Pith. sign in

REVIEW 3 major objections 7 minor 41 references

A closed-loop feedback controller rebalances rare attributes and base objects during diffusion sampling without training.

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

2026-07-11 08:59 UTC pith:YOJPSUSF

load-bearing objection Solid training-free closed-loop controller for rare attribute–object binding; gains look real, CLIP-oracle risk is real but not fatal. the 3 major comments →

arxiv 2607.05088 v1 pith:YOJPSUSF submitted 2026-07-06 cs.CV

RADIANCE: Relative Adaptive Denoising with IP-Adapter for Novel Concept Enhancement

Zi-Xiang Ni , Bo-Lun Huang , Teng-Fang Hsiao , Bo-Kai Ruan , Hong-Han Shuai This is my paper
classification cs.CV
keywords text-to-image diffusionrare concept generationcompositional balanceIP-Adaptertraining-free guidanceclosed-loop feedbackCLIP monitoring
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.

Text-to-image diffusion models still fail on rare attribute–object pairs such as “a thorny dolphin,” either dropping the attribute or letting it erase the object. The paper traces the failure to unbalanced trajectories during denoising and shows that the imbalance can be corrected at inference time. RADIANCE monitors intermediate images with CLIP similarity against simple reference pictures of the object and of each attribute, then applies positive or negative IP-Adapter scales as a restoring force that strengthens the lagging concept and suppresses the dominant one. A lightweight scheduler times these corrections, and two further modules (delayed activation and layer-wise alternation) keep multiple rare objects from fusing prematurely. The result is higher compositional fidelity and higher effective throughput on RareBench and T2I-CompBench while remaining training-free and only modestly slower than the backbone.

Core claim

Rare-concept failures arise from a lack of compositional balance along the denoising trajectory, and that balance can be restored on the fly by treating CLIP similarity signals as feedback that drives signed IP-Adapter scales, without any model training or gradient steps.

What carries the argument

The Bidirectional Scale Controller (BSC): at each step it receives object and attribute similarities from the Compositional Similarity Monitor and sets positive or negative IP-Adapter scales that act as a restoring force, reinforcing the lagging concept while actively suppressing the dominant one.

Load-bearing premise

CLIP cosine similarity between one decoded intermediate image and a few hand-made reference pictures is a reliable enough real-time signal of object-versus-attribute balance to safely drive the signed scale updates.

What would settle it

On a held-out set of rare prompts, replace the CLIP monitor with random or constant scales (or with a deliberately inverted feedback rule) and measure whether compositional alignment and human preference scores still rise above the backbone and R2F baselines; if they do not, the feedback claim collapses.

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

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 / 7 minor

Summary. The paper proposes RADIANCE, a training-free inference-time controller for rare attribute–object composition in T2I diffusion models. It attributes failures (omission, semantic drift, premature fusion) to imbalance along the denoising trajectory and introduces three modules: a Compositional Similarity Monitor (CSM) that scores intermediate VAE-decoded images against object/attribute reference images with CLIP (Eq. 1); a Bidirectional Scale Controller (BSC) that injects positive or negative IP-Adapter scales as a restoring force (Eq. 2); and a Feedback Guidance Scheduler (FGS) that maps consecutive similarity trends to discrete scale pairs (Eq. 3). Multi-object prompts are handled by first synthesizing single-object references, then Delayed Adapter Activation (DAA) and Layer-wise Alternating Guidance (LAG). On RareBench (SD 3.5) the method reports overall 84.7 vs R2F 81.7, with supporting T2I-CompBench numbers, ablations of BSC/FGS/T_early/DAA/LAG, a 52-person preference study, and pipelined latency measurements.

Significance. Rare attribute–object binding remains a practical failure mode even for strong MM-DiT backbones, so a training-free closed-loop controller that reuses IP-Adapter and CLIP is of clear applied interest. Strengths include: (i) an explicit feedback formulation rather than one-shot prompt rewriting; (ii) the empirical observation and use of negative IP-Adapter scales; (iii) multi-object extensions (DAA/LAG) with targeted ablations; (iv) architecture transfer (SD 1.5 and SD 3.5) without re-tuning; and (v) user preference plus wall-clock analysis that frames success rate as effective throughput. If the monitor-driven signed updates are reliable, the work offers a reusable control pattern for compositional T2I beyond rare concepts. The contribution is primarily empirical and systems-level rather than theoretical.

major comments (3)
  1. Sec. 3.2–3.4, Eqs. (1)–(3): The central closed-loop claim rests on CLIP cosine trends between a single decoded intermediate ˆx_t and hand-synthesized references being a trustworthy oracle for large-magnitude signed scale flips (±1). Early latents remain heavily noised, and Fig. 3 itself shows that high joint similarity can reflect co-presence rather than binding. The manuscript never reports monitor accuracy (how often the chosen sign matches a human or VQA judgment of which concept is lagging), the rate of sign flips, or qualitative artifacts from inverted corrections. Table 4 shows that adaptive scales beat fixed scales and that T_early is somewhat stable, but that does not calibrate the sensor. Without such analysis (or a softer continuous controller), it is hard to separate a robust feedback principle from a well-tuned open-loop schedule plus IP-Adapter references.
  2. Sec. 3.2, footnote 1: Reference images for both objects and attributes are “obtained by a brief synthesis with the original backbone” and then used both as CSM targets and as IP-Adapter conditions. The protocol is underspecified: prompts used for pure attributes (e.g., “thorny”, “zebra striped”), number of samples, selection criteria, and failure handling when the backbone itself omits the rare attribute are not stated. Because the same model that struggles on rare pairings supplies the monitoring targets, reference quality is a load-bearing dependency. The paper should document the generation recipe, report sensitivity to reference choice, and discuss failure cases when references are weak or entangled.
  3. Sec. 4.1–4.2, Tables 1–2: Primary RareBench scores are GPT-4o judgments; T2I-CompBench uses BLIP. The scoring prompt, aggregation, and agreement with humans (beyond the separate preference study) are not reported in the main text. Given that the claimed gains over R2F are modest (~3 overall points on RareBench) and that Relation remains weak, the paper needs either (a) transparent automatic-eval protocol plus inter-rater / model–human agreement, or (b) stronger statistical treatment (variance over seeds, significance). Otherwise the headline “consistently enhances” claim is only partially supported by the automatic tables.
minor comments (7)
  1. Title vs. body expansion disagree: title uses “Relative Adaptive Denoising…”, §1 expands RADIANCE as “Rare Attribute Diffusion with Adaptive Novel Concept Enhancement”. Align naming.
  2. Sec. 3.4 and elsewhere: “Appendix??” placeholders remain; multi-attribute scheduler details promised there are not visible in the provided manuscript.
  3. Eq. (3) indexes (s^a_{t−1}, s^o_{t−1}) while comparing sim_t to sim_{t+1}; the time-index convention for “next-step” scales is easy to misread and should be clarified.
  4. Fig. 5 is hard to read in grayscale (overlapping similarity curves); add markers or separate panels.
  5. Table 1 mixes † numbers from R2F with newly run numbers; state seed, step count, and guidance settings for every re-run baseline to ensure fair comparison.
  6. Several typos and spacing issues (e.g., “Text-to-image(T2I)diffusion”, “Despitetheadvances”, “metohds” in Table 1 caption).
  7. Limitations correctly note weak spatial/relational control; a short quantitative breakdown of Relation failures would make that claim more actionable.

Circularity Check

0 steps flagged

No significant circularity: empirical controller evaluated on external benchmarks with ablated hyperparameters.

full rationale

RADIANCE is a training-free inference procedure whose three modules (CSM Eq. 1, BSC Eq. 2, FGS Eq. 3) are defined independently of any evaluation metric. The central claims are purely empirical improvements on the public RareBench and T2I-CompBench suites against independently published baselines (R2F, SD 3.5, FLUX, etc.). Hyper-parameters (T_early = 15, discrete scales ±1) are chosen by ablation (Table 4) rather than fitted to the final GPT-4o / BLIP scores. No equation reduces a reported quantity to a quantity defined by the method itself; self-citations to overlapping-author prior work are incidental and not load-bearing for uniqueness or for the performance numbers. The derivation chain therefore contains no self-definitional, fitted-as-prediction, or self-citation-forced steps.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 3 invented entities

The central claim rests on a small set of engineering choices (CLIP as monitor, fixed ±1 scales, hand-chosen T_early, reference images generated by the same backbone) rather than deep mathematical axioms. These free parameters and domain assumptions are explicit and ablated, but they are still required for the reported gains.

free parameters (3)
  • T_early = 15 (default)
    Timestep that switches the feedback policy from joint formation to object-preservation priority; default 15 (18 for relational prompts), chosen by ablation on RareBench.
  • IP-Adapter scale magnitudes = ±1
    Fixed positive/negative values (±1) used by the Bidirectional Scale Controller; not learned, selected for simplicity and shown effective in Fig. 4.
  • T_delay (DAA)
    Number of initial steps with IP-Adapter disabled for multi-object prompts; required to prevent premature fusion.
axioms (3)
  • domain assumption CLIP cosine similarity between a decoded intermediate latent and a single reference image is a faithful real-time indicator of whether an object or attribute is dominating the generation.
    Stated and used throughout Sec. 3.2–3.4; no formal proof that CLIP tracks the desired semantic balance under all rare-concept distributions.
  • domain assumption Negative IP-Adapter scales suppress the corresponding concept without introducing uncontrolled artifacts or collapsing the overall sample quality.
    Empirically illustrated in Fig. 4 and used as the restoring force in Eq. 3; treated as given rather than derived.
  • ad hoc to paper Reference images for objects and attributes can be obtained by a brief synthesis with the original backbone and remain valid targets for both monitoring and guidance.
    Footnote in Sec. 3.2; quality of these references directly affects both CSM signals and IP-Adapter conditioning.
invented entities (3)
  • Compositional Similarity Monitor (CSM) no independent evidence
    purpose: Provides per-step object/attribute similarity signals from intermediate latents.
    New modular component; independent evidence is only the empirical correlation shown in Fig. 3.
  • Bidirectional Scale Controller (BSC) no independent evidence
    purpose: Applies signed IP-Adapter scales as a restoring force.
    New control policy built on existing IP-Adapter; no external validation outside this paper.
  • Feedback Guidance Scheduler (FGS) no independent evidence
    purpose: Coordinates scale updates across the denoising trajectory according to similarity trends.
    Paper-specific policy (Eq. 3); independent evidence limited to ablations.

pith-pipeline@v1.1.0-grok45 · 19071 in / 2794 out tokens · 22110 ms · 2026-07-11T08:59:03.227725+00:00 · methodology

0 comments
read the original abstract

Text-to-image (T2I) diffusion models have achieved striking progress but still struggle to synthesize rare concepts involving unusual attribute-object pairings, often resulting in concept omission or semantic drift where a dominant entity overwhelms the generation. Tracing these failures to a lack of compositional balance during the denoising trajectory, we propose RADIANCE, a training-free framework that treats inference as a closed-loop feedback process. RADIANCE augments pretrained backbones with three modular components: (1) a Compositional Similarity Monitor (CSM) that tracks the emergence of objects and attributes in intermediate latents via CLIP-based feedback; (2) a Bidirectional Scale Controller (BSC) that applies a reactive "restoring force" using positive and negative IP-Adapter scales to rebalance biased trajectories; and (3) a Feedback Guidance Scheduler (FGS) that coordinates these updates across timesteps without additional training. We further extend the framework to multi-object prompts via Delayed Adapter Activation (DAA) and Layer-wise Alternating Guidance (LAG) to prevent premature concept fusion. By overlapping monitoring and denoising through pipelined execution, RADIANCE maintains competitive latency while significantly enhancing the per-sample success rate and effective throughput. Experiments on RareBench and T2I-CompBench demonstrate that RADIANCE consistently enhances compositional alignment and perceptual quality over state-of-the-art baselines.

Figures

Figures reproduced from arXiv: 2607.05088 by Bo-Kai Ruan, Bo-Lun Huang, Hong-Han Shuai, Teng-Fang Hsiao, Zi-Xiang Ni.

Figure 1
Figure 1. Figure 1: Generated samples from RADIANCE across different rare concept [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the RADIANCE pipeline. At each denoising step, the Feed￾back Guidance Scheduler (Sec. 3.4) computes adaptive guidance. The resulting scale is then applied by the Bidirectional Scale Controller (Sec. 3.3) to inject reference in￾formation into the diffusion process. Finally, multiple rare-concept outputs are merged via the Multiple Rare Concepts Merging (Sec. 3.5) that composed of DAA and LAG. To… view at source ↗
Figure 3
Figure 3. Figure 3: Visualization of reference images and compositional outcomes for [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of IP-Adapter guidance with different scales for the [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Step-wise visualization of adaptive guidance. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Visualization of the effect of DAA and LAG. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative comparison on RareBench. RADIANCE produces more ac￾curate and coherent results compared to baselines. Comparison Basic Advanced Overall Pick Win Pick Win Pick Win Our vs. SD 3.5 95.5% 100% 61.5% 60% 74.3% 75.0% Our vs. R2F 84.6% 100% 60.0% 80% 69.2% 87.5% Our vs. FLUX 79.5% 66.7% 71.5% 100% 74.5% 87.5% Method Time (sec) SD3.5 7.36 R2F 7.38 Ours (w/o parallel) 16.48 Ours (parallel)∗ 9.34 [PITH_… view at source ↗

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

41 extracted references · 10 linked inside Pith

  1. [1]

    Gpt-4oapi documentation.https://platform.openai.com/docs/models/gpt-4o, accessed: 2025-11-12

    2025

  2. [2]

    In: European Conference on Computer Vision

    Butt, M.A., Wang, K., Vazquez-Corral, J., van de Weijer, J.: Colorpeel: Color prompt learning with diffusion models via color and shape disentanglement. In: European Conference on Computer Vision. pp. 456–472. Springer (2024)

    2024

  3. [3]

    ACM trans- actions on Graphics (TOG)42(4), 1–10 (2023)

    Chefer, H., Alaluf, Y., Vinker, Y., Wolf, L., Cohen-Or, D.: Attend-and-excite: Attention-based semantic guidance for text-to-image diffusion models. ACM trans- actions on Graphics (TOG)42(4), 1–10 (2023)

    2023

  4. [4]

    arXiv preprint arXiv:2310.00426 (2023)

    Chen, J., Yu, J., Ge, C., Yao, L., Xie, E., Wu, Y., Wang, Z., Kwok, J., Luo, P., Lu, H., et al.: Pixart-α: Fast training of diffusion transformer for photorealistic text-to-image synthesis. arXiv preprint arXiv:2310.00426 (2023)

    Pith/arXiv arXiv 2023

  5. [5]

    arXiv preprint arXiv:2501.17811 (2025)

    Chen, X., Wu, Z., Liu, X., Pan, Z., Liu, W., Xie, Z., Yu, X., Ruan, C.: Janus-pro: Unified multimodal understanding and generation with data and model scaling. arXiv preprint arXiv:2501.17811 (2025)

    Pith/arXiv arXiv 2025

  6. [6]

    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

  7. [7]

    Advances in neural information processing systems34, 19822–19835 (2021)

    Ding, M., Yang, Z., Hong, W., Zheng, W., Zhou, C., Yin, D., Lin, J., Zou, X., Shao, Z., Yang, H., et al.: Cogview: Mastering text-to-image generation via transformers. Advances in neural information processing systems34, 19822–19835 (2021)

    2021

  8. [8]

    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

  9. [9]

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

    Guo, X., Liu, J., Cui, M., Li, J., Yang, H., Huang, D.: Initno: Boosting text- to-image diffusion models via initial noise optimization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 9380– 9389 (2024)

    2024

  10. [10]

    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

  11. [11]

    arXiv preprint arXiv:2503.15283 (2025)

    Hsiao, T.F., Ruan, B.K., Wu, Y.L., Lin, T.L., Shuai, H.H.: Tf-ti2i: Training-free text-and-image-to-image generation via multi-modal implicit-context learning in text-to-image models. arXiv preprint arXiv:2503.15283 (2025)

    Pith/arXiv arXiv 2025

  12. [12]

    arXiv preprint arXiv:2403.05135 (2024)

    Hu, X., Wang, R., Fang, Y., Fu, B., Cheng, P., Yu, G.: Ella: Equip diffusion models with llm for enhanced semantic alignment. arXiv preprint arXiv:2403.05135 (2024)

    Pith/arXiv arXiv 2024

  13. [13]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)

    Huang, K., Duan, C., Sun, K., Xie, E., Li, Z., Liu, X.: T2i-compbench++: An en- hanced and comprehensive benchmark for compositional text-to-image generation. IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)

    2025

  14. [14]

    In: Thirty- seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track (2023),https://openreview.net/forum?id=weHBzTLXpH RADIANCE 17

    Huang, K., Sun, K., Xie, E., Li, Z., Liu, X.: T2i-compbench: A comprehensive benchmark for open-world compositional text-to-image generation. In: Thirty- seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track (2023),https://openreview.net/forum?id=weHBzTLXpH RADIANCE 17

    2023

  15. [15]

    Advances in Neural Information Processing Systems36, 78723–78747 (2023)

    Huang, K., Sun, K., Xie, E., Li, Z., Liu, X.: T2i-compbench: A comprehensive benchmark for open-world compositional text-to-image generation. Advances in Neural Information Processing Systems36, 78723–78747 (2023)

    2023

  16. [16]

    arXiv preprint arXiv:2302.09778 (2023)

    Huang, L., Chen, D., Liu, Y., Shen, Y., Zhao, D., Zhou, J.: Composer: Cre- ative and controllable image synthesis with composable conditions. arXiv preprint arXiv:2302.09778 (2023)

    Pith/arXiv arXiv 2023

  17. [17]

    International Con- ference on Learning Representations (ICLR) (2014).https://doi.org/10.48550/ arXiv.1312.6114

    Kingma, D.P., Welling, M.: Auto-encoding variational bayes. International Con- ference on Learning Representations (ICLR) (2014).https://doi.org/10.48550/ arXiv.1312.6114

    Pith/arXiv arXiv 2014

  18. [18]

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

    2024

  19. [19]

    arXiv preprint arXiv:2503.09864 (2025)

    Laria, H., Gomez-Villa, A., Qin, J., Butt, M.A., Raducanu, B., Vazquez-Corral, J., van de Weijer, J., Wang, K.: Leveraging semantic attribute binding for free-lunch color control in diffusion models. arXiv preprint arXiv:2503.09864 (2025)

    arXiv 2025

  20. [20]

    In: International confer- ence on machine learning

    Li, J., Li, D., Xiong, C., Hoi, S.: Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In: International confer- ence on machine learning. pp. 12888–12900. PMLR (2022)

    2022

  21. [21]

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

    Li,Y.,Liu,H.,Wu,Q.,Mu,F.,Yang,J.,Gao,J.,Li,C.,Lee,Y.J.:Gligen:Open-set grounded text-to-image generation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 22511–22521 (2023)

    2023

  22. [22]

    arXiv preprint arXiv:2305.13655 (2023)

    Lian, L., Li, B., Yala, A., Darrell, T.: Llm-grounded diffusion: Enhancing prompt understanding of text-to-image diffusion models with large language models. arXiv preprint arXiv:2305.13655 (2023)

    Pith/arXiv arXiv 2023

  23. [23]

    In: International Conference on Learning Repre- sentations (2023)

    Liu, X., Gong, C., qiang liu: Flow straight and fast: Learning to generate and transfer data with rectified flow. In: International Conference on Learning Repre- sentations (2023)

    2023

  24. [24]

    In: Proceedings of the AAAI Conference on Artificial In- telligence

    Mañas, O., Krojer, B., Agrawal, A.: Improving automatic vqa evaluation using large language models. In: Proceedings of the AAAI Conference on Artificial In- telligence. vol. 38, pp. 4171–4179 (2024)

    2024

  25. [25]

    In: Proceedings of the AAAI conference on artificial intelligence

    Mou, C., Wang, X., Xie, L., Wu, Y., Zhang, J., Qi, Z., Shan, Y.: T2i-adapter: Learning adapters to dig out more controllable ability for text-to-image diffusion models. In: Proceedings of the AAAI conference on artificial intelligence. vol. 38, pp. 4296–4304 (2024)

    2024

  26. [26]

    arXiv preprint arXiv:2112.10741 (2021)

    Nichol, A., Dhariwal, P., Ramesh, A., Shyam, P., Mishkin, P., McGrew, B., Sutskever, I., Chen, M.: Glide: Towards photorealistic image generation and editing with text-guided diffusion models. arXiv preprint arXiv:2112.10741 (2021)

    Pith/arXiv arXiv 2021

  27. [27]

    In: The Thirteenth International Conference on Learning Representa- tions

    Park, D., Kim, S., Moon, T., Kim, M., Lee, K., Cho, J.: Rare-to-frequent: Unlock- ing compositional generation power of diffusion models on rare concepts with llm guidance. In: The Thirteenth International Conference on Learning Representa- tions

  28. [28]

    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

  29. [29]

    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

  30. [30]

    Journal of machine learning research21(140), 1–67 (2020) 18 Z.-X

    Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., Liu, P.J.: Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of machine learning research21(140), 1–67 (2020) 18 Z.-X. Ni et al

    2020

  31. [31]

    Advances in Neural Information Processing Systems36, 3536–3559 (2023)

    Rassin, R., Hirsch, E., Glickman, D., Ravfogel, S., Goldberg, Y., Chechik, G.: Lin- guistic binding in diffusion models: Enhancing attribute correspondence through attention map alignment. Advances in Neural Information Processing Systems36, 3536–3559 (2023)

    2023

  32. [32]

    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

  33. [33]

    arXiv preprint arXiv:2505.20808 (2025)

    Ruan, B.K., Ni, Z.X., Huang, B.L., Hsiao, T.F., Shuai, H.H.: Not all thats rare is lost: Causal paths to rare concept synthesis. arXiv preprint arXiv:2505.20808 (2025)

    arXiv 2025

  34. [34]

    Advances in neural information processing systems35, 36479–36494 (2022)

    Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E.L., Ghasemipour, K., Gontijo Lopes, R., Karagol Ayan, B., Salimans, T., et al.: Photorealistic text- to-image diffusion models with deep language understanding. Advances in neural information processing systems35, 36479–36494 (2022)

    2022

  35. [35]

    In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Sun, Q., Cui, Y., Zhang, X., Zhang, F., Yu, Q., Wang, Y., Rao, Y., Liu, J., Huang, T., Wang, X.: Generative multimodal models are in-context learners. In: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 14398–14409 (2024)

    2024

  36. [36]

    Advances in neural information pro- cessing systems30(2017)

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information pro- cessing systems30(2017)

    2017

  37. [37]

    In: The Thirteenth International Conference on Learning Representations

    Xie, J., Mao, W., Bai, Z., Zhang, D.J., Wang, W., Lin, K.Q., Gu, Y., Chen, Z., Yang, Z., Shou, M.Z.: Show-o: One single transformer to unify multimodal under- standing and generation. In: The Thirteenth International Conference on Learning Representations

  38. [38]

    In: Forty- first International Conference on Machine Learning (2024)

    Yang, L., Yu, Z., Meng, C., Xu, M., Ermon, S., Cui, B.: Mastering text-to-image diffusion: Recaptioning, planning, and generating with multimodal llms. In: Forty- first International Conference on Machine Learning (2024)

    2024

  39. [39]

    ACM Comput

    Yang,L.,Zhang,Z.,Song,Y.,Hong,S.,Xu,R.,Zhao,Y.,Zhang,W.,Cui,B.,Yang, M.H.: Diffusion models: A comprehensive survey of methods and applications. ACM Comput. Surv.56(4) (Nov 2023).https://doi.org/10.1145/3626235, https://doi.org/10.1145/3626235

    doi:10.1145/3626235 2023

  40. [40]

    Ye, H., Zhang, J., Liu, S., Han, X., Yang, W.: Ip-adapter: Text compatible image prompt adapter for text-to-image diffusion models (2023)

    2023

  41. [41]

    arXiv preprint arXiv:2406.07844 (2024)

    Zarei, A., Rezaei, K., Basu, S., Saberi, M., Moayeri, M., Kattakinda, P., Feizi, S.: Improving compositional attribute binding in text-to-image generative models via enhanced text embeddings. arXiv preprint arXiv:2406.07844 (2024)

    Pith/arXiv arXiv 2024