Pith. sign in

REVIEW 3 major objections 5 minor 31 references

StateBridge: Training-free Hidden-state Alignment for Latent Communication in LLM Multi-Agent Systems

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Training-free latent communication lets LLM agents beat text baselines on 22 of 26 tasks.

desk verdict A clean, well-ablated method for training-free latent communication whose load-bearing claim about transmitting more than the suffix is backed by one case study, not a control. read the letter →

arxiv 2608.13317 v1 pith:4CVHH6S6 submitted 2026-08-13 cs.AI

Yanwen Peng , Delvin Ce Zhang , Xi Wang , Nikolaos Aletras This is my paper · ORCID
classification cs.AI
keywords latentcommunicationhidden-statealignmentorthogonalProcrustesmulti-agentLLMsystemstraining-freecontinuousprefixembedding-spacecompatibilityKV-cachetransfer
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 paper claims that off-the-shelf LLM agents can communicate through continuous hidden states instead of text, without any training or architectural changes. The key is a closed-form alignment that rotates a sender's final-layer hidden states into the receiver's input embedding space, then calibrates norms and anchors the vectors near real vocabulary embeddings. On math, code, and QA benchmarks across four models, this beats text-based and KV-cache baselines on most tasks, suggesting that the main barrier to latent communication is geometric compatibility, not model training.

What carries the argument

The central object is the closed-form alignment interface: a three-step transformation that maps a sender's final-layer hidden states S to an aligned prefix that the receiver can read. It starts with centering and whitening both S and the reference token embeddings R, then solves an orthogonal Procrustes problem Q* = argmin_Q ||S_w Q - R_w||$_F^{2}$ with Q^T Q = I, whose solution is the SVD product UV^T. This orthogonal rotation preserves pairwise distances and angles in the whitened space. Two compatibility steps follow: norm calibration rescales each vector to the average vocabulary-embedding norm, and vocabulary anchoring moves each vector a fraction $\alpha$ toward its nearest vocabulary embedding. The paper proves that replacing the orthogonal map with ridge regression confines the prefix to the span of the sampled token embeddings, destroying the continuous information that makes latent communication worthwhile.

What would settle it

Take a task where the sender's message is longer than K tokens and the last K tokens are deliberately uninformative (e.g., a repeated filler word). If StateBridge's receiver still recovers the earlier content, the prefix carries information beyond the suffix. If it cannot, the claim that the final K states encode the whole message fails. Also, run the same setup with an isotropic random rotation instead of the Procrustes rotation; if performance stays high, the isometry and alignment-specific SVD is not doing the work.

Watch

Extended reading notes

Core claim

The central claim is that a closed-form orthogonal alignment alone resolves the representation mismatch between final-layer hidden states and input embeddings, enabling training-free and effective latent communication between off-the-shelf LLM agents. The method, StateBridge, extracts the final K=64 hidden states of a sender's generated message, aligns them to the corresponding token embeddings via centering, whitening, and orthogonal Procrustes rotation, then adjusts norms and blends slightly toward vocabulary embeddings. The aligned prefix is prepended to the receiver's prompt embeddings. On 26 model–task pairs spanning Qwen3-4B/8B/32B and OLMo3-7B, StateBridge achieves the best or tied-best score on 22 pairs, improving average accuracy by 2.4–2.9 points over the strongest baseline. Ablations show that the orthogonal constraint is what preserves the geometry that carries information beyond the sampled tokens.

Load-bearing premise

The final K hidden states of a generated message, after a single global rotation to the embedding space, encode enough of the entire message that the receiver can act on it. If those K states summarize only the last K tokens, the method's advantage over text disappears.

Editorial extensions

If this is right

  • If the claim holds, any homogeneous multi-agent system can upgrade its inter-agent channel to a training-free continuous prefix, with no model updates and no learned projectors.
  • Latent communication becomes portable across model families: because StateBridge touches only the input embedding layer, it transfers to architectures where KV-cache injection degrades (e.g., OLMo3-7B).
  • The geometric-preservation result implies that the information channel between agents is no longer bounded by the discrete token capacity K*log2(V) bits; continuous states can carry confidence and alternative-reasoning traces.
  • The method's gains concentrate on harder benchmarks (GPQA, AIME24/25, code generation), suggesting that text handoffs are most lossy precisely where nuanced intermediate reasoning matters.
  • The ablation results indicate that simply injecting raw hidden states or arbitrary continuous vectors is not enough; the prefix must be both informative and compatible, which localizes the design principle for future latent-communication interfaces.

Reading between the lines

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

  • One testable extension the paper leaves implicit: if the alignment preserves geometry, then the same closed-form interface could align hidden states between different models of the same family or even across families, since it only needs a shared embedding dimension and a reference matrix R.
  • The theory suggests that ridge-regression alignment should be strictly worse in any task where the receiver must reconstruct hidden reasoning structure; a direct experiment varying the whitening strength or using anisotropic scaling could further test the isometry hypothesis.
  • The norm-calibration step implies that attention-dot-product scale is a dominant failure mode for raw hidden-state transfer; a cheap diagnostic would compute attention entropy of the receiver on aligned vs. unaligned prefixes.
  • One could integrate StateBridge with learned communication by using the orthogonal alignment as a warm-start, potentially reducing projector training data while keeping portability.
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 / 5 minor

Summary. The paper introduces StateBridge, a training-free protocol for latent communication between homogeneous LLM agents. The sender's last K final-layer hidden states are centered, whitened, and rotated to match the embeddings of the last K decoded tokens via an orthogonal Procrustes solution; the result is norm-calibrated, partially anchored to nearest vocabulary embeddings, and prepended as a continuous prefix to the receiver's prompt. The authors evaluate the method on math, QA, and code benchmarks with Qwen3-4B/8B/32B and OLMo3-7B-Think, reporting best or tied-best on 22 of 26 model-task pairs, with ablations over components and a qualitative case study. The central claim is that closed-form alignment alone resolves the hidden-state/embedding mismatch and permits effective training-free latent communication.

Significance. Training-free, portable latent communication is a valuable goal, and StateBridge's per-message closed-form alignment is a clean idea that avoids trained projectors and layer-wise KV injection. The paper's strengths include a formal derivation of the Procrustes solution, a span-confinement result for ridge regression, a careful ablation isolating the rotation, norm calibration, and vocabulary anchoring, and a random-noise control that rules out a trivial prefix effect. If the empirical claims survive re-evaluation, the method would strengthen the case that continuous inter-agent channels can be built from off-the-shelf models at negligible cost. The significance is therefore substantial, but the current evidence does not yet nail the mechanism: the 'beyond the suffix tokens' claim rests on a single uncontrolled example, and all head-to-head results are single runs.

major comments (3)
  1. [Section 5, Table 1] All evaluations in Table 1 are single runs with stochastic decoding; no standard deviations, seeds, or significance tests are reported. Many decisive comparisons are small (e.g., Qwen3-32B ARC-C is a 1.2-point loss and GSM8K losses are 2.5-2.6 points; Qwen3-4B ARC-C gain is 1.4 points). With run-to-run variance at this scale, the headline '22 of 26' and the conclusion that StateBridge 'consistently outperforms' the strongest baseline are not yet supported. Please report repeated runs with variance estimates or statistical tests for the main comparisons.
  2. [Section 3.2, Table 4] The paper's load-bearing assumption is that the final K hidden states encode the full message, not just the last K tokens. The only direct evidence is the case study in Table 4, which is a single MedQA example; the critic receives the question as text in addition to the prefix, so part of the 'recovery' could be inference from the question rather than from the prefix. The paper itself notes in Section 6.4 that at K=128 'the aligned prefix no longer carries continuous information beyond the suffix tokens.' A control that transmits only the last-K-token text, or a control that directly uses the suffix embeddings R as the prefix, would separate information recovered from earlier context from a generic soft-prefix benefit. Without such a control, the improvement over TextMAS could be due to the interface rather than to recovering content beyond the suffix.
  3. [Section 3.3, Appendix A.3] Proposition A.3 shows only that the rotation step preserves the Gram matrix of the whitened states; the final injected prefix also undergoes dewhitening, norm calibration, and vocabulary anchoring (Eqs. 7-9), none of which is an isometry. Main-text statements such as 'Orthogonal alignment preserves these semantic similarity' (Section 6.1) therefore overstate the guarantee used to explain the code-generation gains. Please state in the main text that the isometry holds only for the rotation in whitened coordinates, and either adjust the explanatory claims or provide an additional result showing the later steps do not materially alter the relevant geometry.
minor comments (5)
  1. [Appendix B.3] Please report the number of independent runs and random seeds used for all main tables; this is needed to assess the stability of the reported accuracies and pass@1 scores.
  2. [Table 2] The 'Random Noise' row should specify how the noise vectors are generated and whether they are norm-calibrated and vocabulary-anchored; if not, the comparison conflates information content with scale and compatibility.
  3. [Table 4] Please state the decoding temperature and sampling protocol for the critic's recovery in the case study, since this is the only direct evidence for the information-preservation mechanism.
  4. [Section 4] The additional benchmarks AIME24, AIME25, and GPQA are reported only for Qwen3-8B and Qwen3-32B; consider reporting OLMo3-7B-Think on these tasks to strengthen the cross-family generalization claim.
  5. [Abstract and Section 3.3] Calling the whole interface a 'closed-form orthogonal transformation' is imprecise: the full pipeline includes whitening, norm calibration, and vocabulary anchoring, and only the Procrustes step is an orthogonal map.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: alignment target is the message's own token embeddings, and evaluation is external to the derivation.

full rationale

The derivation chain is self-contained. The alignment target R in Eq. (2) is the embedding of the sampled tokens of the same generated message that produced the hidden states S; it is not the evaluation label or any downstream task quantity. The Procrustes solution Q* is a closed-form SVD (Eq. 6) with no learnable parameters, and the norm calibration and vocabulary anchoring steps are fixed deterministic rescaling operations. The paper's central claim, that the aligned prefix enables effective latent communication, is tested against external benchmarks (GSM8K, AIME, GPQA, MedQA, ARC-C, MBPP+, HumanEval+) and against baseline communication channels (text and KV-cache transfer), so the evaluation is not forced by the construction. The only near-selection effect is that K=64 and alpha=0.3 are chosen on MedQA with Qwen3-4B before being applied elsewhere, but this is disclosed hyperparameter tuning rather than a fitted parameter renamed as a prediction. Likewise, the self-citations (Cao et al. 2025, Fang et al. 2025) merely support general background statements and do not carry the load of any uniqueness or derivation claim. The anecdotal case study for the 'beyond suffix tokens' claim is weak evidence, but that is an empirical-evidence concern, not circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or abstract entities; its load-bearing content is the closed-form alignment interface. The free parameters are three hyperparameters, of which K and alpha are tuned on one dataset. The main assumptions are about LLM representation geometry and message summarization by final hidden states.

free parameters (3)
  • K = 64
    Prefix length (number of final hidden states). Selected as the best value on MedQA with Qwen3-4B in Section 6.2 and fixed for all other tasks and models.
  • alpha = 0.3
    Vocabulary anchoring coefficient. Selected on MedQA with Qwen3-4B in Section 6.2; sensitivity analysis shows moderate values work across tasks.
  • lambda = 1e-3
    Regularization added to the whitening covariances in Equation (4); fixed by hand.
assumptions (4)
  • domain assumption Pairwise geometric proximity in hidden-state space encodes semantic similarity.
    Motivates the choice of orthogonal Procrustes over ridge regression in Section 3.3, citing Ethayarajh (2019). If false, the geometric-preservation argument loses its force.
  • domain assumption All agents share the same pretrained LLM and hidden dimensionality.
    Problem statement in Section 3.1 explicitly assumes a homogeneous MA system; the method does not address heterogeneous models.
  • domain assumption The last K final-layer hidden states summarize the full generated message.
    Message extraction keeps only the final K states in Section 3.2; the case study in Section 6.4 is the only direct evidence, and it is qualitative.
  • standard math Whitened covariance matrices are invertible; lambda regularization makes them so.
    Equation (4) adds lambda I to the covariances, assuming invertibility.

how reviews work

0 comments
Cite this review

Pith. "Pith review of StateBridge: Training-free Hidden-state Alignment for Latent Communication in LLM Multi-Agent Systems." pith.science (2026). https://pith.science/paper/4CVHH6S6

@misc{pith2026260813317,
  author       = {Pith},
  title        = {Pith review of: StateBridge: Training-free Hidden-state Alignment for Latent Communication in LLM Multi-Agent Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4CVHH6S6}},
  note         = {Machine review of arXiv:2608.13317}
}
read the original abstract

Large language model based multi-agent systems usually communicate in text, i.e., using discrete tokens. However, text introduces a discrete bottleneck. Converting the sender's continuous hidden states into discrete tokens discards information that token identities alone cannot capture. Recent work proposes latent communication as an alternative, where agents transmit hidden representations directly without converting them to text. However, existing latent methods either inject working memory layer by layer across the transformers, or require trained projectors that limit portability. We propose StateBridge, a training-free latent communication approach that aligns the sender's final-layer hidden states to the receiver's input space via a closed-form orthogonal transformation. Lightweight norm calibration and vocabulary anchoring ensure compatibility with the pretrained input distribution. The aligned states are prepended to the input of the receiver agent as a continuous prefix. We evaluate StateBridge on math reasoning, code generation, and question answering with four models from two families. StateBridge achieves the best or tied-best score on 22 out of 26 model-task pairs, consistently outperforming the strongest baseline.

Figures

Figures reproduced from arXiv: 2608.13317 by the authors.

Figure 1
Figure 1. Overview of StateBridge. Standard text communication (a) transforms the sender’s message to discrete token identities via sampling, which causes information loss, whereas StateBridge (b) extracts message hidden states from the sender, aligns them with the input embedding space, and prepends the resulting prefix to the receiver’s prompt. across all transformer layers. However, they transfer internal processing states… view at source ↗
Figure 2
Figure 2. The StateBridge alignment interface transforms message hidden states S into the aligned prefix S¯ for receiver using reference embeddings R. The message hidden states S and the reference em￾beddings R describe the same generated message but occupy different regions of the representation space. The transformation must bridge this gap (Cao et al., 2025). Since geometric proximity in hidden-state space encodes semantic… view at source ↗
Figure 3
Figure 3. Principal component analysis (PCA) density (left) and PCA scatter of the embed [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 14 canonical work pages

  1. [1]

    Progressive depth up-scaling via optimal transport.arXiv preprint arXiv:2508.08011,

    Mingzi Cao, Xi Wang, and Nikolaos Aletras. Progressive depth up-scaling via optimal transport.arXiv preprint arXiv:2508.08011,

    arXiv

  2. [3]

    Evalu- ating large language models trained on code.arXiv preprint arXiv:2107.03374,

    Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, et al. Evalu- ating large language models trained on code.arXiv preprint arXiv:2107.03374,

    arXiv

  3. [5]

    Think you have solved question answering? try arc, the ai2 reasoning challenge.arXiv preprint arXiv:1803.05457,

    Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick, and Oyvind Tafjord. Think you have solved question answering? try arc, the ai2 reasoning challenge.arXiv preprint arXiv:1803.05457,

    arXiv

  4. [9]

    Magentic-one: A generalist multi- agent system for solving complex tasks.arXiv preprint arXiv:2411.04468,

    Adam Fourney, Gagan Bansal, Hussein Mozannar, et al. Magentic-one: A generalist multi- agent system for solving complex tasks.arXiv preprint arXiv:2411.04468,

    arXiv

  5. [11]

    10 Published as a conference paper at COLM 2026 Guohao Li, Hasan Abed Al Kader Hammoud, Hani Itani, Dmitrii Khizbullin, and Bernard Ghanem

    doi: 10.3390/app11146421. 10 Published as a conference paper at COLM 2026 Guohao Li, Hasan Abed Al Kader Hammoud, Hani Itani, Dmitrii Khizbullin, and Bernard Ghanem. Camel: communicative agents for mind exploration of large language model society. InProceedings of the 37th International Conference on Neural Information Processing Systems, NIPS ’23,

    doi:10.3390/app11146421 2026

  6. [12]

    acl-long.353

    doi: 10.18653/v1/2021. acl-long.353. Tian Liang, Zhiwei He, Wenxiang Jiao, Xing Wang, Yan Wang, Rui Wang, Yujiu Yang, Shuming Shi, and Zhaopeng Tu. Encouraging divergent thinking in large language models through multi-agent debate. InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pp. 17889–17904,

    doi:10.18653/v1/2021 2021

  7. [16]

    Wei Tao, Yucheng Zhou, Yanlin Wang, Wenqiang Zhang, Hongyu Zhang, and Yu Cheng

    doi: 10.18653/v1/2024.emnlp-industry.91. Wei Tao, Yucheng Zhou, Yanlin Wang, Wenqiang Zhang, Hongyu Zhang, and Yu Cheng. Magis: Llm-based multi-agent framework for github issue resolution.Advances in Neural Information Processing Systems, 37:51963–51993,

    doi:10.18653/v1/2024.emnlp-industry.91 2024

  8. [17]

    Olmo 3.arXiv preprint arXiv:2512.13961,

    Team OLMo, Allyson Ettinger, Amanda Bertsch, Bailey Kuehl, David Graham, et al. Olmo 3.arXiv preprint arXiv:2512.13961,

    arXiv

Show all 31 references
  1. [18]

    Talk structurally, act hierarchically: A collaborative framework for llm multi-agent sys- tems.arXiv preprint arXiv:2502.11098,

    Zhao Wang, Sota Moriyama, Wei-Yao Wang, Briti Gangopadhyay, and Shingo Takamatsu. Talk structurally, act hierarchically: A collaborative framework for llm multi-agent sys- tems.arXiv preprint arXiv:2502.11098,

    arXiv

  2. [19]

    Talk to right special- ists: Iterative routing in multi-agent systems for question answering.arXiv preprint arXiv:2501.07813,

    Feijie Wu, Zitao Li, Fei Wei, Yaliang Li, Bolin Ding, and Jing Gao. Talk to right special- ists: Iterative routing in multi-agent systems for question answering.arXiv preprint arXiv:2501.07813,

    arXiv

  3. [20]

    Autogen: Enabling next-gen llm applica- tions via multi-agent conversations

    11 Published as a conference paper at COLM 2026 Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Beibin Li, Erkang Zhu, Li Jiang, Xiaoyun Zhang, Shaokun Zhang, Jiale Liu, et al. Autogen: Enabling next-gen llm applica- tions via multi-agent conversations. InFirst Conference on ...

    2026

  4. [21]

    Beyond self-talk: A communication-centric survey of llm-based multi-agent systems.arXiv preprint arXiv:2502.14321,

    Bingyu Yan, Zhibo Zhou, Litian Zhang, Lian Zhang, Ziyi Zhou, Dezhuang Miao, Zhoujun Li, Chaozhuo Li, and Xiaoming Zhang. Beyond self-talk: A communication-centric survey of llm-based multi-agent systems.arXiv preprint arXiv:2502.14321,

    arXiv

  5. [22]

    Qwen3 technical report.arXiv preprint arXiv:2505.09388,

    An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, et al. Qwen3 technical report.arXiv preprint arXiv:2505.09388,

    arXiv

  6. [23]

    Llm-based multi- agent systems: Techniques and business perspectives.arXiv preprint arXiv:2411.14033,

    Yingxuan Yang, Qiuying Peng, Jun Wang, Ying Wen, and Weinan Zhang. Llm-based multi- agent systems: Techniques and business perspectives.arXiv preprint arXiv:2411.14033,

    arXiv

  7. [24]

    Mobile-agent-v3: Fundamental agents for gui automation.arXiv preprint arXiv:2508.15144,

    Jiabo Ye, Xi Zhang, Haiyang Xu, et al. Mobile-agent-v3: Fundamental agents for gui automation.arXiv preprint arXiv:2508.15144,

    arXiv

  8. [25]

    Large language model-brained gui agents: A survey.Transactions on Machine Learning Research, 2025a

    Chaoyun Zhang, Shilin He, Jiaxu Qian, et al. Large language model-brained gui agents: A survey.Transactions on Machine Learning Research, 2025a. Guibin Zhang, Yanwei Yue, Zhixun Li, Sukwon Yun, Guancheng Wan, Kun Wang, Dawei Cheng, Jeffrey Xu Yu, and Tianlong Chen. Cut the cra...

    2025

  9. [26]

    We first formalize the information bottleneck in text communication, then contrast the two approaches at the representation level

    A Theoretical analysis In this section, we analyze the geometric properties of two alignment strategies used in experiments: orthogonal Procrustes and ridge regression. We first formalize the information bottleneck in text communication, then contrast the two approaches at the...

    2026

  10. [27]

    For intermediate γ, the S-dependent coefficients in Hγ still encode continuous variation within span(R), but directions outside this subspace remain inaccessible

    and silence (γ→∞ ). For intermediate γ, the S-dependent coefficients in Hγ still encode continuous variation within span(R), but directions outside this subspace remain inaccessible. A.3 Geometric preservation under Procrustes alignment In contrast, the orthogonal Procrustes a...

    2026

  11. [28]

    Problems span algebra, geometry, number theory, and combinatorics, and require precise numeric answers

    consists of 30 competition-level problems from the 2024 American Invitational Mathematics Examination. Problems span algebra, geometry, number theory, and combinatorics, and require precise numeric answers. • AIME25(Zhang & Math-AI,

    2024

  12. [29]

    Compared with AIME24, this set includes more multi-phase deriva- tions and intricate combinatorial constructions, offering a complementary test of mathematical reasoning

    provides 30 additional problems from the 2025 AIME exam. Compared with AIME24, this set includes more multi-phase deriva- tions and intricate combinatorial constructions, offering a complementary test of mathematical reasoning. Knowledge-Intensive QA. • GPQA-Diamond(Rein et al.,

    2025

  13. [30]

    The dataset emphasizes conceptual depth and cross-disciplinary reasoning

    is the most difficult split of the GPQA bench- mark, featuring 198 graduate-level multiple-choice questions written by domain experts in physics, biology, and chemistry. The dataset emphasizes conceptual depth and cross-disciplinary reasoning. 14 Published as a conference pape...

    2026

  14. [31]

    text format

    augments HumanEval with denser, more challenging test suites, increasing the rigor of functional correctness evaluation. The benchmark emphasizes generalization beyond prompt examples and tests a model’s ability to produce semantically precise, executable Python code. B.2 Agen...

    2026

  15. [1966]

    Zhi Rui Tam, Cheng-Kuang Wu, Yi-Lin Tsai, Chieh-Yen Lin, Hung-yi Lee, and Yun-Nung Chen

    doi: 10.1007/BF02289451. Zhi Rui Tam, Cheng-Kuang Wu, Yi-Lin Tsai, Chieh-Yen Lin, Hung-yi Lee, and Yun-Nung Chen. Let me speak freely? a study on the impact of format restrictions on large lan- guage model performance. InProceedings of the 2024 Conference on Empirical Methods ...

    doi

  16. [2013]

    Chawla, Olaf Wiest, and Xiangliang Zhang

    Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V . Chawla, Olaf Wiest, and Xiangliang Zhang. Large language model based multi-agents: A survey of progress and challenges. InProceedings of the Thirty-Third International Joint Conference on Artificial In...

    2024

  17. [2018]

    Training verifiers to solve math word problems.arXiv preprint arXiv:2110.14168,

    Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. Training verifiers to solve math word problems.arXiv preprint arXiv:2110.14168,

    arXiv

  18. [2019]

    Jinyuan Fang, Yanwen Peng, Xi Zhang, Yingxu Wang, Xinhao Yi, Guibin Zhang, Yi Xu, Bin Wu, Siwei Liu, Zihao Li, Zhaochun Ren, Nikos Aletras, Xi Wang, Han Zhou, and Zaiqiao Meng

    doi: 10.18653/v1/D19-1006. Jinyuan Fang, Yanwen Peng, Xi Zhang, Yingxu Wang, Xinhao Yi, Guibin Zhang, Yi Xu, Bin Wu, Siwei Liu, Zihao Li, Zhaochun Ren, Nikos Aletras, Xi Wang, Han Zhou, and Zaiqiao Meng. A comprehensive survey of self-evolving ai agents: A new paradigm bridgin...

    arXiv

  19. [2021]

    Optima: Optimizing effectiveness and efficiency for llm-based multi-agent system

    Weize Chen, Jiarui Yuan, Chen Qian, Cheng Yang, Zhiyuan Liu, and Maosong Sun. Optima: Optimizing effectiveness and efficiency for llm-based multi-agent system. InFindings of the Association for Computational Linguistics: ACL 2025, pp. 11534–11557,

    2025

  20. [2023]

    AIME 2024 dataset

    Maxwell-Jia. AIME 2024 dataset. https://huggingface.co/datasets/Maxwell-Jia/AIME 2024,

    2024

  21. [2024]

    Predicting multi-agent specialization via task paralleliz- ability.arXiv preprint arXiv:2503.15703,

    Elizabeth Mieczkowski, Ruaridh Mon-Williams, Neil Bramley, Christopher G Lucas, Natalia Velez, and Thomas L Griffiths. Predicting multi-agent specialization via task paralleliz- ability.arXiv preprint arXiv:2503.15703,

  22. [2025]

    Why do multi-agent llm systems fail? InAdvances in Neural Information Processing Systems (NeurIPS 2025, Datasets and Benchmarks Track),

    Mert Cemri, Melissa Z Pan, Shuyi Yang, Lakshya A Agrawal, Bhavya Chopra, Rishabh Tiwari, Kurt Keutzer, Aditya Parameswaran, Dan Klein, Kannan Ramchandran, et al. Why do multi-agent llm systems fail? InAdvances in Neural Information Processing Systems (NeurIPS 2025, Datasets an...

    2025

  23. [2026]

    How contextual are contextualized word representations? comparing the geometry of BERT, ELMo, and GPT-2 embeddings

    Kawin Ethayarajh. How contextual are contextualized word representations? comparing the geometry of BERT, ELMo, and GPT-2 embeddings. InProceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural ...

    2019

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.