Pith. sign in

REVIEW 3 major objections 3 minor 59 references

Translationese is a rational response to the cognitive difficulty of the translation task and can be partly predicted from measures of that difficulty.

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

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

T0 review · grok-4.5

2026-07-14 22:28 UTC pith:Y37OOXIM

load-bearing objection We only have the abstract for the translationese claim; the supplied full text is a different paper (h-transform visual generation), so the operationalizations cannot be checked. the 3 major comments →

arxiv 2603.12050 v2 pith:Y37OOXIM submitted 2026-03-12 cs.CL

Translationese as a Rational Response to Translation Task Difficulty

Maria Kunilovskaya This is my paper
classification cs.CL
keywords translationesecognitive loadtranslation difficultyLLM surprisalcross-lingual transferEnglish-Germansyntactic complexitytranslation entropy
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Translations systematically differ from texts originally written in the target language—a pattern called translationese. The authors argue this is not merely a set of fixed production habits or socio-cultural effects, but a rational response to the cognitive load of the translation task itself. They measure translationese with an automatic classifier that scores how “translated” each segment looks, and they measure task difficulty as the combination of source-text complexity and cross-lingual transfer difficulty, using mainly LLM surprisal and related information-theoretic quantities, plus standard syntactic and semantic features. On a bidirectional English–German corpus of both written and spoken material, these difficulty measures partly explain the translatedness scores, especially for English-to-German. Cross-lingual transfer difficulty usually outweighs source complexity, and source-text syntactic complexity together with translation-solution entropy are the strongest predictors across directions and modes.

Core claim

Observable translationese, operationalised as a segment-level translatedness score from an automatic classifier, can be partly predicted from quantifiable translation-task difficulty that comprises source-text complexity and cross-lingual transfer difficulty. The link is clearer for English-to-German than the reverse; transfer difficulty contributes more than source complexity in most settings; and source-text syntactic complexity plus translation-solution entropy are the most consistent predictors across language pairs and written/spoken modes.

What carries the argument

Translation task difficulty, defined as source-text complexity plus cross-lingual transfer difficulty and measured chiefly by LLM-surprisal (information-theoretic) metrics together with syntactic and semantic alternatives; this difficulty is used to predict the automatic classifier’s segment-level translatedness score.

Load-bearing premise

The automatic classifier’s translatedness score truly indexes cognitive-load-driven translationese, and LLM surprisal plus translation-solution entropy truly measure the cognitive difficulty of translating rather than shared surface cues or model artefacts.

What would settle it

If, after matching or controlling for genre, register and surface features that both the classifier and the surprisal metrics can exploit, the difficulty predictors no longer account for variance in human ratings of cognitive effort or of “how translated” a segment feels, the central claim fails.

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

If this is right

  • Segments whose sources are syntactically complex or whose possible translations have high entropy should show stronger translationese.
  • Cross-lingual transfer difficulty, more than raw source complexity, should drive divergence from original target-language norms.
  • Information-theoretic surprisal features can stand in for traditional syntactic features in written translation but not necessarily in spoken mode.
  • Directionality matters: English-to-German yields clearer difficulty–translationese links than German-to-English.
  • Quality estimation or post-editing tools can use these difficulty predictors to anticipate where translationese (and related effort) will appear.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If cognitive load is the common cause, interventions that lower source complexity or ease transfer (glossaries, constrained decoding, better MT suggestions) should reduce classifier-detected translationese.
  • The same difficulty metrics should also predict process measures such as pauses, fixations or post-editing time if the load account is general.
  • Spoken mode’s lack of benefit from information-theoretic features suggests real-time production constraints can override surprisal-based difficulty.
  • Without independent human validation of both the classifier and the surprisal measures, the predictive link risks being partly circular through shared surface cues.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The manuscript (as represented by its abstract for arXiv:2603.12050) claims that translationese is a rational response to cognitive load in the translation task. Translationese is operationalised as segment-level translatedness scores from an automatic classifier; task difficulty is split into source-text complexity and cross-lingual transfer difficulty, measured mainly by LLM surprisal and related information-theoretic quantities, with syntactic and semantic alternatives. On a bidirectional English–German corpus (written and spoken), the authors report that difficulty partly predicts translatedness, more strongly for English→German; transfer difficulty usually outweighs source complexity; information-theoretic features match or beat traditional ones in written mode only; and source syntactic complexity plus translation-solution entropy are the strongest predictors across pairs and modes.

Significance. If the operationalisations are valid and the partial predictive relationships hold with adequate effect sizes and controls, the paper would offer a unified, testable cognitive-load account of translationese that links translation studies to information-theoretic difficulty measures. That would matter for translation-process research, MT evaluation, and bilingual production models. The abstract alone, however, does not establish those conditions, and the supplied full-text body is an unrelated computer-vision paper (Weighted h-Transform Sampling, arXiv:2603.12057), so the claimed contribution cannot be verified from the materials provided for review.

major comments (3)
  1. The full manuscript text supplied for review is not the translationese paper: it is “Coarse-Guided Visual Generation via Weighted h-Transform Sampling” (arXiv:2603.12057). None of the classifier training, surprisal definitions, regression specifications, corpus construction, or EN–DE written/spoken results exist in the provided body. Load-bearing claims (partial explanation, relative contribution of transfer vs source difficulty, superiority of syntactic complexity and translation-solution entropy) therefore cannot be checked. Review of the central claim is blocked until the correct full text is provided.
  2. From the abstract alone: the claim that translatedness is a rational response to cognitive load rests on two operationalisations whose independence is not shown—(i) an automatic translatedness classifier and (ii) LLM-surprisal / translation-solution entropy difficulty metrics. If both are driven by shared surface or model-based cues (syntax, register, LLM-like distributions), the reported prediction can be circular. The abstract does not report independence tests, residualisation, or out-of-model controls; these are required for the causal/cognitive interpretation.
  3. Abstract results are only directional (“partly explained,” “especially English-to-German,” “for most experiments”). Without effect sizes, R² / variance explained, confidence intervals, or model comparison against genre/register baselines, it is impossible to judge whether the partial explanation is scientifically meaningful or merely statistically detectable. This is load-bearing for the unified-account claim.
minor comments (3)
  1. Abstract does not name the classifier architecture, training data, or how translatedness scores are calibrated against human judgments of translationese.
  2. “Translation-solution entropy” is listed as a strongest predictor but is not defined in the abstract; a one-sentence operational definition would help readers assess construct validity.
  3. Bidirectional EN–DE design is a strength; the abstract should state whether spoken and written subcorpora are balanced for domain and whether mode is modelled as a fixed effect or analysed separately only.

Circularity Check

0 steps flagged

No verifiable circularity: abstract treats translatedness and difficulty as separate operationalizations; full text is a mismatched CV paper so no derivation chain can be checked.

full rationale

The translationese abstract claims an empirical prediction: segment-level translatedness (automatic classifier) is partly explained by independently operationalized task difficulty (source-text complexity plus cross-lingual transfer, mainly LLM surprisal, plus syntactic/semantic alternatives). Difficulty is not defined as translatedness, nor is a parameter fitted to translatedness then re-reported as a prediction of the same quantity. No equations, classifier training details, surprisal definitions, or regression specs appear in the supplied materials. The CACHEABLE full manuscript is an unrelated paper (Weighted h-Transform Sampling, arXiv 2603.12057) and contains none of the load-bearing operationalizations. Under the rule that circularity may be claimed only when a specific reduction can be quoted, no circular step can be exhibited. Shared-cue confounds between classifier and surprisal would be a validity concern, not circularity by construction. Score 0; steps empty.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 1 invented entities

Abstract-only review of the translationese paper. Free parameters and invented entities cannot be exhaustively listed without methods/results sections. Core load-bearing assumptions are the validity of the translatedness classifier and of information-theoretic difficulty proxies as measures of cognitive load in translation.

free parameters (2)
  • Classifier and regression hyperparameters (unspecified)
    Abstract does not report model choices, thresholds, or fitted coefficients for the translatedness classifier or difficulty predictors; any such fits would be free parameters of the central claim.
  • Weighting / relative contribution of source vs transfer difficulty
    Claimed dominance of cross-lingual transfer difficulty implies fitted or estimated relative weights in predictive models not given in the abstract.
axioms (4)
  • domain assumption Segment-level automatic translatedness scores validly measure translationese as studied in the literature.
    Abstract operationalizes translationese solely via a classifier score; validity of that score is assumed for all predictive claims.
  • domain assumption LLM surprisal and related information-theoretic metrics index cognitive load of source processing and cross-lingual transfer.
    Central explanatory story equates quantifiable difficulty metrics with cognitive load inherent in the translation task.
  • domain assumption English–German bidirectional written and spoken subcorpora are representative enough to support claims about language-pair and mode effects.
    Results are reported as differing by direction and mode; generalizability rests on this corpus choice.
  • standard math Standard statistical / predictive modeling assumptions for relating difficulty features to translatedness hold.
    Implicit in any claim that difficulty ‘explains’ or ‘predicts’ translationese.
invented entities (1)
  • Translation task difficulty as dual source-text + cross-lingual transfer construct no independent evidence
    purpose: Unifies prior translationese explanations under a single cognitive-load framing that can be measured and used as a predictor.
    Not a new particle-like entity, but a paper-specific construct bundling metrics; independent evidence would require behavioral/cognitive validation beyond classifier prediction.

pith-pipeline@v1.1.0-grok45 · 20599 in / 2892 out tokens · 35342 ms · 2026-07-14T22:28:18.193345+00:00 · methodology

0 comments
read the original abstract

Translations systematically diverge from texts originally produced in the target language, a phenomenon widely referred to as translationese. Translationese has been attributed to production tendencies (e.g. interference, simplification), socio-cultural variables, and language-pair effects, yet a unified explanatory account is still lacking. We propose that translationese reflects cognitive load inherent in the translation task itself. We test whether observable translationese can be predicted from quantifiable measures of translation task difficulty. Translationese is operationalised as a segment-level translatedness score produced by an automatic classifier. Translation task difficulty is conceptualised as comprising source-text and cross-lingual transfer components, operationalised mainly through information-theoretic metrics based on LLM surprisal, complemented by established syntactic and semantic alternatives. We use a bidirectional English-German corpus comprising written and spoken subcorpora. Results indicate that translationese can be partly explained by translation task difficulty, especially in English-to-German. For most experiments, cross-lingual transfer difficulty contributes more than source-text complexity. Information-theoretic indicators match or outperform traditional features in written mode, but offer no advantage in spoken mode. Source-text syntactic complexity and translation-solution entropy emerged as the strongest predictors of translationese across language pairs and modes.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

59 extracted references · 27 linked inside Pith

  1. [1]

    arXiv preprint arXiv:2111.13606 (2021)

    Batzolis, G., Stanczuk, J., Schönlieb, C.B., Etmann, C.: Conditional image gener- ation with score-based diffusion models. arXiv preprint arXiv:2111.13606 (2021)

    Pith/arXiv arXiv 2021

  2. [2]

    arXiv preprint arXiv:2410.02073 (2024)

    Bochkovskii, A., Delaunoy, A., Germain, H., Santos, M., Zhou, Y., Richter, S.R., Koltun, V.: Depth pro: Sharp monocular metric depth in less than a second. arXiv preprint arXiv:2410.02073 (2024)

    Pith/arXiv arXiv 2024

  3. [3]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Burgert, R., Xu, Y., Xian, W., Pilarski, O., Clausen, P., He, M., Ma, L., Deng, Y., Li, L., Mousavi, M., et al.: Go-with-the-flow: Motion-controllable video diffusion models using real-time warped noise. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 13–23 (2025)

    2025

  4. [4]

    IEEE Transactions on Computational Imaging3(1), 84–98 (2016)

    Chan, S.H., Wang, X., Elgendy, O.A.: Plug-and-play admm for image restoration: Fixed-point convergence and applications. IEEE Transactions on Computational Imaging3(1), 84–98 (2016)

    2016

  5. [5]

    arXiv preprint arXiv:2108.02938 (2021)

    Choi, J., Kim, S., Jeong, Y., Gwon, Y., Yoon, S.: Ilvr: Conditioning method for denoising diffusion probabilistic models. arXiv preprint arXiv:2108.02938 (2021)

    Pith/arXiv arXiv 2021

  6. [6]

    arXiv preprint arXiv:2209.14687 (2022)

    Chung, H., Kim, J., Mccann, M.T., Klasky, M.L., Ye, J.C.: Diffusion posterior sam- pling for general noisy inverse problems. arXiv preprint arXiv:2209.14687 (2022)

    Pith/arXiv arXiv 2022

  7. [7]

    Advances in Neural Information Processing Systems35, 25683–25696 (2022)

    Chung, H., Sim, B., Ryu, D., Ye, J.C.: Improving diffusion models for inverse problems using manifold constraints. Advances in Neural Information Processing Systems35, 25683–25696 (2022)

    2022

  8. [8]

    Advances in neural informa- tion processing systems34, 17695–17709 (2021)

    De Bortoli, V., Thornton, J., Heng, J., Doucet, A.: Diffusion schrödinger bridge with applications to score-based generative modeling. Advances in neural informa- tion processing systems34, 17695–17709 (2021)

    2021

  9. [9]

    Advances in neural information processing systems34, 8780–8794 (2021)

    Dhariwal, P., Nichol, A.: Diffusion models beat gans on image synthesis. Advances in neural information processing systems34, 8780–8794 (2021)

    2021

  10. [10]

    Doob, J.L., Doob, J.: Classical potential theory and its probabilistic counterpart, vol. 262. Springer (1984)

    1984

  11. [11]

    In: Forty-first international conference on machine learning (2024)

    Esser, P., Kulal, S., Blattmann, A., Entezari, R., Müller, J., Saini, H., Levi, Y., Lorenz, D., Sauer, A., Boesel, F., et al.: Scaling rectified flow transformers for high-resolution image synthesis. In: Forty-first international conference on machine learning (2024)

    2024

  12. [12]

    In: Proceedings of the IEEE/CVF In- ternational Conference on Computer Vision

    Gao, Z., Song, J., Zhang, Z., Deng, J., Patras, I.: Frequency-guided diffusion for training-free text-driven image translation. In: Proceedings of the IEEE/CVF In- ternational Conference on Computer Vision. pp. 19195–19205 (2025)

    2025

  13. [13]

    Advances in neural information processing systems30(2017)

    Heusel,M.,Ramsauer,H.,Unterthiner,T.,Nessler,B.,Hochreiter,S.:Ganstrained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems30(2017)

    2017

  14. [14]

    Advances in neural information processing systems33, 6840–6851 (2020)

    Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models. Advances in neural information processing systems33, 6840–6851 (2020)

    2020

  15. [15]

    arXiv preprint arXiv:2207.12598 (2022)

    Ho, J., Salimans, T.: Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598 (2022)

    Pith/arXiv arXiv 2022

  16. [16]

    arXiv preprint arXiv:2410.21276 (2024)

    Hurst, A., Lerer, A., Goucher, A.P., Perelman, A., Ramesh, A., Clark, A., Os- trow, A., Welihinda, A., Hayes, A., Radford, A., et al.: Gpt-4o system card. arXiv preprint arXiv:2410.21276 (2024)

    Pith/arXiv arXiv 2024

  17. [17]

    In: The Twelfth International Conference on Learning Representations (2023)

    Ju, X., Zeng, A., Bian, Y., Liu, S., Xu, Q.: Pnp inversion: Boosting diffusion-based editing with 3 lines of code. In: The Twelfth International Conference on Learning Representations (2023)

    2023

  18. [18]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Karras, T., Laine, S., Aila, T.: A style-based generator architecture for generative adversarial networks. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 4401–4410 (2019) 16 Y. Wang, Z. Jiang, Z. Wang, L. Chen

    2019

  19. [19]

    Advances in neural information processing systems35, 23593–23606 (2022)

    Kawar, B., Elad, M., Ermon, S., Song, J.: Denoising diffusion restoration models. Advances in neural information processing systems35, 23593–23606 (2022)

    2022

  20. [20]

    arXiv preprint arXiv:2505.23145 (2025)

    Kim, J., Hong, Y., Ye, J.C.: Flowalign: Trajectory-regularized, inversion-free flow- based image editing. arXiv preprint arXiv:2505.23145 (2025)

    Pith/arXiv arXiv 2025

  21. [21]

    arXiv preprint arXiv:2503.08136 (2025)

    Kim, J., Kim, B.S., Ye, J.C.: Flowdps: Flow-driven posterior sampling for inverse problems. arXiv preprint arXiv:2503.08136 (2025)

    Pith/arXiv arXiv 2025

  22. [22]

    arXiv preprint arXiv:2412.08629 (2024)

    Kulikov, V., Kleiner, M., Huberman-Spiegelglas, I., Michaeli, T.: Flowedit: Inversion-free text-based editing using pre-trained flow models. arXiv preprint arXiv:2412.08629 (2024)

    Pith/arXiv arXiv 2024

  23. [23]

    Labs, B.F.: Flux.https://github.com/black-forest-labs/flux(2024)

    2024

  24. [24]

    1 kontext: Flow match- ing for in-context image generation and editing in latent space

    Labs, B.F., Batifol, S., Blattmann, A., Boesel, F., Consul, S., Diagne, C., Dock- horn, T., English, J., English, Z., Esser, P., et al.: Flux. 1 kontext: Flow match- ing for in-context image generation and editing in latent space. arXiv preprint arXiv:2506.15742 (2025)

    Pith/arXiv arXiv 2025

  25. [25]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Liang, J., Cao, J., Sun, G., Zhang, K., Van Gool, L., Timofte, R.: Swinir: Image restoration using swin transformer. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 1833–1844 (2021)

    2021

  26. [26]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Ling, L., Sheng, Y., Tu, Z., Zhao, W., Xin, C., Wan, K., Yu, L., Guo, Q., Yu, Z., Lu, Y., et al.: Dl3dv-10k: A large-scale scene dataset for deep learning-based 3d vision. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 22160–22169 (2024)

    2024

  27. [27]

    arXiv preprint arXiv:2210.02747 (2022)

    Lipman, Y., Chen, R.T., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. arXiv preprint arXiv:2210.02747 (2022)

    Pith/arXiv arXiv 2022

  28. [28]

    arXiv preprint arXiv:2209.03003 (2022)

    Liu, X., Gong, C., Liu, Q.: Flow straight and fast: Learning to generate and transfer data with rectified flow. arXiv preprint arXiv:2209.03003 (2022)

    Pith/arXiv arXiv 2022

  29. [29]

    Interna- tional journal of computer vision60(2), 91–110 (2004)

    Lowe, D.G.: Distinctive image features from scale-invariant keypoints. Interna- tional journal of computer vision60(2), 91–110 (2004)

    2004

  30. [30]

    Entropy22(8), 802 (2020)

    Maoutsa, D., Reich, S., Opper, M.: Interacting particle solutions of fokker–planck equations through gradient–log–density estimation. Entropy22(8), 802 (2020)

    2020

  31. [31]

    arXiv preprint arXiv:2108.01073 (2021)

    Meng, C., Song, Y., Song, J., Wu, J., Zhu, J.Y., Ermon, S.: Sdedit: Image synthesis and editing with stochastic differential equations. arXiv preprint arXiv:2108.01073 (2021)

    Pith/arXiv arXiv 2021

  32. [32]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Murez, Z., Kolouri, S., Kriegman, D., Ramamoorthi, R., Kim, K.: Image to image translation for domain adaptation. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4500–4509 (2018)

    2018

  33. [33]

    arXiv preprint arXiv:2304.07193 (2023)

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al.: Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193 (2023)

    Pith/arXiv arXiv 2023

  34. [34]

    arXiv preprint arXiv:2403.12036 (2024)

    Parmar, G., Park, T., Narasimhan, S., Zhu, J.Y.: One-step image translation with text-to-image models. arXiv preprint arXiv:2403.12036 (2024)

    Pith/arXiv arXiv 2024

  35. [35]

    arXiv preprint arXiv:2307.01952 (2023)

    Podell, D., English, Z., Lacey, K., Blattmann, A., Dockhorn, T., Müller, J., Penna, J., Rombach, R.: Sdxl: Improving latent diffusion models for high-resolution image synthesis. arXiv preprint arXiv:2307.01952 (2023)

    Pith/arXiv arXiv 2023

  36. [36]

    In: International conference on machine learning

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al.: Learning transferable visual models from natural language supervision. In: International conference on machine learning. pp. 8748–8763. PmLR (2021)

    2021

  37. [37]

    In: Proceedings of the IEEE/CVF interna- tional conference on computer vision

    Ren, M., Delbracio, M., Talebi, H., Gerig, G., Milanfar, P.: Multiscale structure guided diffusion for image deblurring. In: Proceedings of the IEEE/CVF interna- tional conference on computer vision. pp. 10721–10733 (2023) Coarse-guided-Gen 17

    2023

  38. [38]

    Cambridge university press (2000)

    Rogers,L.C.G.,Williams,D.:Diffusions,Markovprocesses,andmartingales,vol.2. Cambridge university press (2000)

    2000

  39. [39]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10684–10695 (2022)

    2022

  40. [40]

    Advances in neural information processing systems37, 6246– 6266 (2024)

    Shen, F., Tang, J.: Imagpose: A unified conditional framework for pose-guided person generation. Advances in neural information processing systems37, 6246– 6266 (2024)

    2024

  41. [41]

    arXiv preprint arXiv:2511.08633 (2025)

    Singer, A., Rotstein, N., Mann, A., Kimmel, R., Litany, O.: Time-to-move: Training-free motion controlled video generation via dual-clock denoising. arXiv preprint arXiv:2511.08633 (2025)

    arXiv 2025

  42. [42]

    arXiv preprint arXiv:2010.02502 (2020)

    Song, J., Meng, C., Ermon, S.: Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502 (2020)

    Pith/arXiv arXiv 2010

  43. [43]

    Advances in neural information processing systems32(2019)

    Song, Y., Ermon, S.: Generative modeling by estimating gradients of the data distribution. Advances in neural information processing systems32(2019)

    2019

  44. [44]

    arXiv preprint arXiv:2011.13456 (2020)

    Song, Y., Sohl-Dickstein, J., Kingma, D.P., Kumar, A., Ermon, S., Poole, B.: Score- based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456 (2020)

    Pith/arXiv arXiv 2011

  45. [45]

    arXiv preprint arXiv:2203.08382 (2022)

    Su, X., Song, J., Meng, C., Ermon, S.: Dual diffusion implicit bridges for image- to-image translation. arXiv preprint arXiv:2203.08382 (2022)

    Pith/arXiv arXiv 2022

  46. [46]

    SIAM (2005)

    Tarantola, A.: Inverse problem theory and methods for model parameter estima- tion. SIAM (2005)

    2005

  47. [47]

    In: European conference on computer vision

    Teed, Z., Deng, J.: Raft: Recurrent all-pairs field transforms for optical flow. In: European conference on computer vision. pp. 402–419. Springer (2020)

    2020

  48. [48]

    arXiv preprint arXiv:1812.01717 (2018)

    Unterthiner, T., Van Steenkiste, S., Kurach, K., Marinier, R., Michalski, M., Gelly, S.: Towards accurate generative models of video: A new metric & challenges. arXiv preprint arXiv:1812.01717 (2018)

    Pith/arXiv arXiv 2018

  49. [49]

    arXiv preprint arXiv:2503.20314 (2025)

    Wan, T., Wang, A., Ai, B., Wen, B., Mao, C., Xie, C.W., Chen, D., Yu, F., Zhao, H., Yang, J., et al.: Wan: Open and advanced large-scale video generative models. arXiv preprint arXiv:2503.20314 (2025)

    Pith/arXiv arXiv 2025

  50. [50]

    arXiv preprint arXiv:2205.12952 (2022)

    Wang, T., Zhang, T., Zhang, B., Ouyang, H., Chen, D., Chen, Q., Wen, F.: Pretraining is all you need for image-to-image translation. arXiv preprint arXiv:2205.12952 (2022)

    Pith/arXiv arXiv 2022

  51. [51]

    IEEE transactions on pattern analysis and machine intelligence43(10), 3365–3387 (2020)

    Wang, Z., Chen, J., Hoi, S.C.: Deep learning for image super-resolution: A survey. IEEE transactions on pattern analysis and machine intelligence43(10), 3365–3387 (2020)

    2020

  52. [52]

    arXiv preprint arXiv:2406.03293 (2024)

    Yang, X., Chen, C., Yang, X., Liu, F., Lin, G.: Text-to-image rectified flow as plug-and-play priors. arXiv preprint arXiv:2406.03293 (2024)

    Pith/arXiv arXiv 2024

  53. [53]

    arXiv preprint arXiv:2408.06072 (2024)

    Yang, Z., Teng, J., Zheng, W., Ding, M., Huang, S., Xu, J., Yang, Y., Hong, W., Zhang, X., Feng, G., et al.: Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint arXiv:2408.06072 (2024)

    Pith/arXiv arXiv 2024

  54. [54]

    Yu, J., Wang, Y., Zhao, C., Ghanem, B., Zhang, J.: Freedom: Training-free energy- guidedconditionaldiffusionmodel.In:ProceedingsoftheIEEE/CVFInternational Conference on Computer Vision. pp. 23174–23184 (2023)

    2023

  55. [55]

    arXiv preprint arXiv:2409.19365 (2024)

    Zhan, Z., Chen, D., Mei, J.P., Zhao, Z., Chen, J., Chen, C., Lyu, S., Wang, C.: Conditional image synthesis with diffusion models: A survey. arXiv preprint arXiv:2409.19365 (2024)

    Pith/arXiv arXiv 2024

  56. [56]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Zhang, L., Rao, A., Agrawala, M.: Adding conditional control to text-to-image diffusion models. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 3836–3847 (2023) 18 Y. Wang, Z. Jiang, Z. Wang, L. Chen

    2023

  57. [57]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Zhang, R., Isola, P., Efros, A.A., Shechtman, E., Wang, O.: The unreasonable effectiveness of deep features as a perceptual metric. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 586–595 (2018)

    2018

  58. [58]

    arXiv preprint arXiv:2405.15885 (2024)

    Zheng, K., He, G., Chen, J., Bao, F., Zhu, J.: Diffusion bridge implicit models. arXiv preprint arXiv:2405.15885 (2024)

    Pith/arXiv arXiv 2024

  59. [59]

    Zhou, L., Lou, A., Khanna, S., Ermon, S.: Denoising diffusion bridge models. arXiv preprint arXiv:2309.16948 (2023) Coarse-guided-Gen 19 Appendix This appendix is organized as follows: –Section A.1 gives the proof of equivalent marginal distributions between the reversed-SDE and its corresponding PF-ODE. –Section A.2 demonstrates the detailed derivation o...

    Pith/arXiv arXiv 2023