Pith. sign in

REVIEW 4 major objections 6 minor 38 references

Steering Large Text-to-Image Model for Abstract Art Synthesis: Preference-based Prompt Optimization and Visualization

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A two-stage system—semantic injection plus genetic prompt optimization—lets users generate Kandinsky-style abstract art by voting on images, with no prompt writing required.

desk verdict A usable integration of style fine-tuning, GA prompt search, and human voting for Kandinsky-style art, undermined by a seed-confound in the GA and a lack of quantitative evidence. read the letter →

arxiv 2412.14174 v1 pith:HBX6454E submitted 2024-11-18 cs.HC

Aven-Le Zhou , Wei Wu , Yu-Ao Wang , Kang Zhang This is my paper · ORCID
classification cs.HC
keywords text-to-imagegenerationgeneticalgorithmhumanfeedbackpromptoptimizationabstractartKandinskyBauhausstylesemanticinjectionhuman-AIinteraction
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

The paper claims that users can create abstract art in a specific painter's style without writing a single prompt, by voting on images for three to five rounds. It builds a two-stage system: semantic injection embeds Kandinsky's Bauhaus style into a large text-to-image model, and a genetic algorithm then searches prompt space using the user's votes as a fitness signal until it converges on an Optimized Prompting Model that generates images the user prefers. The claim matters because prompt writing is the main barrier for non-experts using text-to-image models, and the approach promises to replace that hurdle with simple preference feedback. The paper supports the claim with a constructed Kandinsky-style dataset, an interactive interface, and in-team convergence tests, while acknowledging that a broader user study is still needed.

What carries the argument

The central mechanism is genetic prompting optimization, where a prompt is a chromosome $C=\{Style, A_Q, A_M, S\}$: $A_M$ groups discrete attributes like form and hue handled by FastLoRA injections, $A_Q$ groups continuous attributes like brightness and composition handled by DiffLoRA injections, and $S$ is the random seed. Fitness of an individual is its vote count $f(i)=V_i$, selection is roulette-wheel based on vote share, crossover and mutation operate separately per gene type, and continuous genes are updated through a normal distribution whose parameters shift with user feedback. This machinery converts a sequence of simple image votes into concrete changes in prompt attributes, guiding the search toward a personalized prompt.

What would settle it

Compare two user groups on the same initial image populations: one group votes according to their genuine preferences and the other votes randomly; if the randomly voting group produces an Optimized Prompting Model that users rate as equally satisfying, then the vote signal is doing no real optimization work and the central claim fails.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is a prompting-free workflow for abstract art: users vote on candidate images, and the system searches prompt space for them. The paper establishes this through semantic injection, which encodes an artist's style into a large text-to-image model by combining FastLoRA for discrete style attributes and DiffLoRA for continuous ones, and through genetic prompting optimization, which treats prompts as chromosomes, uses vote counts as fitness, and iterates selection, crossover, and mutation for three to five rounds. The result is an Optimized Prompting Model that, combined with the Artist Model, generates abstract art matching the user's preferences without any explicit prompt. The paper also contributes a curated Kandinsky Bauhaus text-to-image dataset and an open interactive system that visualizes how prompt attributes evolve.

Load-bearing premise

The claim rests on the assumption that a user's aesthetic preferences stay consistent while they vote; if tastes drift or votes contradict earlier votes, the genetic algorithm's fitness signal becomes noise and the optimized prompt may reflect chance rather than preference.

Editorial extensions

If this is right

  • If the approach works as reported, a user with no prompt-engineering skill can get a personalized abstract art generator in 3 to 5 rounds of voting, in under five minutes.
  • The same two-stage recipe can, in principle, be transferred to any describable visual style, because semantic injection only requires an attribute-value guideline and the genetic loop only requires votes.
  • Because the optimized prompt is itself text, the system produces a shareable, inspectable description of the user's taste, not just the generated images.
  • For artists, this supports a steering rather than training relationship with large models, lowering the resource cost of customizing generative art tools.

Reading between the lines

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

  • The paper does not test for preference drift; a natural extension would be to measure vote consistency across rounds and check whether inconsistent voters still converge to a stable preferred prompt.
  • Automating the semantic descriptive guideline with language models could remove the expert-cost bottleneck, though it would trade away the expert validation that currently grounds the dataset and style labels.
  • The small population size of 16 and one-offspring crossover are pragmatic choices to avoid user fatigue; testing larger populations or multiple offspring per crossover would clarify whether these choices limit the quality of the final prompt.
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. This paper proposes a two-part system for creating abstract art in the Kandinsky Bauhaus style without explicit user prompt engineering. The first part, 'semantic injection,' combines FastLoRA and DiffLoRA to fine-tune Stable Diffusion into a deterministic Artist Model. The second part, 'genetic prompting optimization,' uses a genetic algorithm with real-time human votes to evolve procedural prompts, producing an 'Optimized Prompting Model' that the authors claim adapts to user preferences. The paper also describes a Kandinsky Bauhaus text-to-image dataset, an interactive visualization interface, and reports convergence within 3 to 5 iterations in internal testing.

Significance. If the central claim were fully supported, the approach would be a useful contribution to human-AI interaction for generative art, lowering the prompt-engineering barrier for non-expert users. The paper's strengths include the construction of a curated Kandinsky Bauhaus dataset, the combination of LoRA-based style injection with evolutionary prompt search, and an interactive visualization design that makes the optimization process transparent. However, the current evidence is largely qualitative: the Artist Model evaluation in Sec 5 reports no quantitative metrics, and the prompting-free experience in Sec 6.4 rests on internal testing without a user study. The most serious concern is that the genetic algorithm optimizes the random seed jointly with semantic attributes, so the reported convergence may reflect seed selection rather than learned prompt semantics. These issues make the central claim about automatic, preference-aligned generation not yet convincing, although the system appears functional as a demo.

major comments (4)
  1. [Sec 6.1-6.2] The chromosome C = {Style, A_Q, A_M, S} explicitly includes the Stable Diffusion random seed S, and Sec 6.2 applies uniform crossover (p = 0.5) and uniform mutation (p = 0.05) to the seed gene. Because image output is highly sensitive to the seed, a user can drive the GA to high fitness by selecting a seed that produces a liked image, without the semantic attribute values A_Q/A_M converging to meaningful preferences. The manuscript never states how S is chosen after optimization: if S is fixed, the Optimized Prompting Model replays the voted images, and if S is re-sampled, no evidence shows that the learned attribute weights transfer to new seeds. This directly undermines the central claim in Sec 1.3 that the Optimized Prompting Model automatically generates art matching user preferences. The authors should either fix the seed in a controlled experiment or evaluate generalization over newly sampled seeds.
  2. [Sec 6.2, Fitness Function and Weight Updating] The weight update w'_v = w_v + sum_i V_i for v in C_i, combined with fitness f(i) = V_i, makes the statement that the model 'adapts to user preferences' largely a restatement of the update rule: any attribute value appearing in a voted image receives increased weight. The paper provides no separate evaluation showing that the resulting Optimized Prompting Model generalizes to prompts or images not seen during voting. The reported convergence in Sec 6.4 is therefore partly by construction. To support the central claim, the authors should measure preference alignment on held-out or newly sampled prompts and images after optimization.
  3. [Sec 5.1 and Sec 6.4] Sec 5.1 states that FastLoRA demonstrates 'significantly superior performance' in generating Kandinsky-specific form features, but no numerical results, statistical tests, or inter-rater reliability measures are reported. Similarly, Sec 6.4 reports convergence 'typically within 3 to 5 iterations' based on 'testing within the research team,' with no formal user study. The paper's own Limitation section (Sec 8) concedes limited user evaluation. Given that the central claim is a human-centered interaction claim, at least a small user study or a quantitative preference metric (e.g., vote consistency or preference ranking against a baseline) is needed.
  4. [Sec 3.2] The method assumes that users maintain consistent aesthetic preferences throughout the iterative process. The interface provides no mechanism to detect or correct for preference drift or noisy voting. If a user's tastes shift during the session, the vote signal becomes noise and the GA may converge to an arbitrary prompt. The paper does not discuss robustness to this assumption, so the reliability of the feedback signal remains unexamined. A post-hoc analysis of vote consistency across iterations would help assess this risk.
minor comments (6)
  1. [Sec 3.1] The low-rank factor d is described as 'usually much smaller than d'; this should presumably read 'much smaller than n and m' in the context of W = W + AB^T.
  2. [Sec 6.2] The phrase 'The values of mutedAM are selected without replacement' contains a typo: 'mutedAM' should be 'mutated A_M'.
  3. [Sec 6.2] The fitness value V_i is said to be 'an integer in the range of (0, +∞)', but a vote count should be a non-negative integer including zero; the notation is inconsistent.
  4. [Sec 6.1] The text 'random prompts set' is grammatically awkward; consider 'a random set of prompts'.
  5. [References] Reference [2] is a GitHub repository rather than a peer-reviewed description of FastLoRA; the authors should cite the original FastLoRA paper or provide a versioned repository URL.
  6. [Sec 1.3 and Sec 6.4] The paper claims the system is 'open-sourced' but no repository link or code availability statement is provided in the manuscript; please add one if applicable.

Circularity Check

1 steps flagged · score 3.0 of 10

Preference-adaptive prompting reduces by construction to accumulated user votes; otherwise the paper is externally grounded.

  1. self definitional [Sec 6.2 'Genetic Optimization' (Fitness Function and Weight Updating), coupled with the Sec 1.3 contribution claim]
    "The fitness function is defined as f(i) =Vi. ... In each iteration, the updated weight is computed as w′v = wv +Pn i=0 Vi, where v ∈ Ci, C is the chromosome (prompt) for individual i. ... This iterative process continues until the user is satisfied with the generated images, resulting in an Optimized Prompting Model."

    The Optimized Prompting Model's state (discrete attribute weights and continuous normal-distribution parameters) is updated by directly adding the user's vote counts, and the process terminates when the user is satisfied. Therefore the claim that the model 'adapts to user preferences and generates prompts automatically' is a restatement of the update rule rather than an independently verified result: no held-out preference test or new-image generalization check is reported, so the only evidence that the model encodes user preference is the very votes used to construct it. The circularity is partial because the Artist Model's style fidelity is separately grounded in external expert judgment and a curated Kandinsky dataset.

full rationale

The derivation chain is mostly self-contained. The Artist Model is built by fine-tuning Stable Diffusion with FastLoRA/DiffLoRA and is evaluated against a baseline and with expert judgment using an externally curated Kandinsky dataset, so those components do not reduce to their own inputs. The genetic prompting loop is a standard interactive genetic algorithm: user votes directly update fitness and attribute weights in Sec 6.2, so the statement that the resulting Optimized Prompting Model 'adapts to user preferences' is definitionally tied to the vote-accumulation rule. I flag this one step as a mild self-definitional circularity because no held-out preference data or new-image generalization test is reported; the preference evidence is the same votes that set the weights. This is not a self-citation chain, an imported uniqueness theorem, or a renamed known result, and the seed-in-chromosome confound is a validity/generalization concern rather than a circularity concern. Overall the core system has independent components and external grounding for the Artist Model, so the circularity score is low.

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

The central claims rest on expert-curated semantic attributes, on the reliability of LoRA-based style injection, and on stable user preferences during the interaction. No new physical entities or unvetted scientific constants are introduced; the GA hyperparameters are hand-chosen. The main burden is domain assumptions about Kandinsky's style and user consistency.

free parameters (4)
  • GA population size n = 16
    Chosen to balance user fatigue and diversity; not tuned against external benchmarks (Sec 6.2).
  • Mutation rate p_mut = 0.05
    Standard GA setting, chosen by hand (Sec 6.2).
  • Crossover probability for seed gene = 0.5
    Set fixed for uniform crossover (Sec 6.2).
  • Initial prompt weights w_v = 1
    Initial weights for all attribute values; updated by vote sums (Sec 6.2).
assumptions (4)
  • domain assumption Users maintain consistent tendencies in their aesthetic preferences throughout the process.
    Explicitly stated in Sec 3.2; the GA's fitness signal is the user's votes, so inconsistent or drifting preferences would break convergence.
  • domain assumption The semantic descriptive guideline (7 attributes, 22 values) captures the essential, unambiguous features of Kandinsky's Bauhaus style.
    Derived by experts from Kandinsky's writings (Sec 4.1); if wrong or incomplete, the dataset and Artist Model inherit the error.
  • domain assumption LoRA fine-tuning with FastLoRA and DiffLoRA can independently and reliably encode discrete and continuous stylistic attributes.
    The Artist Model assumes that a small set of low-rank adapter weights suffices to control form, hue, brightness, and composition (Sec 5).
  • standard math Stable Diffusion with a fixed seed is deterministic for a given prompt.
    The paper relies on this for 'deterministically generating' (Sec 1 and 3); this is a property of the deployed inference pipeline.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Steering Large Text-to-Image Model for Abstract Art Synthesis: Preference-based Prompt Optimization and Visualization." pith.science (2026). https://pith.science/paper/HBX6454E

@misc{pith2026241214174,
  author       = {Pith},
  title        = {Pith review of: Steering Large Text-to-Image Model for Abstract Art Synthesis: Preference-based Prompt Optimization and Visualization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HBX6454E}},
  note         = {Machine review of arXiv:2412.14174}
}
read the original abstract

With the advancement of neural generative capabilities, the art community has increasingly embraced GenAI (Generative Artificial Intelligence), particularly large text-to-image models, for producing aesthetically compelling results. However, the process often lacks determinism and requires a tedious trial-and-error process as users often struggle to devise effective prompts to achieve their desired outcomes. This paper introduces a prompting-free generative approach that applies a genetic algorithm and real-time iterative human feedback to optimize prompt generation, enabling the creation of user-preferred abstract art through a customized Artist Model. The proposed two-part approach begins with constructing an Artist Model capable of deterministically generating abstract art in specific styles, e.g., Kandinsky's Bauhaus style. The second phase integrates real-time user feedback to optimize the prompt generation and obtains an Optimized Prompting Model, which adapts to user preferences and generates prompts automatically. When combined with the Artist Model, this approach allows users to create abstract art tailored to their personal preferences and artistic style.

Figures

Figures reproduced from arXiv: 2412.14174 by the authors.

Figure 1
Figure 1. Artist Model (of Kandinsky Bauhaus Style) through Semantic Injection. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Genetic Prompting Optimization with a User’s Real-time Feedback Leads [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Genetic Prompting Optimization with Real-time Iterative Feedback based [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Semantic Descriptive Guideline: Attribute-Value. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Kandinsky Text-to-Image Dataset. We collaborate with experts to categorize and label the selected paintings us￾ing the attribute-value list [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Comparison of Baseline and the Artist Models from Semantic Injection. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Data Sample for Composition-related Attributes. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: The Interface Design charts depict attributes with continuous values, showcasing the distribution of attribute values in the current user-voted and selected prompts. Stream graphs illustrate the progression of attribute-value distributions over multiple iterations, pro…
Figure 9
Figure 9. Figure 9: Flowchart: the Interactive System for Prompting-free Experience [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 21 canonical work pages

  1. [21]

    In: Proceedings of the 46th International ACM SIGIR Confer- ence on Research and Development in Information Retrieval

    Pavlichenko, N., Ustalov, D.: Best prompts for text-to-image models and how to find them. In: Proceedings of the 46th International ACM SIGIR Confer- ence on Research and Development in Information Retrieval. p. 2067–2071. SIGIR ’23, Association for Computing Machinery, New York, NY, USA (Jul2023).https://doi.org/10.1145/3539618.3592000, \url{https://dl.a...

    arXiv

  2. [7]

    https://doi.org/10.48550/arXiv.2212.09611, \url{http://arxiv.org/abs/2212.09611}, arXiv:2212.09611 [cs] 16 A.L

    Hao, Y., Chi, Z., Dong, L., Wei, F.: Optimizing prompts for text-to-image gener- ation (arXiv:2212.09611) (Dec 2023). https://doi.org/10.48550/arXiv.2212.09611, \url{http://arxiv.org/abs/2212.09611}, arXiv:2212.09611 [cs] 16 A.L. Zhou et al

    arXiv doi:10.48550/arxiv.2212.09611 2023

  3. [1]

    Bailey, J.: Why love generative art? (Aug 2018),\url{https://www.artnome.com/ news/2018/8/8/why-love-generative-art}

    work page 2018

  4. [2]

    Cloneofsimo: Cloneofsimo/lora: Using low-rank adaptation to quickly fine-tune dif- fusion models., \url{https://github.com/cloneofsimo/lora}

    work page

  5. [3]

    In: Avidan, S., Brostow, G., Cissé, M., Farinella, G.M., Hassner, T

    Crowson, K., Biderman, S., Kornis, D., Stander, D., Hallahan, E., Castricato, L., Raff, E.: Vqgan-clip: Open domain image generation and editing with natural language guidance. In: Avidan, S., Brostow, G., Cissé, M., Farinella, G.M., Hassner, T. (eds.) Computer Vision – ECCV 2022. p. 88–105. Springer Nature Switzerland, Cham (2022)

    work page 2022

  6. [4]

    https://doi.org/10.48550/arXiv.2212.05032, \url{http://arxiv.org/abs/2212

    Feng, W., He, X., Fu, T.J., Jampani, V., Akula, A., Narayana, P., Basu, S., Wang, X.E., Wang, W.Y.: Training-free structured diffusion guid- ance for compositional text-to-image synthesis (arXiv:2212.05032) (Feb 2023). https://doi.org/10.48550/arXiv.2212.05032, \url{http://arxiv.org/abs/2212. 05032}, arXiv:2212.05032 [cs]

    arXiv doi:10.48550/arxiv.2212.05032 2023

  7. [5]

    Galanter, P.: Generative Art Theory, pp. 146–180. John Wiley and Sons, Ltd (2016), \url{https://onlinelibrary.wiley.com/doi/abs/10.1002/ 9781118475249.ch5},DOI={10.1002/9781118475249.ch5}

    work page doi:10.1002/9781118475249.ch5 2016

  8. [6]

    In: Proceed- ings of the sixth annual conference on Computational learning theory - COLT ’93

    Grefenstette, J.J.: Genetic algorithms and machine learning. In: Proceed- ings of the sixth annual conference on Computational learning theory - COLT ’93. p. 3–4. ACM Press, Santa Cruz, California, United States (1993). https://doi.org/10.1145/168304.168305, \url{http://portal.acm.org/ citation.cfm?doid=168304.168305}

    arXiv 1993

Show all 38 references
  1. [8]

    MIT Press (Apr 1992), google-Books-ID: 5EgGaBkwvWcC

    Holland, J.H.: Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence. MIT Press (Apr 1992), google-Books-ID: 5EgGaBkwvWcC

    1992

  2. [9]

    Hu, E.J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W.: Lora: Low-rank adaptation of large language models (arXiv:2106.09685) (Oct 2021), \url{http://arxiv.org/abs/2106.09685}, arXiv:2106.09685 [cs]

    2021 arXiv

  3. [10]

    Dover Publications, Inc (1977)

    Kandinsky, W.: Concerning the Spiritual in Art. Dover Publications, Inc (1977)

    1977

  4. [11]

    Courier Corporation (1979)

    Kandinsky, W.: Point and Line to Plane. Courier Corporation (1979)

    1979

  5. [12]

    Kovach, B.: A box of chaos: The generative artist’s toolkit (Oct 2018),\url{https: //www.youtube.com/watch?v=kZNTozzsNqk}

    2018

  6. [13]

    Kumari, N., Zhang, B., Zhang, R., Shechtman, E., Zhu, J.Y.: Multi-concept customization of text-to-image diffusion. pp. 1931–1941 (2023), \url{https: //openaccess.thecvf.com/content/CVPR2023/html/Kumari_Multi-Concept_ Customization_of_Text-to-Image_Diffusion_CVPR_2023_paper.html}

    2023

  7. [14]

    International Journal of Machine Learning and Computing 7(1), 9–12 (Feb 2017)

    Lim, S.M., Sultan, A.B.M., Sulaiman, M.N., Mustapha, A., Leong, K.Y.: Crossover and mutation operators of genetic algorithms. International Journal of Machine Learning and Computing 7(1), 9–12 (Feb 2017). https://doi.org/10.18178/ijmlc.2017.7.1.611

    2017 doi

  8. [15]

    Lin, Z., Ehsan, U., Agarwal, R., Dani, S., Vashishth, V., Riedl, M.: Be- yond prompts: Exploring the design space of mixed-initiative co-creativity sys- tems (arXiv:2305.07465) (May 2023),\url{http://arxiv.org/abs/2305.07465}, arXiv:2305.07465 [cs]

    2023 arXiv

  9. [16]

    In: Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems

    Liu, V., Chilton, L.B.: Design guidelines for prompt engineering text-to-image gen- erative models. In: Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. pp. 1–23. CHI ’22, Association for Computing Machin- ery, New York, NY, USA (Apr 2022). https:/...

    2022

  10. [17]

    McCormack, J.: Art, emergence, and the computational sublime

  11. [18]

    Leonardo47(2), 135–141 (Apr 2014)

    McCormack, J., Bown, O., Dorin, A., McCabe, J., Monro, G., Whitelaw, M.: Ten questions concerning generative computer art. Leonardo47(2), 135–141 (Apr 2014). https://doi.org/10.1162/LEON_a_00533

    2014 doi

  12. [19]

    Leonardo42(5), 476–477 (Oct 2009)

    Monro, G.: Emergence and generative art. Leonardo42(5), 476–477 (Oct 2009). https://doi.org/10.1162/leon.2009.42.5.476

    2009 doi

  13. [20]

    In: Proceedings of the 25th International Academic Mindtrek Conference

    Oppenlaender, J.: The creativity of text-to-image generation. In: Proceedings of the 25th International Academic Mindtrek Conference. p. 192–202. Aca- demic Mindtrek ’22, Association for Computing Machinery, New York, NY, USA (Nov 2022). https://doi.org/10.1145/3569219.3569352...

    2022

  14. [22]

    In: Proceedings of the 38th 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., Krueger, G., Sutskever, I.: Learning transfer- able visual models from natural language supervision. In: Proceedings of the 38th International Conference on M...

    2021

  15. [23]

    https://doi.org/10.48550/arXiv.2204.06125, \url{http://arxiv.org/abs/2204

    Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., Chen, M.: Hierarchical text- conditional image generation with clip latents (arXiv:2204.06125) (Apr 2022). https://doi.org/10.48550/arXiv.2204.06125, \url{http://arxiv.org/abs/2204. 06125}, arXiv:2204.06125 [cs]

    2022 doi arXiv

  16. [24]

    In: Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems

    Reynolds, L., McDonell, K.: Prompt programming for large language mod- els: Beyond the few-shot paradigm. In: Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems. pp. 1–7. CHI EA ’21, Association for Computing Machinery, New York, NY, USA (May 2...

    2021

  17. [25]

    Rombach,R.,Blattmann,A.,Lorenz,D.,Esser,P.,Ommer,B.:High-resolutionim- age synthesis with latent diffusion models. p. 10684–10695 (2022),\url{https:// openaccess.thecvf.com/content/CVPR2022/html/Rombach_High-Resolution_ Image_Synthesis_With_Latent_Diffusion_Models_CVPR_2022_pa...

    2022

  18. [26]

    https://doi.org/10.48550/arXiv.2207.13038, \url{http://arxiv.org/abs/ 2207.13038}, arXiv:2207.13038 [cs]

    Rombach, R., Blattmann, A., Ommer, B.: Text-guided synthesis of artis- tic images with retrieval-augmented diffusion models (arXiv:2207.13038) (Jul 2022). https://doi.org/10.48550/arXiv.2207.13038, \url{http://arxiv.org/abs/ 2207.13038}, arXiv:2207.13038 [cs]

    2022 doi arXiv

  19. [27]

    https://doi.org/10.48550/arXiv.2208.12242, \url{http://arxiv.org/abs/2208.12242}, arXiv:2208.12242 [cs]

    Ruiz, N., Li, Y., Jampani, V., Pritch, Y., Rubinstein, M., Aberman, K.: Dream- booth: Fine tuning text-to-image diffusion models for subject-driven genera- tion (arXiv:2208.12242) (Mar 2023). https://doi.org/10.48550/arXiv.2208.12242, \url{http://arxiv.org/abs/2208.12242}, arX...

    2023 doi arXiv

  20. [28]

    Runwayml: Runwayml/stable-diffusion-v1-5 · hugging face, \url{https:// huggingface.co/runwayml/stable-diffusion-v1-5}

  21. [29]

    In: Koyejo, S., Mohamed, S., Agar- wal, A., Belgrave, D., Cho, K., Oh, A

    Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E.L., Ghasemipour, K., Gontijo Lopes, R., Karagol Ayan, B., Salimans, T., Ho, J., Fleet, D.J., Norouzi, M.: Photorealistic text-to-image diffusion mod- els with deep language understanding. In: Koyejo, S., Mohamed, ...

    2022

  22. [30]

    In: Proceedings of the 32nd International Conference on Machine Learning

    Sohl-Dickstein, J., Weiss, E., Maheswaranathan, N., Ganguli, S.: Deep unsuper- vised learning using nonequilibrium thermodynamics. In: Proceedings of the 32nd International Conference on Machine Learning. p. 2256–2265. PMLR (Jun 2015), \url{https://proceedings.mlr.press/v37/so...

    2015

  23. [31]

    https://doi.org/10.48550/arXiv.2303.03751, \url{http://arxiv.org/abs/2303

    Tang, Z., Rybin, D., Chang, T.H.: Zeroth-order optimization meets human feedback: Provable learning via ranking oracles (arXiv:2303.03751) (Mar 2023). https://doi.org/10.48550/arXiv.2303.03751, \url{http://arxiv.org/abs/2303. 03751}, arXiv:2303.03751 [cs]

    2023 doi arXiv

  24. [32]

    Toriato: Toriato/stable-diffusion-webui-wd14-tagger: Labeling exten- sion for automatic1111’s web ui, \url{https://github.com/toriato/ stable-diffusion-webui-wd14-tagger}

  25. [33]

    Tsukisuwa, N.: Difflora (Aug 2023),\url{https://pixai.art/model}

    2023

  26. [34]

    Von Rütte, D., Fedele, E., Thomm, J., Wolf, L.: Fabric: Personalizing diffu- sion models with iterative feedback (arXiv:2307.10159) (Jul 2023), \url{http: //arxiv.org/abs/2307.10159}, arXiv:2307.10159 [cs]

    2023 arXiv

  27. [35]

    https://doi.org/10.48550/arXiv.2302.13848, \url{http://arxiv.org/abs/2302.13848}, arXiv:2302.13848 [cs] 18 A.L

    Wei, Y., Zhang, Y., Ji, Z., Bai, J., Zhang, L., Zuo, W.: Elite: Encoding vi- sual concepts into textual embeddings for customized text-to-image genera- tion (arXiv:2302.13848) (Aug 2023). https://doi.org/10.48550/arXiv.2302.13848, \url{http://arxiv.org/abs/2302.13848}, arXiv:2...

    2023 doi arXiv

  28. [36]

    https://doi.org/10.48550/arXiv.2303.14420, \url{http://arxiv.org/abs/ 2303.14420}, arXiv:2303.14420 [cs]

    Wu, X., Sun, K., Zhu, F., Zhao, R., Li, H.: Human preference score: Better aligning text-to-image models with human preference (arXiv:2303.14420) (Aug 2023). https://doi.org/10.48550/arXiv.2303.14420, \url{http://arxiv.org/abs/ 2303.14420}, arXiv:2303.14420 [cs]

    2023 doi arXiv

  29. [37]

    Transactions on Machine Learning Research (2022),\url{https:// openreview.net/forum?id=AFDcYJKhND}, featured Certification

    Yu, J., Xu, Y., Koh, J.Y., Luong, T., Baid, G., Wang, Z., Vasudevan, V., Ku, A., Yang, Y., Ayan, B.K., Hutchinson, B., Han, W., Parekh, Z., Li, X., Zhang, H., Baldridge, J., Wu, Y.: Scaling autoregressive models for content-rich text-to-image generation. Transactions on Machin...

    2022

  30. [38]

    The intersection of these primary colors producesOrange, Green, andViolet

    Zhang, N., Li, L., Chen, X., Deng, S., Bi, Z., Tan, C., Huang, F., Chen, H.: Dif- ferentiable prompt makes pre-trained language models better few-shot learners (arXiv:2108.13161) (May 2022),\url{http://arxiv.org/abs/2108.13161},DOI= {10.48550/arXiv.2108.13161}, arXiv:2108.1316...

    2022 doi arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.