Pith. sign in

REVIEW 6 minor 47 references

Capability Sheaves for Compositional Agent-Harness Repair: Controlled Quotients and a Real-Repository Stress Test

T0 review · 0 major / 6 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 This paper tries to establish a boundary: quotienting hidden-state nuisance in a capability sheaf halves the candidate search budget in controlled agent-harness tasks, but the same cohomological repair does not beat matched evidence on…

desk verdict Honest boundary paper: controlled invariance mechanism demonstrated, real-repo negative result reported cleanly; worth serious peer review. read the letter →

arxiv 2608.13228 v1 pith:IHYGBLAW submitted 2026-08-13 cs.AI cs.LG

Saveliy Batruin This is my paper · ORCID
classification cs.AIcs.LG MSC 18F2055N3068T20
keywords capabilitysheavesagentharnessrepairrelativecohomologyconstraintsatisfactionhidden-statequotientpatchfusiongluingobstructionreal-repositoryevaluation
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 tries to establish a precise boundary for a sheaf-theoretic repair method for agent harnesses: quotienting out a hidden mediator's stale state reliably halves the candidate search budget in a controlled task family, while the same cohomological construction does not improve real repository patch fusion. The model represents an agent harness as a finite capability sheaf whose stalks are typed behavior signatures and whose restrictions are literal field projections; an exact constraint-satisfaction problem decides acceptance, and a linearized relative cohomology class serves as a diagnostic and search score. In the controlled experiment, all 20 task clusters show the quotient policy reaching first success in one candidate evaluation versus two for the raw stale representative, and aligning the hidden state removes the gap. On the real benchmark, the first pool-level class is constant and cannot rank candidates; a candidate-indexed repair becomes discriminative but resolves only two more issues than a matched noncohomological selector, its abstention gate ties the anchor, and the preregistered development gate fails. The paper's contribution is therefore a boundary: an invariance mechanism that holds in a controlled family and a real-world transfer that fails its own preregistered gate.

What carries the argument

The central object is a finite capability sheaf: a cellular sheaf on a five-vertex incidence graph whose vertices are the requirements localization, contract, ordering, preservation, and verification, with typed behavior-signature stalks and restriction maps that are literal field projections. The acceptance rule is the exact CSP of Equation (1): every local signature must lie in its registered good subset and every registered pair of restrictions must agree. The diagnostic is the relative cellular-sheaf cohomology class $\partial[s_A] = [\delta^0 s_A] \in H^1(X,A;\mathbb{F})$, which vanishes exactly when adjacent typed restrictions agree. The controlled experiment subdivides each overlap with a hidden mediator vertex; changing the hidden value adds an interior coboundary $D_j u = (u,u)$, and the quotient map $P_j = [I\ I]$ kills it, leaving the public endpoint mismatch $\ell_j + r_j$. The real experiment's repair indexes the complex by the candidate action, replacing the constant pool-level class with per-candidate scores $q(S)$ computed from candidate-restricted matrices $D_S$. The exact CSP is the semantic control throughout; cohomology is only an invariant diagnostic, never a replacement for exact feasibility or execution.

What would settle it

On the same 160-issue development split, rerun the candidate-indexed repair with a second, independently constructed evidence model for edit obligations and risks, or with executed-test feedback replacing model scores; if the method then clears the preregistered gate of at least four issue gains, a positive repository-macro effect, six nonzero repositories, and $p \leq 0.2$, the paper's claim that the tested obligation-and-risk complex is too weak for real patch fusion would be overturned.

Watch

Extended reading notes

Core claim

The central discovery, stated on the paper's own terms, is that hidden-state quotienting works as a mechanism but not as a real-world advantage. In the controlled family, inserting a hidden mediator on each overlap coordinate makes the raw residual depend on the mediator's value, while the quotient class $[q_j(h_j)]$ maps to the public endpoint mismatch $\ell_j + r_j$ and is independent of $h_j$; this is what lets the quotient policy stop after one candidate evaluation in every one of the 20 clusters, compared with two for the stale-raw policy, and the aligned-interior ablation confirms that the gain is specifically about removing a nuisance representative. The real-repository stress test then supplies the boundary: because $[b - Dx] = [b]$ in $\operatorname{coker} D$ for any two selections $x_1$ and $x_2$, a single pool-level class cannot rank candidate configurations. The candidate-indexed repair changes the complex per candidate and is nontrivial on 848 of 875 candidates and varies within 120 of 160 issues, but its direct selection gain is two issues at exact repository sign-flip $p = 0.75$ and its leave-one-repository-out abstention ties the strong anchor at 127 of 160 with $p = 1.0$, so the preregistered development gate fails and the confirmatory split stays sealed. The exact CSP with the same restrictions matches or exceeds the quotient in every controlled setting, so the claim is invariance to stale representatives, not superiority over exact reasoning.

Load-bearing premise

The real-repository negative result rests on a single pass of model-generated obligation-and-risk evidence over edit atoms, and if that evidence misses the true semantic conflicts between edits, the failure belongs to the evidence model rather than to the cohomological method.

Editorial extensions

If this is right

  • If the controlled invariance result is correct, agent-harness searches that score raw half-edge residuals can be misled by stale cached representatives, and quotienting those interior coboundaries removes the artifact without changing the public gluing problem.
  • If the identifiability counterexample is taken at face value, a single pool-level cohomology class can never rank candidate configurations, so any future linear diagnostic must index the complex by the candidate or action under consideration.
  • If the real-repository negative result is accepted as the present boundary, cohomological patch fusion will not beat matched semantic evidence until the obligation-and-risk rows capture actual edit-edit semantic conflicts rather than model-generated judgments.
  • In every tested setting, the exact CSP with the same restrictions matches or exceeds the quotient, so cohomological search scores are complements to, not replacements for, exact feasibility and execution.

Reading between the lines

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

  • Editorial extension: the identifiability defect likely applies to any method that attaches one quotient class to an entire candidate pool rather than to each candidate, so the candidate-indexed repair is a template for other topological or linear diagnostics in decision problems.
  • Editorial extension: since the controlled gain disappears when the hidden state is aligned, the mechanism predicts that production harnesses with synchronized caches will show no benefit from quotienting; measuring cache-staleness rates in deployed agents would test how often the controlled situation arises in the wild.
  • Editorial extension: the real-benchmark evidence model is the obvious lever—replacing the single model pass with executed-test feedback or a second independently generated evidence pass on the same 160 issues would tell whether the boundary is fundamental or just a limitation of the available semantic rows.
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

0 major / 6 minor

Summary. The paper introduces a finite 'capability sheaf' model for agent harnesses: typed stalks and literal field-projection restrictions encode local capability signatures, an exact CSP defines global acceptance, and a relative cohomology class over F2 provides a diagnostic score. Four formal results (gluing, portfolio certificate, repair-order separation, rank-truncated spectral stability, and finite-trace recovery) are proved. In a preregistered controlled experiment over 20 task clusters, quotienting hidden interior mediators reduces the first-success candidate budget from 2.000 to 1.000 relative to a stale raw score, and the aligned-state negative control removes the gap. In a real-repository stress test on 160 issues from 20 PatchFuseBench repositories, the full-pool cokernel class is constant by Eq. (7); a candidate-indexed repair is nontrivial on 848/875 candidates but yields only +2 issues over a matched selector (p=0.75) and a LOO abstention ties the anchor (p=1.0), so the development gate fails and the confirmatory split remains sealed. The paper's stated main result is the boundary: controlled invariance holds, while the tested obligation-and-risk complex does not support a real-world cohomological advantage.

Significance. If the results stand, the paper makes a useful methodological contribution: it demonstrates a controlled mechanism by which quotienting hidden-state nuisance variables removes a spurious search gap, and it provides an honest, carefully bounded negative result on a real repository task. The manuscript is unusually disciplined: the controlled design is frozen in advance, exact CSP is used as the semantic control, an aligned-state ablation isolates the mechanism, repository-level inference is used rather than issue-level pseudoreplication, and the confirmatory split is kept sealed after the development gate fails. The mathematical statements are mostly standard but are presented with clean hypotheses and counterexamples, and the proof-backed algebraic identity in Eq. (7) correctly identifies a genuine identifiability defect in the full-pool construction. The negative real-world result is reported with exact p-values and explicit threats to validity, which strengthens rather than weakens the paper. The work is not a general optimizer breakthrough, and the authors do not claim one.

minor comments (6)
  1. [Section 2.3] The cochain dimensions dim C0=246, dim C1=233, rank δ0=173, and the derived H1 dimensions are asserted without derivation; please add the incidence-graph counts or point to the artifact script that computes them so a reader can verify the numbers.
  2. [Section 6.1] The sentence 'In all 4,000 registered candidate–target–intervention checks, the quotient signature is unchanged' is presented alongside empirical outcomes; it would be clearer to state that this is a software-verified algebraic identity implied by Eq. (5), not an empirical observation.
  3. [Abstract and Figure 3] The abstract writes the budget reduction as '2,000 to 1,000' while the body uses '2.000 to 1.000'; please standardize the decimal notation to avoid confusion with thousands separators.
  4. [Section 5.4 / Table 2] The identical token rows for exact CSP and full relative class follow by construction because both policies stop at the same first-success candidate; the prose says this, but adding a table footnote would prevent misreading.
  5. [Section 7.1] The 'strong repository-held-out selector' is central to the abstention benchmark but is described only briefly; please add a sentence or two on how it is constructed or cite the artifact location so the anchor comparison is reproducible.
  6. [Section 8] The statement that 'Independent GLM-5 and Qwen hunk studies earlier in development showed material evidence-model dependence' is an important threat but has no pointer to where those studies are documented; adding a reference or artifact path would make the limitation auditable.

Circularity Check

1 steps flagged · score 4.0 of 10

Controlled quotient-vs-CSP agreement is definitional via Eq. (5), but the real-repository stress test is independent and honestly negative; central claim retains empirical content.

  1. self definitional [Section 2.4, Eq. (5); Section 6.2, Primary result and ablation]
    "Hence Q_j=(V_j⊕V_j)/im D_j ≅ V_j, [q_j(h_j)]↦ℓ_j+r_j. ... For actual one-hot endpoints, the class is zero exactly when ℓ_j=r_j on every registered coordinate. ... Exact CSP also matches the quotient in every cluster, as predicted."

    By Eq. (5), the quotient class is defined to be the public endpoint mismatch ℓ_j+r_j, and exact CSP acceptance requires equality of the projected endpoints on every registered coordinate. Therefore the reported agreement between the quotient and exact CSP in all 20 clusters is an algebraic identity from the construction, not an empirical discovery. The same holds for the aligned-left negative control: h_j=ℓ_j makes the raw half-edge vector (0, ℓ_j+r_j), so the raw aligned score equals the quotient by definition, and a zero gap is a consistency check. The genuinely empirical part is the 2.000→1.000 stopping-budget reduction against the raw-stale policy; the paper explicitly frames the result as invariance to stale representatives, not superiority over exact reasoning.

full rationale

The only load-bearing reduction-by-construction I can exhibit is the controlled quotient's agreement with exact CSP and with the aligned ablation: it follows directly from Eq. (5) that the quotient class is the endpoint mismatch, so 'exact CSP matches the quotient' is not a datum-generated prediction. However, the paper itself states this boundary ('the result demonstrates invariance to stale representatives, not superiority over exact reasoning'), so this is an acknowledged identity rather than a disguised overclaim. The controlled experiment's central empirical content—raw-stale needing 2.000 candidate evaluations versus 1.000 for the quotient across 20 preregistered clusters with p=9.54e-7—is not forced by the definition and depends on the 1,000 model outcomes. The real-repository stress test is fully independent: Eq. (7) is used to expose a defect in the full-pool class, the candidate-indexed repair is evaluated against 153 freshly executed patches, and the paper reports a failed development gate, p=0.75 direct and p=1.0 abstention. That negative result cannot be manufactured by any algebraic identity. There are no load-bearing self-citations (the paper is single-authored and cites only external prior work), no imported uniqueness theorems, and no renamed known result presented as a prediction. The residual circularity is therefore partial and confined to the controlled quotient-vs-CSP identity; the central boundary claim remains independently supported.

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

Everything the central claim rests on: the typed-field model, the sheaf gluing assumption, the one-hot representation, and the trace-freshness assumptions. The controlled result depends on the algebraic quotient identity; the real result depends on the single-pass GLM evidence.

free parameters (4)
  • learned sparse cover (six overlaps) = L-V, L-P, C-P, O-V, P-V, C-O
    The cover is learned from training labels (Section 5.2), and the six overlaps determine the sheaf whose cohomology is computed; different overlaps would change the obstruction class.
  • incidence graph construction = maximum-weight spanning tree plus two strongest redundant edges
    Section 5.2: the incidence graph is selected from trace data; it defines dim C0 and dim C1 and therefore the reported cohomology dimensions.
  • filler cost schedule = workspace partition plus dependency checkpoint at cost 3
    Section 3.1: the outcome-blind completion operator minimizes filler cost; costs are assigned by hand, and the selected bundle depends on them.
  • rank truncation r (Theorem 4) = registered rank
    The stability guarantee requires the true rank r of D to be known or registered; in applications this is a modeling assumption.
assumptions (4)
  • domain assumption Cellular sheaf gluing axiom holds for the finite typed-field complex
    Theorem 1 assumes every overlap needed for gluing is registered; if an unmodeled overlap hides a conflict, the CSP misses it.
  • domain assumption One-hot encoding over F2 exactly represents typed field equality
    Section 2.3 maps field values to one-hot basis vectors; the quotient class is computed in this representation.
  • domain assumption Freshness of complete traces (independence across traces)
    Corollary 5 and Theorem 6 require independent complete traces; the paper notes cached duplicates do not increase n.
  • standard math Eigenvalue gap and norm hypotheses in Theorem 4
    Wedin's theorem is invoked under the stated gap condition; the hypotheses are standard for subspace perturbation bounds.
invented entities (3)
  • capability sheaf independent evidence
    purpose: Formal model of typed local behaviors and their shared-state restrictions
    The sheaf is a mathematical construction, but its predictions are tested in the controlled experiment and the real benchmark.
  • hidden interior mediator vertices
    purpose: Model nuisance hidden states in the quotient experiment
    The mediators are introduced by the paper to demonstrate stale-representative invariance; the aligned ablation is a direct intervention, but the entity is not independently observable.
  • obligation-and-risk evidence rows
    purpose: Real-benchmark features for the quotient router
    The rows are derived from a single GLM-5 pass and are not independently verified; the paper notes evidence-model dependence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Capability Sheaves for Compositional Agent-Harness Repair: Controlled Quotients and a Real-Repository Stress Test." pith.science (2026). https://pith.science/paper/IHYGBLAW

@misc{pith2026260813228,
  author       = {Pith},
  title        = {Pith review of: Capability Sheaves for Compositional Agent-Harness Repair: Controlled Quotients and a Real-Repository Stress Test},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IHYGBLAW}},
  note         = {Machine review of arXiv:2608.13228}
}
abstract

Agent harnesses combine retrieval, routing, state, provenance, and verification, but locally successful components may disagree on shared state. We model this failure with a finite \emph{capability sheaf}: stalks encode typed behavior signatures, restriction maps retain shared fields, and accepted runs are useful global sections. An exact finite constraint-satisfaction problem (CSP) defines acceptance, while a linearized relative cohomology class provides a diagnostic and search feature. A controlled experiment over 20 task clusters introduces hidden interior mediators whose raw states are nuisance variables. Quotienting their coboundaries reduces the candidate budget from 2,000 to 1,000 per cluster; aligning the hidden state removes the gap. Exact CSP matches the quotient, so the result demonstrates invariance to stale representatives, not superiority over exact reasoning. We then test the method on a discovery split from the SWE-bench Multilingual pool of PatchFuseBench: 160 issues from 20 repositories, 875 real candidate patches, 2,579 source-aware edit atoms, and 153 newly executed patches. A first pool-level construction is constant because $[b-Dx]=[b]$ in $\operatorname{coker}D$ and therefore cannot rank configurations. A candidate-indexed repair is nontrivial on 848/875 candidates and varies within 120/160 issues. It resolves 118 issues versus 116 for a matched noncohomological selector, but the difference is not supported across repositories (exact sign-flip $p=0.75$). A leave-one-repository-out abstention gate reaches 127/160, tying the strong anchor and exceeding its matched gate by one issue ($p=1.0$). The discovery gate therefore fails and the confirmatory split remains sealed. The study supports the controlled invariance mechanism and an identifiability correction, but not a real-world cohomological advantage.

Figures

Figures reproduced from arXiv: 2608.13228 by the authors.

Figure 1
Figure 1. A finite and matrix model of the hidden-state quotient. The raw half-edge residual changes when the mediator state ℎ changes. The quotient map removes this interior coboundary and keeps only the endpoint disagreement ℓ + 𝑟. The picture does not replace the exact CSP: class zero checks the registered gluing fields, while useful local sections and execution are separate requirements. 3. Repair and Search 3.1. Typed fi… view at source ↗
Figure 2
Figure 2. One confirmatory experiment. Panel A reports the equal-pool stopping budget. Panel B reports total tokens to first success separately inside each endpoint. Panel C shows heldout balanced accuracy of structural criteria. Panel D compares the outcome-blind generated filler with the registered all-local structural decoy [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Preregistered hidden-interior result. Panel A shows the equal-pool stopping budget. Panel B is the causal check: the stale representative creates a gap, while aligning the hidden state removes it. Panel C shows all 20 paired task-cluster differences, not candidate-level pseudo-replicates. Panel D shows total-token reductions within each endpoint. The figure is computational evidence for this frozen task family, not … view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Real patch-fusion audit. Panel A is a finite and matrix counterexample: the full-pool class is identical for all selections; this is the proof-backed statement in Equation (7). Panel B shows the candidate-indexed repair. Its different scores are an algebraic computatio…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 35 canonical work pages

  1. [1]

    New Journal of Physics , volume =

    Samson Abramsky and Adam Brandenburger , title =. New Journal of Physics , volume =

    work page

  2. [2]

    Proceedings of the 8th International Workshop on Quantum Physics and Logic , series =

    Samson Abramsky and Shane Mansfield and Rui Soares Barbosa , title =. Proceedings of the 8th International Workshop on Quantum Physics and Logic , series =

    work page

  3. [3]

    Kolaitis , title =

    Samson Abramsky and Georg Gottlob and Phokion G. Kolaitis , title =. Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence , pages =

    work page

  4. [4]

    24th EACSL Annual Conference on Computer Science Logic (CSL 2015) , series =

    Samson Abramsky and Rui Soares Barbosa and Kohei Kishida and Raymond Lal and Shane Mansfield , title =. 24th EACSL Annual Conference on Computer Science Logic (CSL 2015) , series =. doi:10.4230/LIPIcs.CSL.2015.211 , year =

    doi:10.4230/lipics.csl.2015.211 2015

  5. [5]

    Journal of Machine Learning Research , volume =

    Henry Adams and Tegan Emerson and Michael Kirby and Rachel Neville and Chris Peterson and Patrick Shipman and Sofya Chepushtanova and Eric Hanson and Francis Motta and Lori Ziegelmeier , title =. Journal of Machine Learning Research , volume =

  6. [6]

    Anton Ayzenberg and Thomas Gebhart and German Magai and Grigory Solomadin , title =

    work page

  7. [7]

    Sheaf Neural Networks with Connection Laplacians , booktitle =

    Federico Barbero and Cristian Bodnar and Haitz S. Sheaf Neural Networks with Connection Laplacians , booktitle =

    work page

  8. [8]

    Bronstein , title =

    Cristian Bodnar and Francesco Di Giovanni and Benjamin Paul Chamberlain and Pietro Li\`o and Michael M. Bronstein , title =. Advances in Neural Information Processing Systems , volume =

    work page

Show all 47 references
  1. [9]

    Bredon , title =

    Glen E. Bredon , title =

  2. [10]

    Journal of Machine Learning Research , volume =

    Peter Bubenik , title =. Journal of Machine Learning Research , volume =

  3. [11]

    Bulletin of the American Mathematical Society , volume =

    Gunnar Carlsson , title =. Bulletin of the American Mathematical Society , volume =

  4. [12]

    On the Cohomology of Contextuality , howpublished =

    Giovanni Car. On the Cohomology of Contextuality , howpublished =

  5. [13]

    Mathematics of Operations Research , volume =

    Vasek Chv\'atal , title =. Mathematics of Operations Research , volume =

  6. [14]

    Discrete and Computational Geometry , volume =

    David Cohen-Steiner and Herbert Edelsbrunner and John Harer , title =. Discrete and Computational Geometry , volume =

  7. [15]

    Chandler Davis and W. M. Kahan , title =. SIAM Journal on Numerical Analysis , volume =. doi:10.1137/0707001 , year =

    doi

  8. [16]

    2014 , note =

    Justin Michael Curry , title =. 2014 , note =

    2014

  9. [17]

    Meyarivan , title =

    Kalyanmoy Deb and Amrit Pratap and Sameer Agarwal and T. Meyarivan , title =. IEEE Transactions on Evolutionary Computation , volume =. doi:10.1109/4235.996017 , year =

  10. [18]

    Herbert Edelsbrunner and John Harer , title =

  11. [19]

    Robert Ghrist , title =

  12. [20]

    Stephan Felber and Bernardo Hummes Flores and Hugo Rincon Galeana , title =

  13. [21]

    Journal of Applied and Computational Topology , volume =

    Jakob Hansen and Robert Ghrist , title =. Journal of Applied and Computational Topology , volume =

  14. [22]

    Proceedings of the 26th International Conference on Artificial Intelligence and Statistics , series =

    Thomas Gebhart and Jakob Hansen and Paul Schrater , title =. Proceedings of the 26th International Conference on Artificial Intelligence and Statistics , series =

  15. [23]

    Olivieri and Roque J

    David N. Olivieri and Roque J. Hern. Sheaf-Theoretic Transport and Obstruction for Detecting Scientific Theory Shift in. doi:10.48550/arXiv.2605.14033 , year =

    doi arXiv

  16. [24]

    doi:10.48550/arXiv.2204.13446 , year =

    Florian Russold , title =. doi:10.48550/arXiv.2204.13446 , year =

    doi arXiv

  17. [25]

    Perturbation Bounds in Connection with Singular Value Decomposition , journal =

    Per-. Perturbation Bounds in Connection with Singular Value Decomposition , journal =. doi:10.1007/BF01932678 , year =

    doi

  18. [26]

    Allen Hatcher , title =

  19. [27]

    Tohoku Mathematical Journal, Second Series , volume =

    Kazuoki Azuma , title =. Tohoku Mathematical Journal, Second Series , volume =. doi:10.2748/tmj/1178243286 , year =

  20. [28]

    Journal of the American Statistical Association , volume =

    Wassily Hoeffding , title =. Journal of the American Statistical Association , volume =

  21. [29]

    International Conference on Learning Representations , year =

    Shengran Hu and Cong Lu and Jeff Clune , title =. International Conference on Learning Representations , year =

  22. [30]

    Joshi and Hanna Moazam and Heather Miller and Matei Zaharia and Christopher Potts , title =

    Omar Khattab and Arnav Singhvi and Paridhi Maheshwari and Zhiyuan Zhang and Keshav Santhanam and Sri Vardhamanan and Saiful Haq and Ashutosh Sharma and Thomas T. Joshi and Hanna Moazam and Heather Miller and Matei Zaharia and Christopher Potts , title =

  23. [31]

    Mengzhuo Chen and Junjie Wang and Zhe Liu and Yawen Wang and Qing Wang , title =

  24. [32]

    Yoonho Lee and Roshen Nair and Qizheng Zhang and Kangwook Lee and Omar Khattab and Chelsea Finn , title =

  25. [33]

    Journal of Machine Learning Research , volume =

    Lisha Li and Kevin Jamieson and Giulia DeSalvo and Afshin Rostamizadeh and Ameet Talwalkar , title =. Journal of Machine Learning Research , volume =

  26. [34]

    Liu and Kevin Lin and John Hewitt and Ashwin Paranjape and Michele Bevilacqua and Fabio Petroni and Percy Liang , title =

    Nelson F. Liu and Kevin Lin and John Hewitt and Ashwin Paranjape and Michele Bevilacqua and Fabio Petroni and Percy Liang , title =. Transactions of the Association for Computational Linguistics , volume =

  27. [35]

    Saunders Mac Lane and Ieke Moerdijk , title =

  28. [36]

    47th International Symposium on Mathematical Foundations of Computer Science (MFCS 2022) , series =

    Adam \'O Conghaile , title =. 47th International Symposium on Mathematical Foundations of Computer Science (MFCS 2022) , series =. doi:10.4230/LIPIcs.MFCS.2022.75 , year =

    2022 doi

  29. [37]

    Patil and Ion Stoica and Joseph E

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

  30. [38]

    Information Fusion , volume =

    Michael Robinson , title =. Information Fusion , volume =

  31. [39]

    Varun Ursekar and Apaar Shanker and Veronica Chatrath and Yuan Xue and Sam Denton , title =

  32. [40]

    International Conference on Learning Representations , year =

    Jiayi Zhang and Jinyu Xiang and Zhaoyang Yu and Fengwei Teng and Xionghui Chen and Jiaqi Chen and Mingchen Zhuge and Xin Cheng and Sirui Hong and Jinlin Wang and Bingnan Zheng and Bang Liu and Yuyu Luo and Chenglin Wu , title =. International Conference on Learning Representat...

  33. [41]

    Lyu and Wei Chen , title =

    Shouyuan Chen and Tian Lin and Irwin King and Michael R. Lyu and Wei Chen , title =. Advances in Neural Information Processing Systems 27 , pages =

  34. [42]

    Journal of Machine Learning Research , volume =

    Eyal Even-Dar and Shie Mannor and Yishay Mansour , title =. Journal of Machine Learning Research , volume =

  35. [43]

    On the Complexity of Best-Arm Identification in Multi-Armed Bandit Models , journal =

    Emilie Kaufmann and Olivier Capp. On the Complexity of Best-Arm Identification in Multi-Armed Bandit Models , journal =

  36. [44]

    Proceedings of the 19th International Conference on Artificial Intelligence and Statistics , series =

    Daniel Russo and James Zou , title =. Proceedings of the 19th International Conference on Artificial Intelligence and Statistics , series =. 2016 , url =

    2016

  37. [45]

    An Information-Theoretic Approach to Minimax Regret in Partial Monitoring , booktitle =

    Tor Lattimore and Csaba Szepesv. An Information-Theoretic Approach to Minimax Regret in Partial Monitoring , booktitle =. 2019 , url =

    2019

  38. [46]

    Jimenez and John Yang and Alexander Wettig and Shunyu Yao and Kexin Pei and Ofir Press and Karthik Narasimhan , title =

    Carlos E. Jimenez and John Yang and Alexander Wettig and Shunyu Yao and Kexin Pei and Ofir Press and Karthik Narasimhan , title =. International Conference on Learning Representations , year =

  39. [47]

    A Single Patch Is Not Enough: Deterministic Fusion of Repair Candidates , howpublished =

    Boyang Yang and Xiangliang Hu and Luyao Ren and Yanjun Chen and Bach Le and Tegawend. A Single Patch Is Not Enough: Deterministic Fusion of Repair Candidates , howpublished =. 2026 , url =

    2026

Pith tools

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