language models are few- shot learners,

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal et al., "Language models are few- shot learners," NeurIPS, vol. 33, pp. 1877–1901, 2020

    work page 1901

  6. [6]

    Chain-of- thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, B. Ichter, F. Xia, E. Chi, Q. Le, and D. Zhou, "Chain-of- thought prompting elicits reasoning in large language models," NeurIPS, vol. 35, 2022. 49

    work page 2022

  7. [7]

    Tree of thoughts: Deliberate problem solving with large language models,

    S. Yao, D. Yu, J. Zhao, I. Shafran, T. Griffiths, Y. Cao, and K. Narasimhan, "Tree of thoughts: Deliberate problem solving with large language models," NeurIPS, vol. 36, 2023

    work page 2023

  8. [8]

    Reflexion: Language agents with verbal reinforcement learning,

    N. Shinn, F. Cassano, A. Gopinath, K. Narasimhan, and S. Yao, "Reflexion: Language agents with verbal reinforcement learning," NeurIPS, vol. 36, 2023

    work page 2023

  9. [9]

    Energy and policy considerations for deep learning in NLP,

    E. Strubell, A. Ganesh, and A. McCallum, "Energy and policy considerations for deep learning in NLP," in Proc. 57th Annual Meeting of the Association for Computational Linguistics (ACL), pp. 3645– 3650, 2019

    work page 2019

  10. [10]

    Kahneman, Thinking, Fast and Slow

    D. Kahneman, Thinking, Fast and Slow. New York: Farrar, Straus and Giroux, 2011

    work page 2011

  11. [11]

    Stress, adaptation, and disease: Allostasis and allostatic load,

    B. S. McEwen, "Stress, adaptation, and disease: Allostasis and allostatic load," Annals of the New York Academy of Sciences, vol. 840, pp. 33–44, 1998

    work page 1998

  12. [12]

    Less is More: Recursive Reasoning with Tiny Networks

    A. Jolicoeur-Martineau, "Less is more: Recursive reasoning with tiny networks," arXiv preprint arXiv:2510.04871, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  13. [13]

    Bio-inspired architecture for parsimonious conversational intelligence: The S-AI-GPT framework,

    S. Slaoui, "Bio-inspired architecture for parsimonious conversational intelligence: The S-AI-GPT framework," International Journal of Artificial Intelligence & Applications (IJAIA), vol. 16, no. 4, 2025, doi: 10.5121/ijaia.2025.16403

    work page doi:10.5121/ijaia.2025.16403 2025

  14. [14]

    S-AI-NET: A sparse AI framework for adaptive and parsimonious autonomous networking,

    S. Slaoui, "S-AI-NET: A sparse AI framework for adaptive and parsimonious autonomous networking," Research Square, 2025, doi: 10.21203/rs.3.rs-4740886/v1

    work page doi:10.21203/rs.3.rs-4740886/v1 2025

  15. [15]

    S-AI-Cyber: A bio-inspired hormonal architecture for real-time cyber-defense,

    S. Slaoui, "S-AI-Cyber: A bio-inspired hormonal architecture for real-time cyber-defense," Research Square, 2025, doi: 10.21203/rs.3.rs-4840888/v1

    work page doi:10.21203/rs.3.rs-4840888/v1 2025

  16. [16]

    Long short-term memory,

    S. Hochreiter and J. Schmidhuber, "Long short-term memory," Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997

    work page 1997

  17. [17]

    Learning phrase representations using RNN encoder–decoder for statistical machine translation,

    K. Cho, B. van Merrienboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y. Bengio, "Learning phrase representations using RNN encoder–decoder for statistical machine translation," in Proc. EMNLP, pp. 1724–1734, 2014

    work page 2014

  18. [18]

    Deep equilibrium models,

    S.-J. Bai, J. Z. Kolter, and V. Koltun, "Deep equilibrium models," in Advances in Neural Information Processing Systems (NeurIPS), vol. 32, 2019

    work page 2019

  19. [19]

    Multiscale deep equilibrium models,

    S.-J. Bai, J. Z. Kolter, and V. Koltun, "Multiscale deep equilibrium models," in Advances in Neural Information Processing Systems (NeurIPS), vol. 33, 2020

    work page 2020

  20. [20]

    Universal Transformers,

    M. Dehghani, S. Gouws, O. Vinyals, J. Uszkoreit, and Ł. Kaiser, "Universal Transformers," in Proc. International Conference on Learning Representations (ICLR), 2019

    work page 2019

  21. [21]

    Neural Turing Machines

    A. Graves, G. Wayne, and I. Danihelka, "Neural Turing machines," arXiv preprint arXiv:1410.5401, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  22. [22]

    S-AI-Anti-Hallucination: A bio-inspired and confidence-aware sparse AI framework for reliable generative systems,

    S. Slaoui, "S-AI-Anti-Hallucination: A bio-inspired and confidence-aware sparse AI framework for reliable generative systems," IJAIA, vol. 16, no. 6, Nov. 2025, doi: 10.5121/ijaia.2025.16601. 50

    work page doi:10.5121/ijaia.2025.16601 2025

  23. [23]

    Triadic S-AI-Antihallucination: Hormonal clarification and metacognitive stabilization in generative reasoning,

    S. Slaoui, "Triadic S-AI-Antihallucination: Hormonal clarification and metacognitive stabilization in generative reasoning," IJAIA, vol. 17, no. 1, Jan. 2026, doi: 10.5121/ijaia.2026.17103

    work page doi:10.5121/ijaia.2026.17103 2026

  24. [24]

    A logical calculus of the ideas immanent in nervous activity,

    W. S. McCulloch and W. Pitts, "A logical calculus of the ideas immanent in nervous activity," Bulletin of Mathematical Biophysics, vol. 5, pp. 115–133, 1943

    work page 1943

  25. [25]

    Neural networks and physical systems with emergent collective computational abilities,

    J. J. Hopfield, "Neural networks and physical systems with emergent collective computational abilities," Proceedings of the National Academy of Sciences, vol. 79, no. 8, pp. 2554–2558, 1982

    work page 1982

  26. [26]

    R. C. O'Reilly and Y. Munakata, Computational Explorations in Cognitive Neuroscience. Cambridge, MA: MIT Press, 2000

    work page 2000

  27. [27]

    Dayan and L

    P. Dayan and L. F. Abbott, Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems. Cambridge, MA: MIT Press, 2001

    work page 2001

  28. [28]

    Bio-inspired hormonal modulation and adaptive orchestration in S-AI-GPT,

    S. Slaoui, "Bio-inspired hormonal modulation and adaptive orchestration in S-AI-GPT," IJAIA, vol. 16, no. 4, 2025, doi: 10.5121/ijaia.2025.16404

    work page doi:10.5121/ijaia.2025.16404 2025

  29. [29]

    Memory architecture in S-AI-GPT: From contextual adaptation to hormonal modulation,

    S. Slaoui, "Memory architecture in S-AI-GPT: From contextual adaptation to hormonal modulation," IJAIA, vol. 16, no. 5, 2025, doi: 10.5121/ijaia.2025.16503

    work page doi:10.5121/ijaia.2025.16503 2025

  30. [30]

    S-AI-EDU: A bio-inspired and modular sparse AI architecture for adaptive and symbolic intelligent educational systems,

    S. Slaoui, "S-AI-EDU: A bio-inspired and modular sparse AI architecture for adaptive and symbolic intelligent educational systems," IJAIA, vol. 17, no. 1, Jan. 2026, doi: 10.5121/ijaia.2026.17102

    work page doi:10.5121/ijaia.2026.17102 2026

  31. [31]

    S. Slaoui, "S-AI-ROBOTICS: A sparse artificial intelligence architecture with hormonal orchestration, parsimonious control, and symbolic memory for adaptive, safe, and explainable embodied robotics," IJAIA, vol. 17, no. 2, March 2026, doi: 10.5121/ijaia.2026.17202

    work page doi:10.5121/ijaia.2026.17202 2026

  32. [32]

    S-AI-DEF: A bio-inspired and parsimonious cognitive architecture for adaptive artificial intelligence in defense systems,

    S. Slaoui, "S-AI-DEF: A bio-inspired and parsimonious cognitive architecture for adaptive artificial intelligence in defense systems," IJAIA, vol. 17, no. 2, March 2026, doi: 10.5121/ijaia.2026.17204

    work page doi:10.5121/ijaia.2026.17204 2026

  33. [33]

    Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity,

    W. Fedus, B. Zoph, and N. Shazeer, "Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity," Journal of Machine Learning Research (JMLR), vol. 23, no. 120, pp. 1–39, 2022

    work page 2022

  34. [34]

    MASAI: Modular architecture for software-engineering AI agents,

    D. Arora, A. Sonwane, N. Wadhwa, A. Mehrotra, S. Utpala, R. Bairi, A. Kanade, and N. Natarajan, "MASAI: Modular architecture for software-engineering AI agents," arXiv preprint arXiv:2406.11638, 2024

    work page arXiv 2024

  35. [35]

    SMoA: Improving multi-agent large language models with sparse mixture-of-agents,

    D. Li, Z. Tan, P. Qian, Y. Li, K. S. Chaudhary, L. Hu, and J. Shen, "SMoA: Improving multi-agent large language models with sparse mixture-of-agents," arXiv preprint arXiv:2411.03284, 2024

    work page arXiv 2024

  36. [36]

    The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink,

    D. Patterson, J. Gonzalez, U. Hölzle, Q. Le, C. Liang, L.-L. Munguia, D. Rothchild, D. So, M. Texier, and J. Dean, "The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink," Computer, vol. 55, no. 7, pp. 18–28, 2022, arXiv preprint arXiv:2204.05149

    work page arXiv 2022

  37. [37]

    H. K. Khalil, Nonlinear Systems, 3rd ed. Upper Saddle River, NJ: Prentice Hall, 2002. 51

    work page 2002

  38. [38]

    Finite-time stability of continuous autonomous systems,

    S. P. Bhat and D. S. Bernstein, "Finite-time stability of continuous autonomous systems," SIAM Journal on Control and Optimization, vol. 38, no. 3, pp. 751–766, 2000

    work page 2000

  39. [39]

    V. S. Borkar, Stochastic Approximation: A Dynamical Systems Viewpoint, 2nd ed. Cambridge: Cambridge University Press, 2022

    work page 2022

  40. [40]

    A mathematical theory of communication,

    C. E. Shannon, "A mathematical theory of communication," Bell System Technical Journal, vol. 27, pp. 379–423, 1948

    work page 1948

  41. [41]

    T. M. Cover and J. A. Thomas, Elements of Information Theory, 2nd ed. Hoboken, NJ: Wiley, 2006

    work page 2006

  42. [42]

    Über Abbildungen von Mannigfaltigkeiten,

    L. E. J. Brouwer, "Über Abbildungen von Mannigfaltigkeiten," Mathematische Annalen, vol. 71, pp. 97–115, 1911

    work page 1911

  43. [43]

    On contraction analysis for non-linear systems,

    W. Lohmiller and J.-J. E. Slotine, "On contraction analysis for non-linear systems," Automatica, vol. 34, no. 6, pp. 683–696, 1998

    work page 1998

  44. [44]

    Wooldridge, An Introduction to MultiAgent Systems, 2nd ed

    M. Wooldridge, An Introduction to MultiAgent Systems, 2nd ed. Chichester: Wiley, 2009

    work page 2009

  45. [45]

    Russell and P

    S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed. Hoboken, NJ: Pearson, 2021

    work page 2021

  46. [46]

    Mamba: Linear-Time Sequence Modeling with Selective State Spaces

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

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  47. [47]

    RWKV: Reinventing RNNs for the transformer era,

    B. Peng, E. Alcaide, Q. Anthony, A. Albalak, S. Arcadinho, H. Cao et al., “RWKV: Reinventing RNNs for the transformer era,” in Proc. EMNLP (Findings), pp. 14048–14077, 2023

    work page 2023

  48. [48]

    Beyond the imitation game: Quantifying and extrapolating the capabilities of language models,

    A. Srivastava et al., “Beyond the imitation game: Quantifying and extrapolating the capabilities of language models,” Transactions on Machine Learning Research, 2023

    work page 2023

  49. [49]

    Measuring mathematical problem solving with the MATH dataset,

    D. Hendrycks, C. Burns, S. Basart, A. Zou, M. Mazeika, D. Song, and J. Steinhardt, “Measuring mathematical problem solving with the MATH dataset,” in Proc. NeurIPS Datasets and Benchmarks Track, 2021

    work page 2021