Pith. sign in

REVIEW 3 major objections 4 minor 64 references

CraftAlign: Feature-Grounded Evaluation and Revision Guidance for AI Stories

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read CraftAlign claims that AI stories become more human-like when revision is driven by energy-selected changes in a 304-feature writing profile, and its targeted guidance beats generic rewriting in both model-space and human tests.

desk verdict Solid evaluation half, plausible revision half, but the headline claim rests on a small descriptive human study rather than Table 2. read the letter →

arxiv 2608.01377 v1 pith:D6INW6AE submitted 2026-08-02 cs.AI

Yang Yang , Boyun Xu , Shaofeng Liang , Yun Han , Zining Zhong , Songning Lai , Kaishen Yuan , Yutao Yue This is my paper · ORCID
classification cs.AI
keywords AIstorytellingstoryrevisionguidancewriting-patternevaluationenergy-basedmodelsfeature-groundednarrativefeatureshuman-likenessLLM
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 that the formulaic 'AI flavor' of machine-written fiction can be reduced by treating an explicit 304-feature writing profile as a space for revision, not just diagnosis. CraftAlign first predicts those features for a story, then uses a prompt-optional class-conditional energy model to score how Human-like versus AI-like the feature configuration is, and finally searches for small schema-valid feature changes that lower the Human/AI energy gap. The chosen changes are rendered as natural-language instructions, and a separate LLM editor rewrites the full story with them. The authors report that this targeted guidance beats generic rewriting and count-matched random guidance across three editor models, and that human readers select the CraftAlign revisions far more often than the original or baseline revisions, while still preferring authentic human-authored stories. If the claim holds, a feature classifier plus energy search is enough to turn evaluation into actionable, story-wide revision guidance.

What carries the argument

The load-bearing object is the shared class-conditional energy function $E_\theta(\hat{\mathbf{z}}, \tilde{e}_p, y)$: a small MLP that scores a predicted 304-feature profile $\hat{\mathbf{z}}$, an optional prompt embedding $\tilde{e}_p$, and a candidate label $y\in\{\mathrm{H}, \mathrm{A}\}$ with a scalar. The difference $D(\hat{\mathbf{z}}, \tilde{e}_p)=E_\theta(\cdot,\mathrm{H})-E_\theta(\cdot,\mathrm{A})$ is the same quantity used to judge Human/AI pattern and to compare revisions. For each single-feature transition $z\to z'$, the Human-directed gain $\Delta D(z\to z'; \tilde{e}_p)=D(z,\tilde{e}_p)-D(z',\tilde{e}_p)$ drives greedy selection; up to five such transitions become natural-lang

What would settle it

Run the same human protocol on a larger held-out sample where guidance is selected to maximize AI-likeness instead of Human-likeness; if readers still prefer those revisions, or if count-matched random guidance matches CraftAlign's selection rate when the sample is large enough, the claim that energy-selected movement toward the Human pattern improves perceived human-likeness collapses.

Watch

Extended reading notes

Core claim

The central discovery, stated on the paper's own terms, is that explicit writing features can serve as a structured intervention space rather than only a diagnostic layer. CraftAlign's local feature estimator reproduces the 304-feature StoryScope profiles well enough that downstream probes retain nearly all Human/AI information (93% Macro-F1 with predicted profiles versus 95% with reference profiles), and its jointly trained energy model reaches 93.86% Macro-F1 when the writing prompt is available. In revision, guidance selected by the largest Human-directed energy gain produces positive mean movement with all three tested editors (average $\Delta D=2.00$ versus $0.24$ for count-matched rand

Load-bearing premise

The whole pipeline stands or falls on whether the 304-feature StoryScope schema, annotated by Gemini-3-Flash rather than human critics, actually captures what makes a story feel human-authored—if the schema misses the real drivers of 'AI flavor,' the energy-guided revisions optimize the wrong target, and the 20-group human study is the only external check.

Editorial extensions

If this is right

  • Feature-grounded guidance transfers across editor models: the same selected transitions improve revisions produced by three different LLM editors, so the method is not tied to one rewrite engine.
  • Predicted profiles, not human reference annotations, are sufficient for deployment: the local estimator preserves nearly all downstream Human/AI signal, making the pipeline applicable to new stories where no reference profile exists.
  • Prompt conditioning strengthens evaluation (0.69 Macro-F1 points) without hurting prompt-free performance, so the same checkpoint can be deployed with or without the original prompt.
  • Randomly selected valid transitions do not help: count-matched random guidance stays near the original in human selection (25.6% vs 24.4%), so targeted selection is the active ingredient.
  • The same energy signal motivates training story-generation models with feature-grounded rewards, a direction the paper explicitly identifies as the next step.

Reading between the lines

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

  • Beyond the paper: the 304-feature space could be used as a controllable stylistic dial—the same energy-guided search, run in reverse, would make a draft more AI-like, which authors might use to imitate or satirize machine-text conventions.
  • Beyond the paper: the evaluation-to-revision recipe is domain-generic; any text domain with an explicit feature taxonomy and a class-conditional energy model could convert a black-box quality judgment into prioritized rewrite instructions, e.g., academic prose or dialogue.
  • Beyond the paper: a natural next experiment is isolating which feature dimensions drive the human preference—ablating single selected transitions would show whether information release, theme explicitness, or causality carries most of the 21-point selection gain.
  • Beyond the paper: the 20-group human study is small; a larger study with more prompts and reviewers would tell whether the observed gap generalizes and whether the human reference advantage (73.8%) can be further closed by longer search budgets.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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. CraftAlign proposes a feature-grounded framework for detecting and revising AI-written stories. It fine-tunes Qwen3.5-9B with 304 feature-specific heads to predict the StoryScope feature profile of a story, then trains a prompt-optional, class-conditional energy model that scores how well a predicted feature configuration matches Human versus AI writing. At inference, the framework fixes Human as target, performs a budgeted greedy search over schema-valid single-feature transitions that maximize the Human-directed energy gain, renders the selected transitions as natural-language instructions from a curated guidance dictionary, and passes them to a general-purpose LLM editor for full-story rewriting. The paper evaluates the framework in three parts: RQ1 measures feature-estimation fidelity; RQ2 measures Human/AI classification with and without the prompt; RQ3 measures revision guidance across three editors in model space and through a small human study. The headline claims are that predicted profiles retain downstream information, the energy model reliably distinguishes Human from AI patterns, and energy-selected targeted guidance outperforms generic rewriting and count-matched random guidance across editors and in human assessment.

Significance. If the central revision claim holds, the paper makes a useful contribution: it demonstrates that an explicit, auditable feature space can serve as an interface between automatic evaluation and full-story revision guidance, rather than stopping at a source label or a holistic score. The design has real strengths. RQ1 and RQ2 are carefully set up with prompt-level splits, out-of-sample predicted profiles, and multiple complementary metrics; the finding that predicted profiles retain most of the reference-profile information is credible and well supported. The paper also deserves credit for explicitly acknowledging the circularity of the model-space revision evaluation and for including a human study as an independent check. The guidance dictionary, the renderability constraints, and the count-matched Random Guidance baseline are thoughtful controls that isolate the value of direction selection. However, the central revision claim currently rests on a model-space analysis that is self-referential and a human study that is small and reported without inferential statistics. The feature-schema premise that the StoryScope LLM-annotated features capture human-likeness is also validated onl

major comments (3)
  1. [RQ3, Table 2, Appendix F] The model-space cross-editor analysis is self-referential in a way that directly affects the abstract's 'across editors' claim. The target set used for Target realized, Non-target drift, and ΔD in Table 2 is the set selected by the same energy model whose D score is the objective being maximized. Random Guidance is evaluated against a target set it never received, so its lower Target realized (47% vs. 55%) partly reflects a mismatch of selection rather than a failure to follow instructions. The paper acknowledges this in §RQ3, but the conclusion still credits the cross-editor evidence. Please either (a) define the target set independently of the search model, e.g., using human-annotated transitions; (b) re-score revisions with a separate evaluator not used during selection; or (c) explicitly restrict the cross-editor claim to the human study. As written, Table 2 cannot independently supp
  2. [Human evaluation, Figure 4, Table 5] The human study is the only independent evidence for the central revision claim, but it is reported without inferential statistics. With 20 prompt groups and 8 reviewers selecting 2 of 5 stories, the chance selection rate is 40%. CraftAlign's 45.6% is only 5.6 points above chance, and the 15.0-point gap to Humanize Only (30.6%) has no confidence interval or p-value. Reviewer-level counts are heterogeneous: R5 chose Humanize 13 times and CraftAlign 10, while R6 chose Random 9 times and CraftAlign 10. The aggregate difference may not be significant once clustered by group. Please report a significance test that accounts for the grouped structure (e.g., bootstrap by prompt group or a mixed-effects logistic regression), and include effect sizes and confidence intervals. Without this, the claim that human readers perceive CraftAlign as more human-like is not established.
  3. [Method / Story Feature Estimation (RQ1) and Energy Model] The target of the entire guidance pipeline is the Human class in the 304-feature StoryScope space, but the reference profiles used to train the feature estimator are produced by Gemini-3-Flash, not by human critics. If this schema or the LLM annotations are biased, then moving predicted features toward 'Human' in this space optimizes the wrong objective. The small human study is the only external validation of this premise. I suggest adding a validation step: show that a sample of StoryScope features correlates with human judgments of human-likeness, or that energy gain predicts human selections beyond the current 20 groups. This is a correctness-risk concern rather than an internal inconsistency; the paper should make the dependency explicit and test it.
minor comments (4)
  1. [Appendix E / Appendix F] The Shared Rewrite Prompt says the revised story must contain 85–115% of the original word count, but Appendix F defines the length-eligibility interval as 0.6–1.4 (60–140%). Please align the two definitions.
  2. [Appendix E / Human evaluation] The planner in Appendix E is referred to as Gemini-3.5-Flash, while the human evaluation section and dataset description refer to Gemini-3-Flash. Clarify whether the planner and the editor are the same model and use consistent naming.
  3. [Table 1] The baseline names 'Profile-only' and 'Prompt-cond.' are not defined in the main text. Add a sentence or a table footnote explaining what these energy specialists are and how they differ from Joint-Zero and Joint-Prompt.
  4. [Figure 3] The left panel reports numeric values for each feature type but the caption does not explain how the percentages are computed. State explicitly that the panel shows average primary scores (F1, κ, Jaccard, or inverse MAE) by feature type.

Circularity Check

1 steps flagged · score 4.0 of 10

Model-space revision metrics reuse the same energy signal that selects the guidance; the paper acknowledges this and supplies a small, descriptive human study as independent support.

  1. self definitional [Method: 'Prompt-Optional Group-Aware Energy Modeling' (ΔD definition) and 'Human-Targeted Structured Search'; evaluated in RQ3, Table 2]
    "For any profile transition z→z′, we define its Human-directed gain as ΔD(z→z′; ep)=D(z,ep)−D(z′,ep). A positive value indicates that the new profile better matches the Human writing pattern under the same condition. This shared quantity is used both to select candidate feature transitions during search and to measure the movement realized after full-story rewriting."

    The energy model D is simultaneously the objective of the search (greedy maximization of ΔD) and the metric of the cross-editor evaluation (mean ΔD, positive-gain rate, H-cross, win rate in Table 2). Any revised story that partially realizes the selected transitions registers positive ΔD because each step was chosen to lower D; Random Guidance, which receives unselected transitions, is scored against goals it was never given. The paper explicitly acknowledges this ('Because this analysis uses the same energy evaluator that selects the guidance...') and relies on a 20-group human study for independent validation, but that study reports only descriptive selection rates without significance tests. Thus the central model-space evidence is partly self-referential, while the independent content

full rationale

CraftAlign's RQ1 and RQ2 are conventional supervised evaluations against StoryScope reference labels and held-out test prompts; they are not circular. The only load-bearing circularity is in RQ3's model-space evaluation, where the ΔD that selects guidance is also used to measure success. This is a genuine self-referential loop. However, the paper does not hide it: it states that the human study is the independent assessment because the cross-editor analysis uses the same evaluator. A human study with 20 groups and eight reviewers does provide external grounding, though its descriptive statistics (no confidence intervals or significance tests) make it weak. Additional evaluation issues—the shared CraftAlign-defined target set penalizes Random Guidance for not realizing instructions it never received—further limit the model-space comparison but are not core circularity. No load-bearing self-citation is present: StoryScope is external work by an unaffiliated group. Overall score 4 reflects partial circularity in the model-space claim, mitigated by an acknowledged but under-powered independent human study.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central claim rests on the adequacy of the StoryScope feature space as annotated by an LLM, on the recovered human-text training data, and on a learned energy model that is used both to select and to measure revisions. These are the main dependencies the reader does not independently verify.

free parameters (4)
  • K_max = 5
    Maximum number of single-feature search steps; chosen by hand and limits how many transitions can be selected.
  • lambda_list = 0.5
    Weight of the listwise loss in the energy-model training objective; chosen by hand and affects within-prompt ranking.
  • epsilon tolerance for target realization = 1e-5
    Tolerance for deciding whether a target feature value is realized; arbitrary and affects reported target-realization rates.
  • length-valid window = 0.6 to 1.4
    Word-count ratio interval used to define the length-valid rate; arbitrary threshold affects one reported metric.
assumptions (4)
  • domain assumption The 304-feature StoryScope schema, annotated by Gemini-3-Flash, captures the writing dimensions that distinguish human from AI stories and that matter for perceived naturalness.
    The whole method searches in this space; if features miss the relevant causes, guidance cannot improve human-likeness.
  • domain assumption Human-authored stories in the StoryScope dataset are a valid target pattern for what AI revisions should emulate.
    Search fixes Human as the target and treats proximity to Human features as improvement; no argument is given that human-like is always better or that these particular human stories are representative.
  • domain assumption The recovered Books3 human texts are faithful reproductions of the original stories used to create the reference profiles.
    The estimator is trained on recovered texts; mismatches would add noise and bias to profile prediction.
  • ad hoc to paper The energy model's Human-versus-AI label, induced from LLM-generated feature profiles, is a serviceable proxy for human-likeness.
    The energy model is a learned classifier, not a validated aesthetic or craft model; it is used for selecting transitions and, in model-space evaluation, for measuring success.
invented entities (1)
  • Guidance dictionary (91 exact transitions plus fallback entries)
    purpose: Maps feature transitions to natural-language rewrite instructions; central to rendering search results into editor guidance.
    It is a hand-curated resource introduced in this paper; only one example is shown, and no validation of the dictionary itself is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CraftAlign: Feature-Grounded Evaluation and Revision Guidance for AI Stories." pith.science (2026). https://pith.science/paper/D6INW6AE

@misc{pith2026260801377,
  author       = {Pith},
  title        = {Pith review of: CraftAlign: Feature-Grounded Evaluation and Revision Guidance for AI Stories},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D6INW6AE}},
  note         = {Machine review of arXiv:2608.01377}
}
read the original abstract

Large language models can now generate fluent and complete stories, yet many outputs still feel formulaic and unnatural because of cliches, over-explanation, linear causal progression, and stereotyped endings, an immediately recognizable AI flavor. Existing detection and evaluation methods often stop at source labels or holistic scores, while revision methods typically target predefined issues through localized edits, limiting their ability to support multiple plausible revision strategies or guide story-wide changes in information release, causal organization, and ending treatment. We introduce CraftAlign, a framework that aligns AI stories with the craft of human storytelling by both assessing Human/AI writing patterns and providing revision guidance. CraftAlign comprises two learned modules and an inference-time guidance pipeline. A feature estimator built on Qwen3.5-9B predicts 304 explicit writing features spanning style and narrative. A class-conditional energy model scores the resulting feature configuration against Human and AI writing patterns, conditioning on the original writing prompt when available. At inference time, CraftAlign applies schema-valid structured perturbations, selects changes that move the feature configuration toward the Human writing pattern, and converts them into natural-language guidance for a separate editor to rewrite the full story. Experiments show that CraftAlign accurately distinguishes Human and AI writing patterns and that its guidance outperforms revision baselines across editors and in a human study.

Figures

Figures reproduced from arXiv: 2608.01377 by the authors.

Figure 1
Figure 1. Comparison between direct Human/AI judgment [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. CraftAlign overview. StoryScope reference profiles supervise only the local feature estimator. Downstream energy [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. RQ1 feature-estimation reliability. Left: primary [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Mean Selection Rates from eight human reviewers. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 50 canonical work pages

  1. [1]

    Cao, Q.; Wang, X.; Yuan, Y.; Liu, Y.; Luo, F.; and Song, R. 2026. Evaluating Text Creativity across Diverse Domains: A Dataset and Large Language Model Evaluator. In Proceedings of the 14th International Conference on Learning Representations

    work page 2026

  2. [2]

    Cao, Z.; Qin, T.; Liu, T.-Y.; Tsai, M.-F.; and Li, H. 2007. Learning to Rank: From Pairwise Approach to Listwise Approach. In Proceedings of the 24th International Conference on Machine Learning, 129--136. ACM

    work page 2007

  3. [3]

    Chakrabarty, T.; Laban, P.; and Wu, C.-S. 2025. Can AI Writing Be Salvaged? Mitigating Idiosyncrasies and Improving Human- AI Alignment in the Writing Process through Edits. In Proceedings of the CHI Conference on Human Factors in Computing Systems

    work page 2025

  4. [4]

    Espinosa Zarlenga, M.; Barbiero, P.; Ciravegna, G.; Marra, G.; Giannini, F.; Diligenti, M.; Shams, Z.; Precioso, F.; Melacci, S.; Weller, A.; Lio, P.; and Jamnik, M. 2022. Concept Embedding Models: Beyond the Accuracy-Explainability Trade-Off. In Advances in Neural Information Processing Systems

    work page 2022

  5. [5]

    Espinosa Zarlenga, M.; Collins, K.; Dvijotham, K.; Weller, A.; Shams, Z.; and Jamnik, M. 2023. Learning to Receive Help: Intervention-Aware Concept Embedding Models. In Advances in Neural Information Processing Systems, volume 36, 37849--37875. Curran Associates, Inc

    work page 2023

  6. [6]

    Fein, D.; Russo, S.; Xiang, V.; Jolly, K.; Rafailov, R.; and Haber, N. 2026. LitBench : A Benchmark and Dataset for Reliable Evaluation of Creative Writing. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), 7740--7755. Rabat, Morocco: Association for Computational Linguistics

    work page 2026

  7. [7]

    Hamilton, S.; Wilkens, M.; and Piper, A. 2026. NarraBench : A Comprehensive Framework for Narrative Benchmarking. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), 3786--3801. Rabat, Morocco: Association for Computational Linguistics

    work page 2026

  8. [8]

    Hans, A.; Schwarzschild, A.; Cherepanova, V.; Kazemi, H.; Saha, A.; Goldblum, M.; Geiping, J.; and Goldstein, T. 2024. Spotting LLM s with Binoculars: Zero-Shot Detection of Machine-Generated Text. In Proceedings of the 41st International Conference on Machine Learning

    work page 2024

Show all 64 references
  1. [9]

    Karimi, A.-H.; Barthe, G.; Sch \"o lkopf, B.; and Valera, I. 2022. A Survey of Algorithmic Recourse: Contrastive Explanations and Consequential Recommendations. ACM Computing Surveys, 55(5)

    2022

  2. [10]

    W.; Nguyen, T.; Tang, Y

    Koh, P. W.; Nguyen, T.; Tang, Y. S.; Mussmann, S.; Pierson, E.; Kim, B.; and Liang, P. 2020. Concept Bottleneck Models. In Proceedings of the 37th International Conference on Machine Learning

    2020

  3. [11]

    E.; and van der Schaar, M

    Laguna, S.; Kobalczyk, K.; Vogt, J. E.; and van der Schaar, M. 2025. Interpretable Reward Modeling with Active Concept Bottlenecks. arXiv:2507.04695

    2025 arXiv

  4. [12]

    LeCun, Y.; Chopra, S.; Hadsell, R.; Ranzato, M.; and Huang, F. J. 2006. A Tutorial on Energy-Based Learning. MIT Press

    2006

  5. [13]

    K.; Sharma, A.; and Tan, C

    Mothilal, R. K.; Sharma, A.; and Tan, C. 2020. Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations. In Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency, 607--617. ACM

    2020

  6. [14]

    Qin, L.; Welleck, S.; Khashabi, D.; and Choi, Y. 2022. COLD Decoding: Energy-Based Constrained Text Generation with Langevin Dynamics. In Advances in Neural Information Processing Systems, volume 35, 9538--9551

    2022

  7. [15]

    Russell, J.; Karpinska, M.; and Iyyer, M. 2025. People Who Frequently Use ChatGPT for Writing Tasks Are Accurate and Robust Detectors of AI -Generated Text. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 5342...

    2025

  8. [17]

    Schr \"o der, T.; Ou, Z.; Li, Y.; and Duncan, A. B. 2024. Energy-Based Modelling for Discrete and Mixed Data via Heat Equations on Structured Spaces. In Advances in Neural Information Processing Systems

    2024

  9. [19]

    Shi, X.; Cao, W.; and Raschka, S. 2023. Deep Neural Networks for Rank-Consistent Ordinal Regression Based on Conditional Probabilities. Pattern Analysis and Applications, 26: 941--955

    2023

  10. [20]

    Shin, S.; Jo, Y.; Ahn, S.; and Lee, N. 2023. A Closer Look at the Intervention Procedure of Concept Bottleneck Models. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, 31504--31520. PMLR

    2023

  11. [21]

    Z.; and Liu, Z

    Sun, M.; Yin, Y.; Xu, Z.; Kolter, J. Z.; and Liu, Z. 2025. Idiosyncrasies in Large Language Models. In Proceedings of the 42nd International Conference on Machine Learning, volume 267 of Proceedings of Machine Learning Research, 57854--57885. PMLR

    2025

  12. [22]

    Ustun, B.; Spangher, A.; and Liu, Y. 2019. Actionable Recourse in Linear Classification. In Proceedings of the Conference on Fairness, Accountability, and Transparency, 10--19

    2019

  13. [23]

    Wachter, S.; Mittelstadt, B.; and Russell, C. 2017. Counterfactual Explanations without Opening the Black Box: Automated Decisions and the GDPR . Harvard Journal of Law & Technology, 31: 841--887

    2017

  14. [24]

    Wang, H.; Xiong, W.; Xie, T.; Zhao, H.; and Zhang, T. 2024. Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts. In Findings of the Association for Computational Linguistics: EMNLP 2024, 10582--10592

    2024

  15. [25]

    Yang, K.; Tian, Y.; Peng, N.; and Klein, D. 2022. Re3: Generating Longer Stories With Recursive Reprompting and Revision. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, 4393--4479. Abu Dhabi, United Arab Emirates: Association for Com...

    2022

  16. [26]

    StoryScope: Investigating Idiosyncrasies in

    Russell, Jenna and Rajendhran, Rishanth and Pham, Chau Minh and Iyyer, Mohit and Wieting, John , year =. StoryScope: Investigating Idiosyncrasies in. 2604.03136 , archivePrefix =

    arXiv

  17. [27]

    2026 , address =

    Fein, Daniel and Russo, Sebastian and Xiang, Violet and Jolly, Kabir and Rafailov, Rafael and Haber, Nick , booktitle =. 2026 , address =. doi:10.18653/v1/2026.eacl-long.362 , url =

    2026 doi

  18. [28]

    Proceedings of the 14th International Conference on Learning Representations , year =

    Evaluating Text Creativity across Diverse Domains: A Dataset and Large Language Model Evaluator , author =. Proceedings of the 14th International Conference on Learning Representations , year =

  19. [29]

    Chakrabarty, Tuhin and Laban, Philippe and Wu, Chien-Sheng , booktitle =. Can. 2025 , doi =

    2025

  20. [30]

    Help Me Write a Story: Evaluating

    Lee, Mina and others , booktitle =. Help Me Write a Story: Evaluating. 2025 , note =

    2025

  21. [31]

    and Finn, Chelsea , booktitle =

    Mitchell, Eric and Lee, Yoonho and Khazatsky, Alexander and Manning, Christopher D. and Finn, Chelsea , booktitle =. 2023 , url =

    2023

  22. [32]

    Proceedings of the 40th International Conference on Machine Learning , year =

    A Watermark for Large Language Models , author =. Proceedings of the 40th International Conference on Machine Learning , year =

  23. [33]

    Spotting

    Hans, Abhimanyu and Schwarzschild, Avi and Cherepanova, Valeriia and Kazemi, Hamid and Saha, Aniruddha and Goldblum, Micah and Geiping, Jonas and Goldstein, Tom , booktitle =. Spotting. 2024 , url =

    2024

  24. [34]

    Proceedings of the 37th International Conference on Machine Learning , year =

    Concept Bottleneck Models , author =. Proceedings of the 37th International Conference on Machine Learning , year =

  25. [35]

    Advances in Neural Information Processing Systems , year =

    Concept Embedding Models: Beyond the Accuracy-Explainability Trade-Off , author =. Advances in Neural Information Processing Systems , year =

  26. [36]

    Findings of the Association for Computational Linguistics: EMNLP 2024 , pages =

    Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts , author =. Findings of the Association for Computational Linguistics: EMNLP 2024 , pages =. 2024 , doi =

    2024

  27. [37]

    2025 , eprint =

    Interpretable Reward Modeling with Active Concept Bottlenecks , author =. 2025 , eprint =

    2025

  28. [38]

    Predicting structured data , booktitle =

    A Tutorial on Energy-Based Learning , author =. Predicting structured data , booktitle =. 2006 , publisher =

    2006

  29. [39]

    Advances in Neural Information Processing Systems , year =

    Energy-Based Modelling for Discrete and Mixed Data via Heat Equations on Structured Spaces , author =. Advances in Neural Information Processing Systems , year =

  30. [40]

    Pattern Analysis and Applications , volume =

    Deep Neural Networks for Rank-Consistent Ordinal Regression Based on Conditional Probabilities , author =. Pattern Analysis and Applications , volume =. 2023 , doi =

    2023

  31. [41]

    Counterfactual Explanations without Opening the Black Box: Automated Decisions and the

    Wachter, Sandra and Mittelstadt, Brent and Russell, Chris , journal =. Counterfactual Explanations without Opening the Black Box: Automated Decisions and the. 2017 , url =

    2017

  32. [42]

    Proceedings of the Conference on Fairness, Accountability, and Transparency , pages =

    Actionable Recourse in Linear Classification , author =. Proceedings of the Conference on Fairness, Accountability, and Transparency , pages =. 2019 , doi =

    2019

  33. [43]

    ACM Computing Surveys , volume =

    A Survey of Algorithmic Recourse: Contrastive Explanations and Consequential Recommendations , author =. ACM Computing Surveys , volume =. 2022 , doi =

    2022

  34. [44]

    Advances in Neural Information Processing Systems , year =

    Learning to Summarize with Human Feedback , author =. Advances in Neural Information Processing Systems , year =

  35. [45]

    Advances in Neural Information Processing Systems , year =

    Training Language Models to Follow Instructions with Human Feedback , author =. Advances in Neural Information Processing Systems , year =

  36. [46]

    Advances in Neural Information Processing Systems , year =

    Direct Preference Optimization: Your Language Model is Secretly a Reward Model , author =. Advances in Neural Information Processing Systems , year =

  37. [47]

    and Zhang, Hao and Gonzalez, Joseph E

    Zheng, Lianmin and Chiang, Wei-Lin and Sheng, Ying and Zhuang, Siyuan and Wu, Zhanghao and Zhuang, Yonghao and Lin, Zi and Li, Zhuohan and Li, Dacheng and Xing, Eric P. and Zhang, Hao and Gonzalez, Joseph E. and Stoica, Ion , booktitle =. Judging. 2023 , url =

    2023

  38. [48]

    and Hatfield-Dodds, Zac and Mann, Ben and Amodei, Dario and Joseph, Nicholas and McCandlish, Sam and Brown, Tom and Kaplan, Jared , year =

    Bai, Yuntao and Kadavath, Saurav and Kundu, Sandipan and Askell, Amanda and Kernion, Jackson and Jones, Andy and Chen, Anna and Goldie, Anna and Mirhoseini, Azalia and McKinnon, Cameron and Chen, Carol and Olsson, Catherine and Hernandez, Danny and Drain, Dawn and Ganguli, Dee...

    arXiv

  39. [49]

    Advances in Neural Information Processing Systems , year =

    Self-Refine: Iterative Refinement with Self-Feedback , author =. Advances in Neural Information Processing Systems , year =

  40. [50]

    2024 , url =

    Lin, Zicheng and Gou, Zhibin and Liang, Tian and Luo, Ruilin and Liu, Haowei and Yang, Yujiu , booktitle =. 2024 , url =

    2024

  41. [51]

    and Xiong, Caiming and Socher, Richard , year =

    Keskar, Nitish Shirish and McCann, Bryan and Varshney, Lav R. and Xiong, Caiming and Socher, Richard , year =

  42. [52]

    International Conference on Learning Representations , year =

    Plug and Play Language Models: A Simple Approach to Controlled Text Generation , author =. International Conference on Learning Representations , year =

  43. [53]

    2021 , url =

    Yang, Kevin and Klein, Dan , booktitle =. 2021 , url =

    2021

  44. [54]

    Proceedings of the 42nd International Conference on Machine Learning , pages =

    Idiosyncrasies in Large Language Models , author =. Proceedings of the 42nd International Conference on Machine Learning , pages =. 2025 , volume =

    2025

  45. [55]

    , year =

    Shaib, Chantal and Chakrabarty, Tuhin and Garcia-Olano, Diego and Wallace, Byron C. , year =. Measuring. 2509.19163 , archivePrefix =

  46. [56]

    The Fourteenth International Conference on Learning Representations , year =

    Death of the Novel(ty): Beyond N-Gram Novelty as a Metric for Textual Creativity , author =. The Fourteenth International Conference on Learning Representations , year =

  47. [57]

    and Dhillon, Paramveer , year =

    Chakrabarty, Tuhin and Ginsburg, Jane C. and Dhillon, Paramveer , year =. Readers Prefer Outputs of. 2510.13939 , archivePrefix =

  48. [58]

    2025 , url =

    Namuduri, Ramya and Wu, Yating and Zheng, Anshun Asher and Wadhwa, Manya and Durrett, Greg and Li, Junyi Jessy , booktitle =. 2025 , url =

    2025

  49. [59]

    2026 , address =

    Hamilton, Sil and Wilkens, Matthew and Piper, Andrew , booktitle =. 2026 , address =. doi:10.18653/v1/2026.eacl-long.176 , url =

    2026 doi

  50. [60]

    People Who Frequently Use

    Russell, Jenna and Karpinska, Marzena and Iyyer, Mohit , booktitle =. People Who Frequently Use. 2025 , month = jul, address =. doi:10.18653/v1/2025.acl-long.267 , url =

    2025 doi

  51. [61]

    Proceedings of the 40th International Conference on Machine Learning , pages =

    A Closer Look at the Intervention Procedure of Concept Bottleneck Models , author =. Proceedings of the 40th International Conference on Machine Learning , pages =. 2023 , volume =

    2023

  52. [62]

    Advances in Neural Information Processing Systems , volume =

    Learning to Receive Help: Intervention-Aware Concept Embedding Models , author =. Advances in Neural Information Processing Systems , volume =. 2023 , publisher =

    2023

  53. [63]

    Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing , pages =

    Re3: Generating Longer Stories With Recursive Reprompting and Revision , author =. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing , pages =. 2022 , address =. doi:10.18653/v1/2022.emnlp-main.296 , url =

    2022 doi

  54. [64]

    Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency , pages =

    Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations , author =. Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency , pages =. 2020 , publisher =. doi:10.1145/3351095.3372850 , url =

    2020

  55. [65]

    2022 , url =

    Qin, Lianhui and Welleck, Sean and Khashabi, Daniel and Choi, Yejin , booktitle =. 2022 , url =

    2022

  56. [66]

    Proceedings of the 24th International Conference on Machine Learning , pages =

    Learning to Rank: From Pairwise Approach to Listwise Approach , author =. Proceedings of the 24th International Conference on Machine Learning , pages =. 2007 , publisher =

    2007

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.