Early to Share, Late to Save: Synchronisation-Driven Communication Gating in Bandwidth-Constrained Cooperative VLN

Arav Gupta; Avinash Gautam; Nivedan Yakolli

Under tight bandwidth, cooperative navigation agents learn to share early and navigate alone later, not when lost.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-10 06:19 UTC pith:K4N2UWSM

load-bearing objection Clean budgeted-comm VLN setup and a real early-sync finding, but SR barely moves and the asymmetric pairing may be doing more work than the paper admits. the 4 major comments →

arxiv 2607.08504 v1 pith:K4N2UWSM submitted 2026-07-09 cs.MA cs.RO

Early to Share, Late to Save: Synchronisation-Driven Communication Gating in Bandwidth-Constrained Cooperative VLN

Arav Gupta , Nivedan Yakolli , Avinash Gautam This is my paper

classification cs.MA cs.RO

keywords Vision-Language NavigationMulti-Agent CommunicationBandwidth-Constrained CommunicationRecurrent Neural NetworksHindsight SupervisionHidden-State AlignmentCooperative VLN

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Most multi-robot vision-language navigation work assumes agents can talk freely at every step. This paper studies the realistic case where each agent may send only a few messages per episode. Instead of learning when to talk with high-variance trial-and-error, the authors label steps after the fact: a step is communication-critical if one agent was about to take a wrong action while its partner already knew the right one. A small supervised gate trained on those labels fires almost exclusively in the first few steps and when agents are confident, the opposite of the usual “talk when uncertain” rule. The explanation is recurrent: early messages inject aligned trajectory information into the agents’ GRU hidden states, and that alignment persists and compounds even after communication stops. With only three transmissions the learned gate nearly matches unlimited communication on hidden-state alignment and is far more efficient per message than random or entropy-based policies. The practical regime that emerges is simple: synchronise representations early, then navigate independently.

Core claim

Trained communication gates under hard per-episode budgets fire predominantly in early episode steps and at higher agent confidence, producing cumulative GRU hidden-state alignment of +0.072 with B=3 that approaches unconstrained communication (+0.078) and is 260% and 320% more alignment-efficient than random and entropy-based gating. This establishes a synchronisation-driven regime rather than uncertainty recovery.

What carries the argument

Hindsight gating: post-hoc binary labels mark a step communication-critical only when one agent erred and its partner already knew the correct action; a lightweight MLP gate is then trained by ordinary binary cross-entropy on the agent’s hidden state and remaining budget, converting a long-horizon policy-gradient problem into stable supervised classification.

Load-bearing premise

That higher cosine similarity between partner GRU hidden states is a meaningful measure of communication value even when actual navigation success rates barely move.

What would settle it

On a stronger single-agent backbone whose val-unseen success rate is high enough for messages to be useful, replace the learned early gate with late high-entropy gating at matched budget and check whether success rate and late-episode alignment reverse the reported ordering.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

Clean budgeted-comm VLN setup and a real early-sync finding, but SR barely moves and the asymmetric pairing may be doing more work than the paper admits.

read the letter

The useful core here is the problem, not the SR numbers. They define cooperative VLN with a hard per-episode transmission budget, replace REINFORCE gates with a simple hindsight BCE label (wrong while partner is right), and then show a consistent pattern: the gate fires early and at higher confidence, not when the agent is lost. At B=3 the cumulative GRU alignment gain (+0.072) nearly matches always-on (+0.078) and beats random and entropy by a wide margin, with the gain still growing after communication stops. That is a concrete, counter-intuitive timing result for recurrent multi-agent systems, and the baselines (random, entropy, always, none) are the right ones. Code is public.

What they do well: honest discussion that alignment is a proxy and can mean “similarly wrong,” explicit note that the backbone is weak without speaker data, and a gate that only sees hidden state plus remaining budget—so the early/high-confidence pattern is learned, not hard-coded. The formulation fills a real gap relative to Co-NavGPT/CAMON (free broadcast) and Farooq (compression, no language).

Soft spots, in proportion. First, navigation barely moves (8.7% → 8.9% val-unseen; std ≤0.4%), and the prose claim that Agent 0 exceeds the single-agent 9.2% baseline does not match Table 1 (8.9%). So the practical “helps navigation” story is thin; the paper’s own discussion almost admits this. Second, the stress-test concern is fair: Agent 1 starts at the midpoint with goal-region privilege, and hindsight labels only fire when one is wrong and the other already correct—so early steps are structurally more label-rich. Without a symmetric-pairing control, “synchronise early” may partly be an artifact of that construction rather than a general regime. Third, they never run a REINFORCE gate head-to-head, so the “avoids high variance” claim is methodological, not empirical. None of these kill the paper; they bound how far the regime claim travels.

Who it is for: people working on budgeted multi-agent communication, recurrent policies, or cooperative embodied agents who care about when to spend a message. Not for pure VLN leaderboard chasers—the absolute SR is too low. Math and citations look fine; no load-bearing formal errors. I would send it to referees: the problem is clean, the finding is sharp enough to argue about, and the limitations are mostly stated. Engage if you care about communication timing under hard budgets; treat the alignment numbers as evidence of a learned policy, not yet as proof of navigation value.

Referee Report

4 major / 0 minor

Summary. The paper introduces bandwidth-constrained cooperative VLN on R2R with a hard per-agent transmission budget B, and proposes hindsight gating: a BCE-supervised MLP gate trained from post-hoc labels that mark steps where one agent is wrong and its partner is correct (Eq. 2), avoiding REINFORCE. On an asymmetric pairing (Agent 0 full path; Agent 1 from the midpoint), trained gates fire mostly in early steps and at higher confidence (Table 2). The authors attribute this to recurrent hidden-state synchronisation: early context injection yields cumulative GRU cosine-alignment gains of +0.072 at B=3, near always-communicate (+0.078) and far above random (+0.020) and entropy-based (+0.017) gating (Table 3), with residual alignment after communication ceases. They conclude a regime of “synchronise early, navigate independently later.” Code is released.

Significance. If the early-synchronisation regime is real and not an artifact of the asymmetric construction, the paper supplies a useful, falsifiable design principle for bandwidth-limited multi-agent systems with recurrent policies, plus a practical alternative to high-variance REINFORCE gates. Strengths include matched-budget baselines (random, entropy, always, none), reported significance on alignment deltas, an explicit remaining-budget input, frozen-gate fine-tuning, and a public codebase. The counter-intuitive high-confidence early-firing pattern is interesting even if navigation SR gains remain small. The work is incremental relative to IC3Net-style gating and Co-NavGPT-style cooperative VLN, but the timing/alignment analysis is a genuine complementary contribution.

major comments (4)

§5.2 and Table 1: the prose states that Hindsight gate (B=3) “exceeds the single-agent baseline (9.2%)” and that selective communication “can improve individual navigation performance.” Table 1 reports Agent 0 val-unseen SR of 8.9% for Hindsight/Full-comm versus 9.2% single-agent and 8.7% No Comm (std ≤0.4%). 8.9% does not exceed 9.2%, and the 0.2-point lift over No Comm is within the stated noise. This is a load-bearing factual error for any claim of navigation benefit; the abstract/conclusion should not imply practical SR gains that Table 1 does not support. Align all SR claims strictly with Table 1 and treat alignment, not SR, as the primary evidence unless significance is shown.
§3 Asymmetric Path Assignment and Eq. (2): the central regime claim (early high-confidence gating → recurrent alignment) is measured under a strongly asymmetric pairing in which Agent 1 starts at v⌊T/2⌋ with privileged goal-region observations. Hindsight labels y_i(t)=1 only when the partner is already correct; under this construction early steps are systematically more likely to be labelled critical for Agent 0. Table 2’s early concentration (82.6% of fires in steps 0–2 at B=3) and Table 3’s large Δ vs random may therefore partly reflect the label distribution induced by role asymmetry rather than a general “synchronise early” principle. A symmetric-pairing control (or at least an ablation that randomises/removes midpoint privilege) is needed to show the regime is not an artifact; without it, §6 Limitations should substantially strengthen this caveat and the abstract should not present
Table 3 vs Table 4 cumulative alignment: Table 3 reports learned-gate ΣΔ = +0.072 on val-unseen at B=3; Table 4 reports +0.057 for B=3. The abstract and §5.4 quote +0.072. These cannot both be correct under the same definition of cumulative alignment. Reconcile the numbers, state the exact aggregation (which agents, which steps, which split), and ensure efficiency claims (260%/320%) use a single consistent baseline.
§5.4–§6 Relationship between Alignment and Navigation: cosine similarity of partner GRU states is treated as the primary success metric while val-unseen SR is essentially flat. The paper correctly notes alignment is an indirect proxy that could mean agents become “similarly wrong,” yet the title, abstract, and conclusion still sell a practically useful communication regime. Either (i) provide evidence that higher Δ predicts better downstream decisions (e.g., action-agreement with ground truth, recovery after early sync), or (ii) reframe the contribution as a characterisation of gate timing under hindsight labels, not as establishing a deployable regime. As written, the practical force of the strongest claim rests on an unvalidated proxy at a competence level the authors themselves flag as below the IC3Net threshold.

Circularity Check

0 steps flagged

No significant circularity: empirical gate training and post-hoc alignment measurements do not reduce to inputs by construction.

full rationale

This is an empirical multi-agent VLN methods paper, not a first-principles derivation. Hindsight labels (Eq. 2) supervise a BCE gate from observed failure asymmetry; the gate is then frozen and evaluated. Cumulative hidden-state alignment Δ is a post-hoc measurement against no-comm, random, entropy, and always baselines at matched budget—not the training objective—so the +0.072 vs +0.020/+0.017 comparisons are not tautological. Early high-confidence firing is reported as an observation under the trained gate, not derived from a uniqueness theorem or fitted parameter renamed as prediction. The paper’s own note that partner ranking at label-1 steps holds “by construction of Equation 2” is a transparent, non-load-bearing proxy check, not a circular claim about the synchronisation regime. Asymmetric pairing may confound when labels concentrate (a validity concern), but confounding is not circular reduction of a claimed derivation to its inputs. No self-citation uniqueness chain, no ansatz smuggled via prior author work, and no renaming of a known closed-form result. Score 0 is the honest finding.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 3 invented entities

The central claim rests on a constructed information asymmetry (complementary sub-paths), a proxy label for ‘communication-critical’ steps, and alignment as a stand-in for communication value when SR is flat. Free knobs (τ, B, architecture sizes) are ordinary engineering choices; the load-bearing domain assumptions are the pairing scheme, the label proxy, and GRU persistence of early messages. Invented entities are methodological (problem name, gate, regime label), not physical particles.

free parameters (4)

gate decision threshold τ = 0.4
Inference uses g_i(t)=1[p_send>τ] with τ=0.4; firing rates and budget use depend on this hand-chosen cutoff.
per-agent transmission budget B = primary B=3
Hard constraint B∈{1,3,5} defines the operating regime; primary claims use B=3.
hidden / context dimension = 512
Fixed at 512 to match CLIP features; shapes capacity of messages and GRU state used in alignment.
max path length / candidates = 20 / 15
Episode horizon 20 and max 15 candidates bound when early vs late steps are defined.

axioms (5)

domain assumption Complementary sub-path pairing (Agent 0 full path, Agent 1 from midpoint) creates genuine, useful information asymmetry for cooperative VLN.
§3 Asymmetric Path Assignment; all multi-agent results depend on this construction rather than independent co-located agents.
ad hoc to paper Hindsight label y_i(t)=1 only when agent i is wrong and partner j is correct on their own ground-truth actions is a valid proxy for expected communication value of broadcasting context c_i(t).
Eq. 2 and §4.2; necessary-but-conservative proxy; partner correctness is not receiver-action correctness.
domain assumption Cosine similarity of GRU hidden states measures beneficial synchronisation rather than shared error.
§5.4–§6; paper admits this is an indirect proxy when SR does not improve.
domain assumption Early injected messages persist and compound through subsequent GRU updates without further transmissions.
Recurrent propagation claim in abstract and §5.4; standard GRU property assumed to explain rising Δ after gates stop firing.
domain assumption Teacher-forced imitation on R2R without speaker augmentation is an acceptable backbone for studying communication timing.
§5.1–5.2; yields 9.2% val-unseen SR and large seen/unseen gap that limits message usefulness.

invented entities (3)

bandwidth-constrained cooperative VLN no independent evidence
purpose: Name the problem of hard per-agent message budgets on language-guided multi-agent navigation.
Problem formulation contribution; no independent field definition outside this paper.
hindsight gating no independent evidence
purpose: Replace REINFORCE communication gates with BCE on post-hoc failure-derived labels.
Core method; evidence is internal experiments only.
synchronisation-driven communication regime no independent evidence
purpose: Interpret early high-confidence sends as intentional hidden-state alignment rather than uncertainty recovery.
Interpretive label for observed gate behaviour and alignment curves; not independently measured outside this setup.

pith-pipeline@v1.1.0-grok45 · 13343 in / 3934 out tokens · 54156 ms · 2026-07-10T06:19:54.563802+00:00 · methodology

0 comments

read the original abstract

Most cooperative Vision-Language Navigation (VLN) methods assume unlimited communication, not considering real-world applications where bandwidth is restricted and information efficiency is critical. We introduce \textbf{bandwidth-constrained cooperative VLN} and propose \textbf{hindsight gating}: a lightweight supervised gate that labels communication-critical steps post-hoc from navigation failures, avoiding the high variance of REINFORCE. Contrary to the intuition that agents should communicate when uncertain, we observe a consistent counter-intuitive pattern: trained gates fire predominantly in early episode steps and more often when agents are confident, across all budget levels ($B \in \{1,3,5\}$). We explain this through \textbf{recurrent hidden-state alignment}: early communication injects grounded trajectory representations that persist and compound through subsequent Gated Recurrent Unit (GRU) updates, achieving $+0.072$ cumulative alignment gain with $B{=}3$ transmissions, approaching unconstrained communication ($+0.078$) at 260\% greater alignment efficiency than random gating ($+0.020$) and 320\% greater efficiency than entropy-based gating ($+0.017$). Our results establish a new communication regime for bandwidth-limited embodied agents: synchronise representations early, navigate independently later. Our codebase is available at: https://github.com/AravG13/bandwidth-constrained-cooperative-vln

discussion (0)

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages · 3 internal anchors

[1]

In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2018)

Anderson, P., Wu, Q., Teney, D., et al.: Vision-and-language navigation: Interpret- ing visually-grounded navigation instructions in real environments. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2018)

work page 2018
[2]

In: International Conference on 3D Vision (3DV) (2017)

Chang, A., Dai, A., Funkhouser, T., et al.: Matterport3d: Learning from rgb-d data in indoor environments. In: International Conference on 3D Vision (3DV) (2017)

work page 2017
[3]

In: Advances in Neural Information Processing Systems (NeurIPS) (2021)

Chen, S., Guhur, P.L., Schmid, C., Laptev, I.: History aware multimodal trans- former for vision-and-language navigation. In: Advances in Neural Information Processing Systems (NeurIPS) (2021)

work page 2021
[4]

In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022)

Chen, S., Guhur, P.L., Tapaswi, M., Schmid, C., Laptev, I.: Think global, act lo- cal: Dual-scale graph transformer for vision-and-language navigation. In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022)

work page 2022
[5]

In: International Conference on Machine Learning (ICML) (2019)

Das, A., Gerber, S., Mukherjee, S., Schwing, A.: TarMAC: Targeted multi-agent communication. In: International Conference on Machine Learning (ICML) (2019)

work page 2019
[6]

In: IEEE International Conference on Robotics and Automation (ICRA) (2026)

Farooq, M.U., Iqbal, S.: Bandwidth-efficient multi-agent communication through information bottleneck and vector quantization. In: IEEE International Conference on Robotics and Automation (ICRA) (2026)

work page 2026
[7]

In: Advances in Neural Information Processing Systems (NeurIPS) (2018)

Fried, D., Hu, R., Cirik, V., et al.: Speaker-follower models for vision-and-language navigation. In: Advances in Neural Information Processing Systems (NeurIPS) (2018)

work page 2018
[8]

In: European Confer- ence on Computer Vision (ECCV) (2020)

Krantz, J., Wijmans, E., Majumdar, A., Batra, D., Lee, S.: Beyond the nav-graph: Vision-and-language navigation in continuous environments. In: European Confer- ence on Computer Vision (ECCV) (2020)

work page 2020
[9]

Emergent Multi-Agent Communication in the Deep Learning Era

Lazaridou, A., Baroni, M.: Emergent multi-agent communication in the deep learn- ing era. arXiv preprint arXiv:2006.02419 (2020)

work page internal anchor Pith review Pith/arXiv arXiv 2006
[10]

Scalable 3D Registration via Truncated Entry-wise Absolute Residuals

Liu, P., Mu, Y., Wu, H., et al.: CAMON: Cooperative agents for multi-object navigation with llm-based conversations. arXiv preprint arXiv:2404.00915 (2024)

work page internal anchor Pith review Pith/arXiv arXiv 2024
[11]

In: Advances in Neu- ral Information Processing Systems (NeurIPS) (2017)

Lowe, R., Wu, Y., Tamar, A., Harb, J., Abbeel, P., Mordatch, I.: Multi-agent actor-critic for mixed cooperative-competitive environments. In: Advances in Neu- ral Information Processing Systems (NeurIPS) (2017)

work page 2017
[12]

In: Proceedings of the AAAI Conference on Artificial Intelli- gence (2018)

Mordatch, I., Abbeel, P.: Emergence of grounded compositional language in multi- agent populations. In: Proceedings of the AAAI Conference on Artificial Intelli- gence (2018)

work page 2018
[13]

In: International Conference on Machine Learn- ing (ICML) (2021)

Radford, A., Kim, J.W., Hallacy, C., et al.: Learning transferable visual models from natural language supervision. In: International Conference on Machine Learn- ing (ICML) (2021)

work page 2021
[14]

In: International Conference on Learn- ing Representations (ICLR) (2019)

Singh, A., Jain, T., Sukhbaatar, S.: Learning when to communicate at scale in mul- tiagent cooperative and competitive tasks. In: International Conference on Learn- ing Representations (ICLR) (2019)

work page 2019
[15]

In: Advances in Neural Information Processing Systems (NeurIPS) (2016) 12 A

Sukhbaatar, S., Szlam, A., Fergus, R.: Learning multiagent communication with backpropagation. In: Advances in Neural Information Processing Systems (NeurIPS) (2016) 12 A. Gupta et al

work page 2016
[16]

Co-NavGPT: Multi-Robot Cooperative Visual Semantic Navigation Using Vision Language Models

Yu, B., Kasaei, H., Cao, M.: Co-NavGPT: Multi-robot cooperative visual semantic navigation using large language models. arXiv preprint arXiv:2310.07937 (2023)

work page internal anchor Pith review Pith/arXiv arXiv 2023

[1] [1]

In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2018)

Anderson, P., Wu, Q., Teney, D., et al.: Vision-and-language navigation: Interpret- ing visually-grounded navigation instructions in real environments. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2018)

work page 2018

[2] [2]

In: International Conference on 3D Vision (3DV) (2017)

Chang, A., Dai, A., Funkhouser, T., et al.: Matterport3d: Learning from rgb-d data in indoor environments. In: International Conference on 3D Vision (3DV) (2017)

work page 2017

[3] [3]

In: Advances in Neural Information Processing Systems (NeurIPS) (2021)

Chen, S., Guhur, P.L., Schmid, C., Laptev, I.: History aware multimodal trans- former for vision-and-language navigation. In: Advances in Neural Information Processing Systems (NeurIPS) (2021)

work page 2021

[4] [4]

In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022)

Chen, S., Guhur, P.L., Tapaswi, M., Schmid, C., Laptev, I.: Think global, act lo- cal: Dual-scale graph transformer for vision-and-language navigation. In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022)

work page 2022

[5] [5]

In: International Conference on Machine Learning (ICML) (2019)

Das, A., Gerber, S., Mukherjee, S., Schwing, A.: TarMAC: Targeted multi-agent communication. In: International Conference on Machine Learning (ICML) (2019)

work page 2019

[6] [6]

In: IEEE International Conference on Robotics and Automation (ICRA) (2026)

Farooq, M.U., Iqbal, S.: Bandwidth-efficient multi-agent communication through information bottleneck and vector quantization. In: IEEE International Conference on Robotics and Automation (ICRA) (2026)

work page 2026

[7] [7]

In: Advances in Neural Information Processing Systems (NeurIPS) (2018)

Fried, D., Hu, R., Cirik, V., et al.: Speaker-follower models for vision-and-language navigation. In: Advances in Neural Information Processing Systems (NeurIPS) (2018)

work page 2018

[8] [8]

In: European Confer- ence on Computer Vision (ECCV) (2020)

Krantz, J., Wijmans, E., Majumdar, A., Batra, D., Lee, S.: Beyond the nav-graph: Vision-and-language navigation in continuous environments. In: European Confer- ence on Computer Vision (ECCV) (2020)

work page 2020

[9] [9]

Emergent Multi-Agent Communication in the Deep Learning Era

Lazaridou, A., Baroni, M.: Emergent multi-agent communication in the deep learn- ing era. arXiv preprint arXiv:2006.02419 (2020)

work page internal anchor Pith review Pith/arXiv arXiv 2006

[10] [10]

Scalable 3D Registration via Truncated Entry-wise Absolute Residuals

Liu, P., Mu, Y., Wu, H., et al.: CAMON: Cooperative agents for multi-object navigation with llm-based conversations. arXiv preprint arXiv:2404.00915 (2024)

work page internal anchor Pith review Pith/arXiv arXiv 2024

[11] [11]

In: Advances in Neu- ral Information Processing Systems (NeurIPS) (2017)

Lowe, R., Wu, Y., Tamar, A., Harb, J., Abbeel, P., Mordatch, I.: Multi-agent actor-critic for mixed cooperative-competitive environments. In: Advances in Neu- ral Information Processing Systems (NeurIPS) (2017)

work page 2017

[12] [12]

In: Proceedings of the AAAI Conference on Artificial Intelli- gence (2018)

Mordatch, I., Abbeel, P.: Emergence of grounded compositional language in multi- agent populations. In: Proceedings of the AAAI Conference on Artificial Intelli- gence (2018)

work page 2018

[13] [13]

In: International Conference on Machine Learn- ing (ICML) (2021)

Radford, A., Kim, J.W., Hallacy, C., et al.: Learning transferable visual models from natural language supervision. In: International Conference on Machine Learn- ing (ICML) (2021)

work page 2021

[14] [14]

In: International Conference on Learn- ing Representations (ICLR) (2019)

Singh, A., Jain, T., Sukhbaatar, S.: Learning when to communicate at scale in mul- tiagent cooperative and competitive tasks. In: International Conference on Learn- ing Representations (ICLR) (2019)

work page 2019

[15] [15]

In: Advances in Neural Information Processing Systems (NeurIPS) (2016) 12 A

Sukhbaatar, S., Szlam, A., Fergus, R.: Learning multiagent communication with backpropagation. In: Advances in Neural Information Processing Systems (NeurIPS) (2016) 12 A. Gupta et al

work page 2016

[16] [16]

Co-NavGPT: Multi-Robot Cooperative Visual Semantic Navigation Using Vision Language Models

Yu, B., Kasaei, H., Cao, M.: Co-NavGPT: Multi-robot cooperative visual semantic navigation using large language models. arXiv preprint arXiv:2310.07937 (2023)

work page internal anchor Pith review Pith/arXiv arXiv 2023