Pith. sign in

REVIEW 3 major objections 4 minor 33 references

WorldMark: A Plug-and-Play World Knowledge Interface for Cross-Host Language Model Watermarking

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read WorldMark claims a plug-and-play knowledge-saliency interface that improves attacked detection and text quality at the same time, across three adaptive-strength MorphMark host variants, without changing the detector or retraining the…

desk verdict Novel knowledge-saliency watermarking idea, but the headline gains rest on reproductions that underperform the original MorphMark numbers, so the central claim is conditional. read the letter →

arxiv 2608.06416 v1 pith:ZCTRUDWW submitted 2026-08-05 cs.CR cs.AI

Song Xiao , Yuqi Yuan , Yanshuo Zhang , Kejun Zhang This is my paper · ORCID
classification cs.CRcs.AI
keywords LLMwatermarkingknowledgesaliencyworldmemoryasymmetricmodulationadaptive-strengthplug-and-playinterfaceattackeddetectionC4evaluation
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

WorldMark proposes that where an LLM watermark is placed should depend on which tokens carry confirmed world knowledge, not just on local token probability. The paper builds a semantic-episodic memory graph from the prompt and observed context, scores each decoding position by how strongly the candidate distribution aligns with retrieved knowledge, and uses that saliency score to relax the watermark on anchored tokens while strengthening it on unanchored ones. On the primary C4 protocol with three MorphMark host variants, the complete interface reports higher clean and attacked detection and slightly lower perplexity than the reproduced baselines, with average attacked TPR +0.0296 and attacked Best F1 +0.0108. Direct memory injection without saliency-aware modulation is unstable across hosts, which the paper treats as evidence that the modulation mechanism, rather than the knowledge content alone, carries the benefit. If the pattern holds, knowledge-aware placement offers a host-agnostic way to make watermarks more attack-resistant without retraining the generator or modifying the detector.

What carries the argument

The load-bearing object is the knowledge saliency score $s_t$, a soft cosine similarity between the serialized retrieved knowledge context and the top-$m$ candidate tokens at each decoding step. It is converted into the asymmetric modulation factor $\mu_t = (1-\beta_p s_t)(1+\beta_d(1-s_t))$, which is the mechanism that moves watermark strength away from knowledge-anchored positions and toward unanchored ones. Because the knowledge context $c_0$ is fixed per generation episode while the candidate distribution $p_t^k$ changes with the prefix, the score is position-dependent without requiring per-token retrieval or access to future tokens. The same $\mu_t$ is applied to adaptive-logits strengths, sampling probabilities, and entropy-gate thresholds, which is what makes the interface host-agnostic while leaving the original detector untouched.

What would settle it

Independently re-run the primary C4 evaluation with a verified faithful implementation of MorphMark that matches the original published attacked TPR values (0.9600 for exponential, 0.9275 for linear, 0.9375 for logarithmic) rather than the reproduction values; if the average attacked TPR gain over that faithful baseline is not positive, the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that a token-level knowledge saliency signal derived from a world-knowledge memory graph can be spliced into existing watermarks as a plug-and-play interface, improving attacked detection and text quality at the same time. Specifically, WorldMark combines World Knowledge Memory (a semantic-episodic graph with factual triples and supporting observations), a Knowledge Saliency Estimator computing $s_t = \sigma(\lambda \cdot \mathrm{sim}(\varphi(c_0), \varphi(\mathrm{top}_m(p_t^k))))$, and Asymmetric Knowledge Modulation with a quality-relief coefficient $\rho_t = 1-\beta_p s_t$ and a detection-boost coefficient $\eta_t = 1+\beta_d(1-s_t)$. The composite factor $\mu_t = \rho_t \eta_t$ scales the host watermark strength, and because it correlates with $1-s_t$ on high-entropy positions, it concentrates signal on weakly grounded tokens while keeping knowledge-anchored tokens close to their natural form. The paper reports consistent improvements for exponential, linear, and logarithmic MorphMark strength functions on OPT-1.3B over its own reproductions, with the linear variant showing the largest gain under synonym replacement.

Load-bearing premise

The load-bearing premise is that the internally reproduced MorphMark baselines are faithful implementations of the host watermarks; the reproduction numbers fall below the original published values, so the reported gains could shrink if those baselines are too weak.

Editorial extensions

If this is right

  • A placement signal based on external knowledge rather than local entropy can be shared across watermark families that otherwise use different signal constructions.
  • The quality versus detectability trade-off can be eased in both directions at once: anchored tokens are perturbed less, while unanchored positions carry a stronger signal, so attacked detection rises as perplexity falls.
  • The host detector stays unchanged, so existing deployment and verification pipelines could adopt WorldMark by modifying only decoding-time modulation.
  • The paper claims the improvements carry over to LLaMA-3-8B, Mistral-7B, TriviaQA, and NaturalQuestions, and that the benefit widens on knowledge-intensive datasets where entity-bearing tokens carry greater semantic weight.

Reading between the lines

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

  • If the mechanism generalizes, the same saliency-guided placement could be combined with stronger attacks such as LLM-based paraphrase or translation round-tripping: attackers would be forced to rewrite anchored factual tokens, which should make the watermark harder to erase than under uniform placement.
  • The ablations suggest that knowledge similarity outperforms random and entropy-based saliency, implying that semantic grounding carries information beyond local uncertainty; a direct test would swap in other knowledge sources, such as retrieval-augmented passages, to see whether the saliency signal rather than the graph structure is what matters.
  • Because the memory graph is built only from the prompt and already observed text, the interface could in principle be deployed in black-box settings where the watermarking party controls decoding but not the model weights or detector; a direct test would be to attach WorldMark to a hosted API generation endpoint.
  • The use of a small 22.7M-parameter encoder for saliency suggests that knowledge-aware placement could be added to existing watermarking pipelines with only modest auxiliary compute at generation time, though whether this overhead stays negligible on slower or larger configurations remains an open question.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes WorldMark, a plug-and-play interface that retrieves semantic and episodic knowledge from a World Knowledge Memory graph, computes a token-level knowledge saliency score, and adjusts the strength of a host watermark via Asymmetric Knowledge Modulation without changing the host detector. The primary experiments on C4 with OPT-1.3B compare three MorphMark host variants with and without WorldMark, reporting gains in clean and attacked detection and small perplexity reductions relative to internally reproduced MorphMark baselines. Extended evaluations on larger backbones and QA datasets, a scaled cross-family study, and a small pilot that injects memory directly without modulation are also presented. The paper includes ablations, hyperparameter tables, seed-level results, and paired t-tests.

Significance. If the reported effects are real, WorldMark is a useful contribution: it is detector-preserving, host-agnostic, and shows a simultaneous improvement in robustness and perplexity, a direction of genuine interest in LLM watermarking. The paper is methodologically careful in several ways: validation, threshold calibration, and test data are disjoint; five seed runs with paired t-tests are reported; and the ablations distinguish knowledge content, retrieval quality, saliency source, and modulation mechanism, including negative controls such as shuffled retrieval and random saliency. There is no obvious circularity: the detector is unchanged and the saliency score is derived from the model's own distribution. The main uncertainty is empirical: the central claim is defined as a delta over reproduced baselines that differ substantially from the originally published MorphMark values.

major comments (3)
  1. [§4.1, Tables 1–2, Eqs. (8)–(9)] The headline result of an average +0.0296 robust TPR gain is a delta against the internally reproduced MorphMark baselines, and those baselines are not at the operating point of the published host watermark. In Table 1, MorphMark-exp robust TPR@1% is 0.9000 (Repro) vs 0.9600 (Paper), MorphMark-log is 0.8525 vs 0.9375, and the repro PPL values are noticeably lower (10.9404 vs 11.3569 for exp), consistent with a lower-strength regime. Because AKM multiplies the host strength by μ_t (Eqs. 8–9), a low-strength baseline can be boosted trivially, and for exp and log the +WorldMark values (0.9119 and 0.8800) still do not reach the published MorphMark values. The central claim therefore depends on these reproductions being faithful implementations at a matched operating point. Please provide either (i) a reproduction configuration whose robust TPR and PPL match the published MorphMark values, (ii) a direct comparison of +WorldMark against the published values, or (iii) released code and the exact attack/evaluation configuration so the operating point can be verified. The paper's own statement that 'reproduced baselines do not always match' (§4.2) makes this issue explicit rather than hypothetical.
  2. [Figure 2, Tables 1] Figure 2 contains numerical values that contradict Table 1, which is the paper's main results table. In panel (a), for example, MorphMark-exp is shown as 87.2/87.9/92.1 for Paper/Repro/+WorldMark, while Table 1 reports robust TPR@1% of 0.9600/0.9000/0.9119; for KGW the figure shows 46.6/45.6 whereas Table 1 reports 0.8050/0.6775. Panel (b) similarly reports values such as 59.8/59.2/66.2 for MorphMark-exp robust Best F1 while Table 1 reports 0.9778/0.9672/0.9783. The figure therefore does not illustrate the results it is intended to present. Please reconcile the figure with Table 1 or remove it.
  3. [§4.3, Table 2, Abstract] The claim that every monitored metric moves in the favorable direction is overstated. For the log variant, ΔTPR@1% is 0.0000, and for exp and linear the clean TPR gains are only +0.0025 and +0.0050 on already near-saturated values (0.9975–1.0000). The abstract's statement that WorldMark 'improves clean and attacked detection' should be qualified, or the clean TPR differences should be shown to be statistically significant. The robust metrics, which show larger and significant deltas, are the appropriate focus of the paper's contribution.
minor comments (4)
  1. [Table 2] The caption states 'All checked entries are improvements,' but the log variant's ΔTPR@1% is 0.0000; use 'non-decreasing' or qualify the statement.
  2. [Appendix A, Tables 3–4] The extended evaluations claim to 'confirm' transfer to LLaMA-3-8B, Mistral-7B, TriviaQA, and NaturalQuestions, but these tables report only robust TPR and Robust F1 without standard deviations, confidence intervals, significance tests, or perplexity values; consider describing these as consistent exploratory evidence rather than confirmation.
  3. [§3.3, Table 1] The claim of no measurable latency penalty is based on wall-clock differences of about 0.01–0.16 s that are within normal runtime variation, but the mechanism is stated as GPU asynchrony; please report the isolated per-step Sentence-BERT inference time and clarify how it is hidden in batch size 1.
  4. [Algorithm 1, §3.3] In Algorithm 1 the notation 'topm(p^k_t)' should be 'top_m(p^k_t)' to denote the set of top m tokens, and the loop bound T should be defined in the algorithm or by reference to the generation-length protocol.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: WorldMark's modulation is an empirically evaluated interface on held-out data, not a derivation that reduces to its inputs.

full rationale

The paper's derivation chain is explicit and non-circular: WKM retrieval (Eqs. 1-3) produces a knowledge context c0; the saliency score (Eq. 5) is a cosine similarity between the embedding of c0 and the embedding of the top-m tokens of the pre-watermark distribution pk_t; AKM (Eqs. 6-8) converts the saliency into a multiplicative factor μ_t; and the host watermark strength is modulated by μ_t (Eqs. 9, 13, 14) while the detector remains the original z-score (Eq. 12). None of these quantities is defined in terms of the target detection outcome, and no parameter is fitted to the reported test set. The hyperparameters (λ, m, β_p, β_d, d, w, k) are tuned on a disjoint 50-sample validation set and then fixed for the 400-sample test evaluation, so the reported improvements are measured rather than forced by construction. The ablation study (Table 9) provides appropriate negative controls—shuffled retrieval, irrelevant context, random saliency, and entropy saliency all underperform the full method—showing that the knowledge signal carries information beyond the modulation mechanism itself. The paper's acknowledged limitations, such as 'The reproduced baselines do not always match the values reported in the original papers' (Section 4.2) and the small pilot sample sizes (Appendix C), are correctness and baseline-fidelity risks, not circularity. There is also no load-bearing self-citation: the host MorphMark and the AriGraph memory are external prior works with disjoint authorship. Overall, the derivation is self-contained and the central claim is an empirical, held-out comparison.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim rests on an empirical heuristic (saliency from embedding similarity) and on the fidelity of internally reproduced baselines. No new physical entities are posited; World Knowledge Memory and the saliency score are internal components, not independently evidenced entities. The main free parameters are the tuned saliency and modulation coefficients, all selected on a 50-sample validation set.

free parameters (6)
  • lambda (saliency sharpness) = 5.0
    Tuned by grid search on 50 held-out C4 samples; controls the steepness of the sigmoid in Eq. (5).
  • m (top-m candidates) = 20
    Tuned on 50 C4 samples; number of candidate tokens used in the saliency computation.
  • beta_p (quality-relief coefficient) = 0.3
    Tuned on 50 C4 samples; scales the relief term rho_t = 1 - beta_p * s_t in Eq. (6).
  • beta_d (detection-boost coefficient) = 0.3
    Tuned on 50 C4 samples; scales the boost term eta_t = 1 + beta_d * (1 - s_t) in Eq. (7).
  • d, w, k (retrieval depth, width, episodic vertices) = 2, 5, 10
    Selected by grid search over {1,2,3} x {3,5,10} x {5,10,20} on 50 C4 samples; used in the WKM retrieval of Eq. (2).
  • hybrid gate coefficients kappa_alpha, kappa_beta = not reported
    Used in Eq. (14) to shift entropy gates for the hybrid host, but no values or tuning protocol are reported anywhere in the paper.
assumptions (5)
  • domain assumption Knowledge saliency s_t = sigma(lambda * cos_sim(phi(c0), phi(top_m(p_k_t)))) is a valid proxy for which tokens are semantically anchored and safe to watermark.
    This is the load-bearing modeling assumption in Eq. (5). The paper itself states the score is not a factuality judgment and gives no formal guarantee; only indirect ablation support is provided.
  • domain assumption The WKM extraction and retrieval pipeline (LLaMA-3-8B fact extraction, Sentence-BERT semantic search, AriGraph episodic memory) returns knowledge that is relevant and mostly new relative to the prompt.
    Section 3.2 and Appendix D. The only reported quality statistic is that 68.3% of retrieved triples are not verbatim in the prompt; there is no evaluation of fact correctness or retrieval relevance.
  • domain assumption The reproduced MorphMark and other host watermark baselines in Table 1 are faithful implementations of the original methods.
    The paper evaluates against its own 'Repro' numbers rather than the 'Paper' numbers because they differ substantially; the fidelity of these reproductions is not independently verified.
  • standard math Prompt-only conditioning with a fixed knowledge context c0 preserves causal decoding and does not leak future-token information.
    Stated in Section 3.2; retrieval happens once before generation using only x and y_obs, and c0 is not refreshed per token.
  • ad hoc to paper The asymmetric modulation forms rho_t = 1 - beta_p * s_t and eta_t = 1 + beta_d * (1 - s_t) are appropriate functional choices.
    Eqs. (6) and (7) are introduced without derivation and are justified only by the ablation results in Appendix E.

how reviews work

0 comments
Cite this review

Pith. "Pith review of WorldMark: A Plug-and-Play World Knowledge Interface for Cross-Host Language Model Watermarking." pith.science (2026). https://pith.science/paper/ZCTRUDWW

@misc{pith2026260806416,
  author       = {Pith},
  title        = {Pith review of: WorldMark: A Plug-and-Play World Knowledge Interface for Cross-Host Language Model Watermarking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZCTRUDWW}},
  note         = {Machine review of arXiv:2608.06416}
}
read the original abstract

Watermarking traces the provenance of text produced by large language models by embedding statistically detectable signals during decoding. Existing schemes fall into logits-based, sampling-based, entropy-aware, and adaptive-strength families, yet all of them place watermark signals according to local token statistics. In the open-ended text-generation settings evaluated in this work, local statistics may provide insufficient guidance for placing robust watermark signals. We introduce WorldMark, a plug-and-play interface that uses World Knowledge Memory (WKM) to organize semantic and episodic knowledge in a memory graph, converts the retrieved knowledge into a token-level knowledge saliency score, and adjusts the strength of a host watermark through Asymmetric Knowledge Modulation (AKM). WorldMark requires no backbone retraining and introduces no additional detector-side model or parameter. On the primary C4 evaluation, the complete WorldMark interface improves clean and attacked detection across three adaptive-strength host variants while slightly reducing perplexity. Additional pilot experiments on C4 and OpenGen show that direct memory conditioning transfers across multiple watermark families but can be unstable without saliency-aware modulation. WorldMark requires no additional detector-side model or parameter and introduces negligible overhead under the primary protocol.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 14 canonical work pages

  1. [1]

    A watermark for large language models.CoRR, abs/2301.10226, 2023

    John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, and Tom Goldstein. A watermark for large language models.CoRR, abs/2301.10226, 2023

    arXiv 2023

  2. [2]

    Provable robust watermarking for ai-generated text.CoRR, abs/2306.17439, 2023

    Xuandong Zhao, Prabhanjan Ananth, Lei Li, and Yu-Xiang Wang. Provable robust watermarking for ai-generated text.CoRR, abs/2306.17439, 2023

    arXiv 2023

  3. [3]

    Watermarking of large language models

    Scott Aaronson and Hendrik Kirchner. Watermarking of large language models. Talk / OpenAI blog post, 2022. No arXiv posting; see https://scottaaronson.blog/?p=6823

    work page 2022

  4. [4]

    Robust distortion-free watermarks for language models.CoRR, abs/2307.15593, 2023

    Rohith Kuditipudi, John Thickstun, Tatsunori Hashimoto, and Percy Liang. Robust distortion-free watermarks for language models.CoRR, abs/2307.15593, 2023

    arXiv 2023

  5. [5]

    Who wrote this code? watermarking for code generation.CoRR, abs/2305.15060, 2023

    Taehyun Lee, Seokhee Hong, Jaewoo Ahn, Ilgee Hong, Hwaran Lee, Sangdoo Yun, Jamin Shin, and Gunhee Kim. Who wrote this code? watermarking for code generation.CoRR, abs/2305.15060, 2023

    arXiv 2023

  6. [6]

    An entropy-based text watermarking detection method.CoRR, abs/2403.13485, 2024

    Yijian Lu, Aiwei Liu, Dianzhi Yu, Jingjing Li, and Irwin King. An entropy-based text watermarking detection method.CoRR, abs/2403.13485, 2024

    arXiv 2024

  7. [7]

    Invisible Entropy: Towards Safe and Efficient Low-Entropy LLM Watermarking

    Tianle Gu, Zongqi Wang, Kexin Huang, Yuanqi Yao, Xiangliang Zhang, Yujiu Yang, and Xiuying Chen. Invisible entropy: Towards safe and efficient low-entropy LLM watermarking.CoRR, abs/2505.14112, 2025

    work page Pith review arXiv 2025

  8. [8]

    Morphmark: Flexible adaptive watermarking for large language models.CoRR, abs/2505.11541, 2025

    Zongqi Wang, Tianle Gu, Baoyuan Wu, and Yujiu Yang. Morphmark: Flexible adaptive watermarking for large language models.CoRR, abs/2505.11541, 2025

    arXiv 2025

Show all 33 references
  1. [9]

    Sorokin, Dmitry Evseev, Mikhail Burtsev, and Evgeny Burnaev

    Petr Anokhin, Nikita Semenov, Artyom Y. Sorokin, Dmitry Evseev, Mikhail Burtsev, and Evgeny Burnaev. Ari- graph: Learning knowledge graph world models with episodic memory for LLM agents.CoRR, abs/2407.04363, 2024. 15

    2024 arXiv

  2. [10]

    Undetectable watermarks for language models.CoRR, abs/2306.09194, 2023

    Miranda Christ, Sam Gunn, and Or Zamir. Undetectable watermarks for language models.CoRR, abs/2306.09194, 2023

    2023 arXiv

  3. [11]

    Waterbench: Towards holistic evaluation of watermarks for large language models.CoRR, abs/2311.07138, 2023

    Shangqing Tu, Yuliang Sun, Yushi Bai, Jifan Yu, Lei Hou, and Juanzi Li. Waterbench: Towards holistic evaluation of watermarks for large language models.CoRR, abs/2311.07138, 2023

    2023 arXiv

  4. [12]

    Julien Piet, Chawin Sitawarin, Vivian Fang, Norman Mu, and David A. Wagner. Mark my words: Analyzing and evaluating language model watermarks.CoRR, abs/2312.00273, 2023

    2023 arXiv

  5. [13]

    Markllm: An open-source toolkit for LLM watermarking.CoRR, abs/2405.10051, 2024

    Leyi Pan, Aiwei Liu, Zhiwei He, Zitian Gao, Xuandong Zhao, Yijian Lu, Binglin Zhou, Shuliang Liu, Xuming Hu, Lijie Wen, and Irwin King. Markllm: An open-source toolkit for LLM watermarking.CoRR, abs/2405.10051, 2024

    2024 arXiv

  6. [14]

    Task-agnostic language model watermarking via high entropy passthrough layers

    Vaden Masrani, Mohammad Akbari, David Ming Xuan Yue, Ahmad Rezaei, and Yong Zhang. Task-agnostic language model watermarking via high entropy passthrough layers. In Toby Walsh, Julie Shah, and Zico Kolter, editors,Thirty-Ninth AAAI Conference on Artificial Intelligence, Thirty...

    2025

  7. [15]

    Entropy-guided watermarking for llms: A test-time framework for robust and traceable text generation.CoRR, abs/2504.12108, 2025

    Shizhan Cai, Liang Ding, and Dacheng Tao. Entropy-guided watermarking for llms: A test-time framework for robust and traceable text generation.CoRR, abs/2504.12108, 2025

    2025 arXiv

  8. [16]

    Watme: Towards lossless watermarking through lexical redundancy

    Liang Chen, Yatao Bian, Yang Deng, Deng Cai, Shuaiyi Li, Peilin Zhao, and Kam-Fai Wong. Watme: Towards lossless watermarking through lexical redundancy. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors, Proceedings of the 62nd Annual Meeting of the Association for Com...

    2024

  9. [17]

    From trade-off to synergy: A versatile symbiotic watermarking framework for large language models

    Yidan Wang, Yubing Ren, Yanan Cao, and Binxing Fang. From trade-off to synergy: A versatile symbiotic watermarking framework for large language models. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mohammad Taher Pilehvar, editors,Proceedings of the 63rd Annual Meetin...

    2025

  10. [18]

    Robust data watermarking in lan- guage models by injecting fictitious knowledge

    Xinyue Cui, Johnny Tian-Zheng Wei, Swabha Swayamdipta, and Robin Jia. Robust data watermarking in lan- guage models by injecting fictitious knowledge. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mo- hammad Taher Pilehvar, editors,Findings of the Association for Comp...

    2025

  11. [19]

    Reflexion: an autonomous agent with dynamic memory and self-reflection.CoRR, abs/2303.11366, 2023

    Noah Shinn, Beck Labash, and Ashwin Gopinath. Reflexion: an autonomous agent with dynamic memory and self-reflection.CoRR, abs/2303.11366, 2023

    2023 arXiv

  12. [20]

    O’Brien, Carrie J

    Joon Sung Park, Joseph C. O’Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, and Michael S. Bernstein. Generative agents: Interactive simulacra of human behavior.CoRR, abs/2304.03442, 2023

    2023 arXiv

  13. [21]

    From local to global: A graph RAG approach to query-focused summarization.CoRR, abs/2404.16130, 2024

    Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, and Jonathan Larson. From local to global: A graph RAG approach to query-focused summarization.CoRR, abs/2404.16130, 2024

    2024 arXiv

  14. [22]

    Hausknecht, Layla El Asri, Mahmoud Adada, Wendy Tay, and Adam Trischler

    Marc-Alexandre Cˆ ot´ e,´Akos K´ ad´ ar, Xingdi Yuan, Ben Kybartas, Tavian Barnes, Emery Fine, James Moore, Matthew J. Hausknecht, Layla El Asri, Mahmoud Adada, Wendy Tay, and Adam Trischler. Textworld: A learning environment for text-based games. In Tristan Cazenave, Abdallah...

    2018

  15. [23]

    Miller, Roberta Raileanu, Marco Selvatici, Edward Grefenstette, and Tim Rockt¨ aschel

    Heinrich K¨ uttler, Nantas Nardelli, Alexander H. Miller, Roberta Raileanu, Marco Selvatici, Edward Grefenstette, and Tim Rockt¨ aschel. The nethack learning environment. In Hugo Larochelle, Marc’Aurelio Ranzato, Raia Hadsell, Maria-Florina Balcan, and Hsuan-Tien Lin, editors,...

    2020

  16. [24]

    On the reliability of watermarks for large language models.CoRR, abs/2306.04634, 2023

    John Kirchenbauer, Jonas Geiping, Yuxin Wen, Manli Shu, Khalid Saifullah, Kezhi Kong, Kasun Fernando, Anirud- dha Saha, Micah Goldblum, and Tom Goldstein. On the reliability of watermarks for large language models.CoRR, abs/2306.04634, 2023

    2023 arXiv

  17. [25]

    Postmark: A robust blackbox watermark for large language models

    Yapei Chang, Kalpesh Krishna, Amir Houmansadr, John Wieting, and Mohit Iyyer. Postmark: A robust blackbox watermark for large language models. In Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen, editors,Proceedings of the 2024 Conference on Empirical Methods in Natural Langu...

    2024

  18. [26]

    Catmark: A context-aware thresholding framework for robust cross-task watermarking in large language models.CoRR, abs/2510.02342, 2025

    Yu Zhang, Shuliang Liu, Xu Yang, and Xuming Hu. Catmark: A context-aware thresholding framework for robust cross-task watermarking in large language models.CoRR, abs/2510.02342, 2025

    2025

  19. [27]

    Sorokin, and Mikhail Burtsev

    Yuri Kuratov, Aydar Bulatov, Petr Anokhin, Ivan Rodkin, Dmitry Sorokin, Artyom Y. Sorokin, and Mikhail Burtsev. Babilong: Testing the limits of llms with long context reasoning-in-a-haystack. In Amir Globersons, Lester Mackey, Danielle Belgrave, Angela Fan, Ulrich Paquet, Jaku...

    2024

  20. [28]

    Sentence-bert: Sentence embeddings using siamese bert-networks

    Nils Reimers and Iryna Gurevych. Sentence-bert: Sentence embeddings using siamese bert-networks. In Kentaro Inui, Jing Jiang, Vincent Ng, and Xiaojun Wan, editors,Proceedings of the 2019 Conference on Empirical Meth- ods in Natural Language Processing and the 9th International...

    2019

  21. [29]

    Improved bounds for private and robust alignment.CoRR, abs/2512.23816, 2025

    Wenqian Weng, Yi He, and Xingyu Zhou. Improved bounds for private and robust alignment.CoRR, abs/2512.23816, 2025

    2025 arXiv

  22. [30]

    The llama 3 herd of models.CoRR, abs/2407.21783, 2024

    Llama Team. The llama 3 herd of models.CoRR, abs/2407.21783, 2024

    2024 arXiv

  23. [31]

    Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de Las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, L´ elio Renard Lavaud, Marie- Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thom...

    arXiv

  24. [32]

    Weld, and Luke Zettlemoyer

    Mandar Joshi, Eunsol Choi, Daniel S. Weld, and Luke Zettlemoyer. Triviaqa: A large scale distantly supervised challenge dataset for reading comprehension. In Regina Barzilay and Min-Yen Kan, editors,Proceedings of the 55th Annual Meeting of the Association for Computational Li...

    2017

  25. [33]

    Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, Kristina Toutanova, Llion Jones, Matthew Kelcey, Ming-Wei Chang, Andrew M

    Tom Kwiatkowski, Jennimaria Palomaki, Olivia Redfield, Michael Collins, Ankur P. Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, Kristina Toutanova, Llion Jones, Matthew Kelcey, Ming-Wei Chang, Andrew M. Dai, Jakob Uszkoreit, Quoc Le, and S...

    2019

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.