Pith. sign in

REVIEW 4 major objections 6 minor 36 references

HiCaM: A Hierarchical-Causal Modification Framework for Long-Form Text Modification

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Structuring a document as an entity summary tree plus a causal graph before editing makes LLMs revise long texts more faithfully than direct prompting, with win rates up to 79.50%.

desk verdict Useful training-free editing idea, but the win rates are uninterpretable until the scaffolding model is disclosed and the evaluation is anchored by human judgment. read the letter →

arxiv 2505.24319 v1 pith:N4UKKO34 submitted 2025-05-30 cs.CL

Yuntao Shi , Yi Luo , Yeyun Gong , Chen Lin This is my paper
classification cs.CL
keywords long-formtextmodificationhierarchicalsummarytreecausalgraphentity-centricstructureLLM-as-judgeevaluationmulti-domainbenchmarktraining-freeframeworkcontrolledgeneration
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 large language models fail at long-form text modification in two predictable ways—rewriting or summarizing content that should stay untouched, and missing updates to passages only implicitly connected to the requested change—and that both failures can be fixed without any retraining. The fix, called HiCaM, pre-structures a document along two axes before any editing happens: a hierarchical summary tree that maps every text region to the entities it concerns, and a causal graph of directed influence relations among the key entities. Edits are then planned node by node in the tree, while the causal graph forces each change to propagate through every passage it logically touches, so untouched regions stay verbatim. On a new multi-domain benchmark assembled from seven existing long-text corpora, the paper reports that four LLMs (GPT-4o, GPT-4o-mini, Deepseek-V3, QWQ-32B) augmented with HiCaM win pairwise comparisons against direct prompting at rates up to 79.50%. The claim matters because over-editing and under-editing are exactly the failure modes that make LLM editing of stories, protocols, and manuscripts impractical today.

What carries the argument

The load-bearing objects are two analytical structures built before generation. The hierarchical entity-oriented summary tree is a recursive decomposition of the text in which each node covers a span (marked by opening and closing phrases), carries a summary focused on the key entities, and either proposes a modification for its span or is omitted entirely; its job is to confine edits to relevant regions so that irrelevant content survives verbatim. The causal graph is a set of directed edges among entities, each edge labeled with the kind of influence ('causes,' 'depends on,' 'affects'), built chunk-by-chunk and merged into a global graph; its job is to convert implicitly relevant passages into explicitly tracked dependencies so that a change to one entity propagates through all downstream relations. The third component is the tree-structured modification suggestion itself, the joint artifact that turns the two structures into a concrete editing plan for the final pass.

What would settle it

Run a human preference study on a few hundred of the benchmark's comparison pairs: with readers blind to which output came from which method, compare their choices with GPT-4o's verdicts—if agreement is near chance, the reported win rates measure the judge's tastes rather than editing quality. A complementary check targets the mechanism directly: because HiCaM promises to leave irrelevant spans untouched, measure the fraction of unedited regions kept byte-for-byte in HiCaM outputs versus direct-prompt outputs.

Watch

Extended reading notes

Core claim

The central claim is that the two observed failure modes of long-form text modification are structural, not parametric: the model does not know which regions are safe and which passages depend on one another. HiCaM supplies both pieces of knowledge explicitly. It first extracts the entities named in the modification suggestion, each with an importance score and a description of the intended change, in a chain-of-thought-inspired step. It then builds, in parallel, a hierarchical entity-oriented summary tree, which recursively decomposes the text into entity-anchored sub-sections with summaries and span boundaries, and a causal graph whose directed edges record how entities influence one another across the whole document. Each tree node that needs editing carries its own modification suggestion; nodes that need no change are simply omitted, so untouched regions stay verbatim. The final revision is generated from these tree-structured suggestions with the causal graph consulted to extend every change through the entity relations it triggers. The paper's experiments claim that this pipeline consistently beats direct prompting on faithfulness, logical coherence, and fluency, with the causal graph as the larger contributor on logically dense texts.

Load-bearing premise

Every quality judgment in the experiments comes from a single AI judge, GPT-4o, comparing pairs of outputs, and the paper reports no check of whether human readers would agree with that judge's calls.

Editorial extensions

If this is right

  • Direct prompting of an LLM to 'edit this text' is the wrong baseline for long-form editing; the paper argues that pre-structuring the document around its entities and their causal relations is what separates faithful edits from destructive ones.
  • The two failure modes are separable: the summary tree alone confines edits to relevant regions, and adding the causal graph yields the largest extra gains on texts with dense internal logic, such as narratives, legal documents, and meeting transcripts.
  • The framework is training-free and its largest net gain appears on the weakest model (GPT-4o-mini, 39.40%), so the value of the approach concentrates precisely where baseline editing quality is lowest.
  • The paper ships a reusable eight-dataset, multi-domain benchmark for long-form text modification, with metadata-guided modification suggestions, so later methods can be measured against the same pairs.
  • Ablations attribute most of the win on logically dense texts to the causal graph (an average 20.74-point win-rate drop without it), while weakly structured datasets lose little, suggesting the graph's role varies by input.

Reading between the lines

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

  • The framework's design implies a cheap diagnostic the paper does not run: extract the causal graph alone, show its edges to a human editor, and ask whether those edges identify the passages a careful editor would deem implicitly relevant; strong agreement would isolate the mechanism behind the win rates.
  • The same two structures could support iterative editing sessions: rebuild the summary tree and causal graph after each revision so a user can apply successive changes without the model drifting into over- or under-editing.
  • Because the dataset is built by asking an LLM to generate modification suggestions for existing QA and summarization texts, benchmark difficulty inherits the generator's skill; adding contradiction-inducing or multi-hop suggestions would stress-test whether the causal graph is actually doing propagation rather than merely being present.
  • The evaluation aggregates faithfulness, coherence, and fluency into one judge call; scoring the three criteria separately, and measuring how often unedited regions survive byte-for-byte, would directly test the mechanism the framework claims.
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

4 major / 6 minor

Summary. The paper introduces HiCaM, a training-free framework for long-form text modification. Given an original text and a modification suggestion, HiCaM first extracts key entities, optionally chunks the text, and then builds two entity-centric structures: a hierarchical summary tree whose nodes mark text spans with summaries and local modification instructions, and a causal graph of directional entity relations. The final modified text is generated by prompting an LLM with tree-structured suggestions derived from these structures. The authors also construct an evaluation dataset by taking documents from seven existing benchmarks and generating modification suggestions with an LLM. They report pairwise GPT-4o-judged comparisons of HiCaM-augmented versus direct-prompt baselines for four LLMs, with win rates mostly in the 56-72% range (with one 79.50% cell), and ablation studies varying chunk size, causal graph inclusion, and entity filtering.

Significance. If the empirical claims hold, HiCaM would be a useful plug-and-play contribution: it requires no training, is model-agnostic, and addresses two clearly motivated failure modes (over-editing and missed implicit updates). The dataset, though LLM-generated, is a plausible first resource for a task that currently lacks benchmarks. The causal-graph ablation (Tables 4-5) shows substantial drops on logically structured datasets, which is a nice falsifiable component. However, the current evidence is almost entirely internal: the evaluation set, the modification suggestions, and the quality judgments are all produced by LLMs, with no human validation and no disclosed identity of the scaffolding model. The claimed consistency across models and domains is therefore not yet established at the level claimed.

major comments (4)
  1. [§2.1, §2.2.2, §4.1] The manuscript never identifies which LLM performs entity extraction, tree construction, and causal-graph construction; it only says 'we use a large language model (LLM)' and that a temperature of 0.7 is applied 'to the model.' If a fixed strong model (e.g., GPT-4o) builds the structures for all four baselines, then Table 2 does not compare 'model M + HiCaM' against model M; it compares 'model M + scaffolding-by-GPT-4o + final generation by M' against direct M. The win rates would then partially reflect the scaffolding model's planning ability rather than the framework itself. Please specify the scaffolding model for each baseline, or run the scaffolding with each baseline model, and report both variants.
  2. [§3, §4.1, Appendix E] The evaluation is fully internal. The modification suggestions in the dataset are LLM-generated and never human-validated (Section 3), and the pairwise quality judgments are made by GPT-4o with no human agreement statistics. GPT-4o also serves as a baseline and, as far as can be told, as the scaffolding model, so self-enhancement and style preferences cannot be ruled out by the statement in Appendix E that 'we observe no evidence of bias toward its own outputs' without reporting numbers. Please add a human-annotated sample (e.g., 100-200 pairs across datasets) with inter-annotator agreement, and report judge agreement with GPT-4o.
  3. [Introduction, Abstract, Table 2] The reported win-rate range is internally inconsistent. The Introduction states 'win rates ranging from 56.81% to 72.47%, with net win rates between 13.87% and 59.50%,' but Table 2 reports GPT-4o-mini on MultiFieldQA-zh at 79.50% win rate, and the Abstract claims 'up to a 79.50% win rate.' Please reconcile these statements; as written, the headline ranges exclude a cell in the paper's own main table, which matters because the central claim is 'consistent performance improvements.'
  4. [Table 2 vs Table 1] The evaluation units are undefined and the counts do not match the dataset sizes. For example, Table 1 reports QASPER with 200 documents, but the Deepseek-V3 row in Table 2 sums to 382 pairwise judgments; MultiFieldQA-zh has 200 documents but the GPT-4o-mini row sums to 400 judgments. The paper does not state how many modification suggestions per document were generated or how multiple judgments on the same document were aggregated. Please define the evaluation unit (document, suggestion, or judgment), state whether judgments on the same document are independent, and report confidence intervals or significance tests for the win rates.
minor comments (6)
  1. [Algorithm 1, line 22] The stopping condition uses a threshold τ but never defines its value; please state the default and whether it was tuned.
  2. [Table 13 vs Table 1] Table 13 lists LVEval and LongBench as sources, but Table 1 and Section 3 list a different set of benchmark datasets; clarify the mapping (for instance, is MultiFieldQA drawn from LongBench?).
  3. [Figure 5] The length-ratio analysis is an indirect proxy for undesired modifications; consider reporting direct evidence such as overlap between the spans that were modified and the spans that should have been preserved.
  4. [§4.1] The sentence 'During the pipeline, a temperature value of 0.7 is applied to the model' does not say whether this applies to scaffolding, final generation, or both; please clarify.
  5. [Appendix B.3] The cross-reference 'Table B.2' should point to Table 12; the table is numbered inconsistently.
  6. [§4.1] The term 'net win rate' is used without definition; please define it explicitly (presumably (Win - Lose) / total judgments).

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: HiCaM's win rates are head-to-head outcomes, not fitted inputs; the main risks are evaluation validity, not circularity.

full rationale

HiCaM is a scaffolding/prompting method rather than a fitted model, so there is no parameter that is fitted to a subset of data and then renamed a prediction. The reported win rates are pairwise comparisons between an LLM augmented with HiCaM and the same LLM prompted directly; the framework can lose (e.g., DeepSeek-V3 on QASPER, 56.81% win rate), so the outcome is not forced by construction. The hierarchical summary tree and causal graph are not derived from the evaluation metric; they are generated from the input text and modification suggestions, and the final output is generated separately from the judge. There is no load-bearing self-citation: the chain-of-thought motivation and Graph RAG comparison are external, and no uniqueness theorem is invoked. The main weaknesses are experimental-validity concerns rather than circularity: the scaffolding model is not disclosed (Section 2.1: 'we use a large language model (LLM)'), the judge (GPT-4o) is from the same model family as some augmented systems and is itself a baseline, no human agreement study is reported, and the Introduction's win-rate range (56.81%-72.47%) is inconsistent with Table 2's 79.50% for GPT-4o-mini on MultiFieldQA-zh. These issues bear on whether the empirical claim is convincing, but they do not make the derivation equivalent to its inputs. Accordingly, no circular step is identified.

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

The empirical claim rests on three hand-chosen hyperparameters (top-5 entities, 4096-token chunks, tree depth 1), each selected on the authors' own evaluation set. The evaluation additionally assumes that LLM-generated suggestions define a valid task, that entity-causal structure captures coherence, and that GPT-4o's judgments measure quality without human validation. No new physical or representational entities are postulated beyond the causal graph's edges, which restructure entities already present in the text.

free parameters (3)
  • top-k entity filter = 5
    Chosen by the authors after ablating k in {3, 5, 10, none} on their own evaluation set (Appendix B.1); top-5 gives the best win rate, so the paper selects the optimum of its own metric.
  • chunk size = 4096 tokens
    Default chunk size; ablation (Figures 6-7 and Appendix B.2) shows 2048 and no-limit settings can outperform 4096 on average, so the default is a hand-picked middle value rather than the empirical optimum.
  • hierarchical tree depth limit = 1
    Section 4.1 sets the recursion depth to 1, making the reported 'hierarchical' tree a single level of sub-sections below the root; this hand-set engineering choice directly shapes the method's structure.
assumptions (4)
  • domain assumption Key entities extracted from the modification suggestion, with LLM-assigned importance scores, identify the complete scope of required edits.
    Section 2.1 assumes editing intent decomposes into entities; if entity extraction misses or over-ranks entities, the tree and graph inherit the error.
  • domain assumption Document coherence is captured by entity-level causal relations among the extracted entities.
    Section 2.2.2 models implicit relevance as directed causal edges; this is a modeling assumption about how coherence failures propagate, not an established fact.
  • domain assumption GPT-4o pairwise judgments on faithfulness, logical coherence, and fluency are valid measures of modification quality.
    Section 4.1 and Appendix E rely on this without human agreement validation; it is the weakest premise under the empirical claim.
  • domain assumption LLM-generated modification suggestions on texts from seven benchmarks form a valid evaluation task.
    Section 3 and Appendix D construct the dataset with LLM-generated suggestions and no human quality check of the suggestions themselves.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HiCaM: A Hierarchical-Causal Modification Framework for Long-Form Text Modification." pith.science (2026). https://pith.science/paper/N4UKKO34

@misc{pith2026250524319,
  author       = {Pith},
  title        = {Pith review of: HiCaM: A Hierarchical-Causal Modification Framework for Long-Form Text Modification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N4UKKO34}},
  note         = {Machine review of arXiv:2505.24319}
}
read the original abstract

Large Language Models (LLMs) have achieved remarkable success in various domains. However, when handling long-form text modification tasks, they still face two major problems: (1) producing undesired modifications by inappropriately altering or summarizing irrelevant content, and (2) missing necessary modifications to implicitly related passages that are crucial for maintaining document coherence. To address these issues, we propose HiCaM, a Hierarchical-Causal Modification framework that operates through a hierarchical summary tree and a causal graph. Furthermore, to evaluate HiCaM, we derive a multi-domain dataset from various benchmarks, providing a resource for assessing its effectiveness. Comprehensive evaluations on the dataset demonstrate significant improvements over strong LLMs, with our method achieving up to a 79.50\% win rate. These results highlight the comprehensiveness of our approach, showing consistent performance improvements across multiple models and domains.

Figures

Figures reproduced from arXiv: 2505.24319 by the authors.

Figure 1
Figure 1. Illustration of challenges in LTM: (a) Unde [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the HICAM framework. In Step 3, the framework performs parallel processing: constructing causal graphs for key entities and building hierarchical entity-oriented summary trees, where different colors represent analytical structures in different text segments. These components jointly produce tree-structured modification suggestions, transforming the original text to the modified text while maintaining co… view at source ↗
Figure 3
Figure 3. Example of extracted entities with importance [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Example of recursive processing in building [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Distribution of output lengths across different datasets. The x-axis represents length bins, where outputs [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Win rate for dif￾ferent chunk sizes. 2048 4096 8192 no_limit Chunk Size 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 M e a n N o d e C o u nt [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

36 extracted references · 17 canonical work pages

  1. [1]

    Yushi Bai, Xin Lv, Jiajie Zhang, Hongchang Lyu, Jiankai Tang, Zhidian Huang, Zhengxiao Du, Xiao Liu, Aohan Zeng, Lei Hou, Yuxiao Dong, Jie Tang, and Juanzi Li. 2024. https://doi.org/10.18653/V1/2024.ACL-LONG.172 Longbench: A bilingual, multitask benchmark for long context understanding . In Proceedings of the 62nd Annual Meeting of the Association for Com...

    doi:10.18653/v1/2024.acl-long.172 2024

  2. [2]

    Longwriter: Unleashing 10,000+ word generation from long context llms, 2024

    Yushi Bai, Jiajie Zhang, Xin Lv, Linzhi Zheng, Siqi Zhu, Lei Hou, Yuxiao Dong, Jie Tang, and Juanzi Li. Longwriter: Unleashing 10,000+ word generation from long context llms, 2024. URL https://arxiv. org/abs/2408.07055

    arXiv 2024

  3. [3]

    Josef Dai, Xuehai Pan, Ruiyang Sun, Jiaming Ji, Xinbo Xu, Mickel Liu, Yizhou Wang, and Yaodong Yang. 2024. https://openreview.net/forum?id=TyFrPOKYXw Safe RLHF: safe reinforcement learning from human feedback . In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024 . OpenReview.net

    work page 2024

  4. [4]

    Smith, and Matt Gardner

    Pradeep Dasigi, Kyle Lo, Iz Beltagy, Arman Cohan, Noah A. Smith, and Matt Gardner. 2021. https://doi.org/10.18653/V1/2021.NAACL-MAIN.365 A dataset of information-seeking questions and answers anchored in research papers . In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Te...

    doi:10.18653/v1/2021.naacl-main.365 2021

  5. [5]

    Sumanth Dathathri, Andrea Madotto, Janice Lan, Jane Hung, Eric Frank, Piero Molino, Jason Yosinski, and Rosanne Liu. 2020. https://openreview.net/forum?id=H1edEyBKDS Plug and play language models: A simple approach to controlled text generation . In 8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 20...

    work page 2020

  6. [6]

    Zhang, Han Bao, Hanwei Xu, Haocheng Wang, Haowei Zhang, Honghui Ding, Huajian Xin, Huazuo Gao, Hui Li, Hui Qu, J

    DeepSeek - AI, Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dongjie Ji, Erhang Li, Fangyun Lin, Fucong Dai, Fuli Luo, Guangbo Hao, Guanting Chen, Guowei Li, H. Zhang, Han Bao, Hanwei Xu, Haocheng Wang, Haowei Zhang, Honghui Ding, Hu...

    arXiv doi:10.48550/arxiv.2412.19437 2024

  7. [7]

    Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, Dasha Metropolitansky, Robert Osazuwa Ness, and Jonathan Larson. 2024. From local to global: A graph rag approach to query-focused summarization. arXiv preprint arXiv:2404.16130

    arXiv 2024

  8. [8]

    Xingang Guo, Fangxu Yu, Huan Zhang, Lianhui Qin, and Bin Hu. 2024. https://openreview.net/forum?id=yUxdk32TU6 Cold-attack: Jailbreaking llms with stealthiness and controllability . In Forty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024 . OpenReview.net

    work page 2024

Show all 36 references
  1. [9]

    Xingwei He. 2021. https://doi.org/10.18653/V1/2021.EMNLP-MAIN.681 Parallel refinements for lexically constrained text generation with BART . In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021, Virtual Event / Punta Cana, Domin...

    2021 doi

  2. [10]

    Xinyu Hua and Lu Wang. 2020. https://doi.org/10.18653/V1/2020.EMNLP-MAIN.57 PAIR: planning and iterative refinement in pre-trained transformers for long text generation . In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, EMNLP 2020, Onl...

    2020 doi

  3. [11]

    Luyang Huang, Shuyang Cao, Nikolaus Parulian, Heng Ji, and Lu Wang. 2021. https://doi.org/10.18653/v1/2021.naacl-main.112 Efficient attentions for long document summarization . In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computati...

    2021 doi

  4. [12]

    Aaron Hurst, Adam Lerer, Adam P. Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, Aleksander Madry, Alex Baker - Whitcomb, Alex Beutel, Alex Borzunov, Alex Carney, Alex Chow, Alex Kirillov, Alex Nichol, Alex Paino, Alex ...

    2024 doi arXiv

  5. [13]

    Muhammad Khalifa, Hady Elsahar, and Marc Dymetman. 2021. https://openreview.net/forum?id=jWkw45-9AbL A distributional approach to controlled text generation . In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . OpenReview.net

    2021

  6. [15]

    David Landsman, Jerry Zikun Chen, and Hussain Zaidi. 2022. https://aclanthology.org/2022.findings-aacl.40 Beamr: Beam reweighing with attribute discriminators for controllable text generation . In Findings of the Association for Computational Linguistics: AACL-IJCNLP 2022, Onl...

    2022

  7. [16]

    Shilong Li, Yancheng He, Hangyu Guo, Xingyuan Bu, Ge Bai, Jie Liu, Jiaheng Liu, Xingwei Qu, Yangguang Li, Wanli Ouyang, Wenbo Su, and Bo Zheng. 2024. https://aclanthology.org/2024.findings-emnlp.746 Graphreader: Building graph-based agent to enhance long-context abilities of l...

    2024

  8. [17]

    OpenAI. 2023. https://doi.org/10.48550/ARXIV.2303.08774 GPT-4 technical report . CoRR, abs/2303.08774

    2023 doi arXiv

  9. [18]

    Richard Yuanzhe Pang, Alicia Parrish, Nitish Joshi, Nikita Nangia, Jason Phang, Angelica Chen, Vishakh Padmakumar, Johnny Ma, Jana Thompson, He He, and Samuel R. Bowman. 2022. https://doi.org/10.18653/V1/2022.NAACL-MAIN.391 Quality: Question answering with long input texts, ye...

    2022 doi

  10. [19]

    Shanghaoran Quan, Tianyi Tang, Bowen Yu, An Yang, Dayiheng Liu, Bofei Gao, Jianhong Tu, Yichang Zhang, Jingren Zhou, and Junyang Lin. 2024. https://doi.org/10.48550/ARXIV.2410.23933 Language models can self-lengthen to generate long texts . CoRR, abs/2410.23933

    2024 doi arXiv

  11. [20]

    Angela Ramirez, Kartik Agarwal, Juraj Juraska, Utkarsh Garg, and Marilyn A. Walker. 2023. https://doi.org/10.18653/V1/2023.SIGDIAL-1.32 Controllable generation of dialogue acts for dialogue systems via few-shot response generation and ranking . In Proceedings of the 24th Meeti...

    2023 doi

  12. [21]

    Damith Chamalke Senadeera and Julia Ive. 2022. https://doi.org/10.48550/ARXIV.2212.02924 Controlled text generation using T5 based encoder-decoder soft prompt tuning and analysis of the utility of generated text in AI . CoRR, abs/2212.02924

    2022 doi arXiv

  13. [22]

    Qwen Team. 2025. https://qwenlm.github.io/blog/qwq-32b/ Qwq-32b: Embracing the power of reinforcement learning

    2025

  14. [23]

    Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022. https://doi.org/10.1162/TACL\_A\_00475 musique: Multihop questions via single-hop question composition . Trans. Assoc. Comput. Linguistics, 10:539--554

    2022 doi

  15. [24]

    Hao Wang and Lei Sha. 2024. https://doi.org/10.48550/ARXIV.2402.04160 Harnessing the plug-and-play controller by prompting . CoRR, abs/2402.04160

    2024 doi arXiv

  16. [25]

    Yidong Wang, Qi Guo, Wenjin Yao, Hongbo Zhang, Xin Zhang, Zhen Wu, Meishan Zhang, Xinyu Dai, Min Zhang, Qingsong Wen, Wei Ye, Shikun Zhang, and Yue Zhang. 2024. http://papers.nips.cc/paper\_files/paper/2024/hash/d07a9fc7da2e2ec0574c38d5f504d105-Abstract-Conference.html Autosur...

    2024

  17. [26]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. 2022. http://papers.nips.cc/paper\_files/paper/2022/hash/9d5609613524ecf4f15af0f7b31abca4-Abstract-Conference.html Chain-of-thought prompting elicits reasoning...

    2022

  18. [27]

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Me...

    2024 arXiv

  19. [28]

    Dian Yu, Zhou Yu, and Kenji Sagae. 2021. https://doi.org/10.18653/V1/2021.FINDINGS-EMNLP.194 Attribute alignment: Controlling text generation from pre-trained language models . In Findings of the Association for Computational Linguistics: EMNLP 2021, Virtual Event / Punta Cana...

    2021 doi

  20. [29]

    Yongcheng Zeng, Guoqing Liu, Weiyu Ma, Ning Yang, Haifeng Zhang, and Jun Wang. 2024. https://openreview.net/forum?id=1RZKuvqYCR Token-level direct preference optimization . In Forty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 202...

    2024

  21. [30]

    Glass, and Tianxing He

    Jingyu Zhang, James R. Glass, and Tianxing He. 2023. https://doi.org/10.18653/V1/2023.STARSEM-1.27 Pcfg-based natural language interface improves generalization for controlled text generation . In Proceedings of the The 12th Joint Conference on Lexical and Computational Semant...

    2023 doi

  22. [31]

    Yizhe Zhang, Guoyin Wang, Chunyuan Li, Zhe Gan, Chris Brockett, and Bill Dolan. 2020. https://doi.org/10.18653/V1/2020.EMNLP-MAIN.698 POINTER: constrained progressive text generation via insertion-based generative pre-training . In Proceedings of the 2020 Conference on Empiric...

    2020 doi

  23. [32]

    Jun Zhao, Can Zu, Xu Hao, Yi Lu, Wei He, Yiwen Ding, Tao Gui, Qi Zhang, and Xuanjing Huang. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.912 LONGAGENT : Achieving question answering for 128k-token-long documents through multi-agent collaboration . In Proceedings of the 20...

    2024 doi

  24. [33]

    Xing, Hao Zhang, Joseph E

    Lianmin Zheng, Wei - Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric P. Xing, Hao Zhang, Joseph E. Gonzalez, and Ion Stoica. 2023. http://papers.nips.cc/paper\_files/paper/2023/hash/91f18a1287b398d378ef22505bf41832-Abstr...

    2023

  25. [34]

    Ming Zhong, Da Yin, Tao Yu, Ahmad Zaidi, Mutethia Mutuma, Rahul Jha, Ahmed Hassan Awadallah, Asli Celikyilmaz, Yang Liu, Xipeng Qiu, and Dragomir R. Radev. 2021. https://doi.org/10.18653/V1/2021.NAACL-MAIN.472 Qmsum: A new benchmark for query-based multi-domain meeting summari...

    2021 doi

  26. [35]

    Wangchunshu Zhou, Yuchen Eleanor Jiang, Ethan Wilcox, Ryan Cotterell, and Mrinmaya Sachan. 2023. https://proceedings.mlr.press/v202/zhou23g.html Controlled text generation with natural language instructions . In International Conference on Machine Learning, ICML 2023, 23-29 Ju...

    2023

  27. [36]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  28. [37]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.