Pith. sign in

REVIEW 3 major objections 5 minor 4 cited by

SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Training

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read SnapGen is a 379M-parameter text-to-image model that, for the first time, generates 1024x1024 images on a mobile phone in about 1.4 seconds while matching or beating billion-parameter models on standard benchmarks.

desk verdict Genuinely new mobile 1K T2I result, but the headline 1.4s latency does not survive contact with the paper's own per-component numbers; worth refereeing after the measurement protocol is clarified. read the letter →

arxiv 2412.09619 v1 pith:D7I7HD7X submitted 2024-12-12 cs.CV

Dongting Hu , Jierun Chen , Xijie Huang , Huseyin Coskun , Arpit Sahni , Aarush Gupta , Anujraaj Goyal , Dishani Lahiri
show 11 more authors
Rajesh Singh Yerlan Idelbayev Junli Cao Yanyu Li Kwang-Ting Cheng S.-H. Gary Chan Mingming Gong Sergey Tulyakov Anil Kag Yanwu Xu Jian Ren
This is my paper · ORCID
classification cs.CV
keywords text-to-imagegenerationdiffusionmodelsknowledgedistillationstepmobiledevicesefficientUNetflowmatchinglatent
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 asks whether a text-to-image diffusion model can be trained from scratch to create 1024x1024 images on a phone, and answers yes. Its model, SnapGen, packs a 379M-parameter denoising UNet, scores 0.66 on GenEval and 81.1 on DPG-Bench, and beats several billion-parameter models on those benchmarks while running on an iPhone 16 Pro-Max in roughly 1.4 seconds. The route to this result is a set of architectural cuts, a tiny decoder, cross-architecture knowledge distillation from a much larger diffusion transformer, and adversarial step distillation for few-step generation. If the result stands, it would mean high-quality text-to-image generation is feasible privately and offline on ordinary mobile hardware.

What carries the argument

The argument is carried by three coupled components. The efficient UNet keeps self-attention only in the lowest-resolution stage, replaces convolutions with expanded separable convolutions, trims feed-forward expansion, uses multi-query attention, injects text conditions from the first stage, and adds QK-RMSNorm and 2D rotary position embeddings; each change is ablated on ImageNet FID, latency, and FLOPs. The decoder removes attention entirely, keeps minimal GroupNorm, and uses separable convolutions to reach 1.38M parameters. Training combines flow matching with multi-level knowledge distillation that matches both the teacher's velocity output and its last-layer features, reweighted by a timestep-aware scaling function, and then applies adversarial step distillation to reach 4-8 steps.

What would settle it

Run the released model on an iPhone 16 Pro-Max and time a 1024x1024 generation: a median above the reported 1.2-2.3 second range would falsify the deployment claim. Independently, training the same architecture and recipe from scratch should reproduce GenEval near 0.66 and DPG-Bench near 81.1; a large miss would show the public numbers depend on undisclosed data or evaluation details.

Watch

Extended reading notes

Core claim

The central claim is that a small UNet, deliberately thinned and shortened, can be the backbone of mobile-runnable text-to-image generation without sacrificing quality. On ImageNet-1K, the 372M-parameter model reaches an FID of 2.06 at 256x256, matching SiT-XL while using 45% fewer parameters and 68% fewer FLOPs. For open-domain text-to-image, the 379M-parameter model scores 0.66 on GenEval and 81.1 on DPG-Bench, exceeding SDXL, Playground, and IF-XL, and its step-distilled 4- and 8-step variants keep most of that quality. The tiny 1.38M-parameter decoder reconstructs images with PSNR close to SD3's decoder while being 36x smaller and 54x faster, and the full 1024x1024 pipeline runs on-device in 1.2-2.3 seconds.

Load-bearing premise

The whole recipe assumes that a much larger teacher's output and internal features, compared in the shared latent space, can push a much smaller student to the same quality, and that if this cross-architecture transfer breaks at high resolution, the claimed quality would not appear.

Editorial extensions

If this is right

  • A text-to-image model under 400M parameters can outperform billion-parameter models on prompt-following benchmarks, so model scale is not the only path to quality.
  • On-device 1024x1024 generation in about 1.4 seconds makes private, offline image generation practical on consumer phones.
  • Step-distilled 4- and 8-step inference retains most of the 28-step quality (GenEval 0.61-0.63 vs 0.66), so mobile latency and quality do not have to trade off sharply.
  • The architectural ablations show that removing self-attention from high-resolution stages and using separable convolutions can lower FLOPs and latency while improving FID.
  • Distilling from a large DiT teacher into a compact UNet student, in a shared latent space, transfers capability across heterogeneous architectures.

Reading between the lines

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

  • Beyond the paper's claims, the recipe suggests that a compact UNet distilled from a large DiT teacher in a shared latent space could generalize to other flow-matching generators (video, 3D, audio), where such teachers exist.
  • The paper fixes the teacher feature layer to the last transformer layer; ablating earlier or later teacher layers would test whether that choice is optimal for heterogeneous distillation.
  • The latency figures are for a single generation; repeated-run energy and thermal measurements would complete the mobile-deployment picture.
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

3 major / 5 minor

Summary. The paper presents SnapGen, a compact text-to-image diffusion model comprising a 379M-parameter efficient UNet, a 1.38M-parameter decoder, multi-level knowledge distillation from SD3.5-Large, and adversarial step distillation. The authors report an ImageNet-1K FID of 2.06 at 256x256 with 372M parameters, GenEval 0.66 and DPG-Bench 81.1 at 1024x1024, human evaluation results competitive with SD3.5-Large, and claim on-device 1024x1024 generation on an iPhone 16 Pro-Max in approximately 1.4 seconds.

Significance. If the claims hold, this is a significant step for on-device text-to-image generation: a carefully ablated efficient architecture, a tiny high-quality decoder, and a training recipe that transfers knowledge from a much larger DiT teacher are all useful contributions. The ablations in Fig. 3 and Supp. Tab. 3 are clear and internally consistent, and the ImageNet FID result at 372M parameters is strong. The main caveats are that the headline mobile latency claim is not backed by a complete measurement protocol and that the 379M-parameter number refers only to the UNet backbone while the deployed system also carries large text encoders. These issues are fixable in revision but currently prevent the central mobile-feasibility claim from being verified.

major comments (3)
  1. [Abstract, Sec. 1, Sec. 3.2 (Discussion of Total On-Device Latency), Sec. 4.1 (Hyper-parameters)] The headline claim that SnapGen generates 1024x1024 images on a mobile device in ~1.4 s is not supported by the timing data given in the paper. Sec. 3.2 reports a 119 ms decoder and a 274 ms per-step UNet latency, which gives 4x274 + 119 = 1,215 ms for a 4-step run before any VAE encoding, text-encoder inference, or scheduling overhead. The training recipe uses classifier-free guidance with 46.4% text-encoder dropout (Sec. 4.1), but the paper never states whether CFG is applied in the timed mobile demo; with standard CFG the UNet cost doubles and a 4-step run becomes approximately 2.3 s. In addition, Fig. 3 lists a final per-step UNet latency of 397 ms on an iPhone 15 Pro while Sec. 3.2 uses 274 ms on an iPhone 16 Pro-Max, and the paper does not explain whether this difference is due to hardware, quantization, step count, or other settings. No measurement protocol (device state, warm-up, model loading, text-encoder inclusion, CFG on/off) is provided for the 'around 1.4 seconds' figure, so the central mobile-feasibility contribution cannot be independently verified or reproduced.
  2. [Sec. 4.1 (Model Details) and Table 3] The repeated claim that SnapGen has 'merely 379M parameters' refers only to the UNet backbone, as the Table 3 caption states, but the deployed pipeline described in Sec. 4.1 also requires CLIP-L, CLIP-G, and Gemma-2-2b text encoders, adding more than 2.5B parameters to the full system. Since the paper motivates the work by on-device deployment and by size comparisons with SDXL (2.6B) and IF-XL (5.5B), the authors should report the total on-device model size and memory footprint for the configurations they actually deploy, and state which text-encoder configuration was used for the benchmark results versus the mobile demo. Without this, the '7x smaller than SDXL' comparison conflates backbone size with full-system size.
  3. [Sec. 3.3, Eq. (7)] The timestep-aware weighting in Eq. (7) uses the ratio |L_task^t| / |L_kd^t|, but the paper does not specify how these magnitudes are computed in practice: per-batch estimates, running averages, or a fixed schedule estimated from a reference run. Since loss magnitudes change over the course of training, a fixed ratio could become stale, and because lambda(t) is described as a logit-normal density rather than a normalized probability, it is not clear that the weights on the two loss terms remain nonnegative. Please specify the exact computation and how it is updated during training.
minor comments (5)
  1. [Fig. 3] The three vertical bars for each design choice are not labeled with a legend indicating which bar corresponds to latency, FLOPs, and parameter count; please add clear labels or a legend.
  2. [Table 3] The table does not include the SD3.5-Large teacher model even though the human evaluation in Fig. 8 compares against it; adding the teacher's GenEval, DPG-Bench, CLIP, and ImageReward scores would make the comparison complete.
  3. [Sec. 3.2 and Fig. 4] The decoder latency is reported on both ANE and GPU in Table 2, but it is not stated which hardware path the 119 ms decoder time in Sec. 3.2 refers to; please clarify.
  4. [Supplementary Sec. G] The training-data description says the collection and filtering pipeline 'follows the approach described by Kag et al. [32]' but does not report dataset size, filtering thresholds, or deduplication details; please provide a self-contained summary for reproducibility.
  5. [Sec. 3.3, Eq. (5)] The feature-distillation loss is defined over layer pairs (l_T, l), but the text says only that the last layer is used in both models; please specify the exact layer indices and the output dimensions before and after the projector.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: central claims rest on external benchmarks and measured latencies; self-citations are contextual, not load-bearing.

full rationale

The paper's derivation chain is not circular. The efficient-UNet and decoder claims are supported by measured ImageNet FID, PSNR, FLOPs, and on-device latency tables (Tabs. 1-2, Figs. 3-4); the training method (Eqs. 1-9) is a set of objectives, not a fit to the reported GenEval/DPG scores; timestep-aware scaling (Eq. 7) normalizes loss magnitudes by construction but is an optimization weighting scheme, and the paper separately ablates it (Tab. 3) on held-out benchmarks. The T2I quality claims (GenEval 0.66, DPG-Bench 81.1, ImageReward 1.32, human study) are measured against external benchmarks and external baselines (SDXL, Sana, IF-XL), not derived from the teacher SD3.5-Large by an equation that collapses into the input. Self-citations (SnapFusion [43], MobileDiffusion [88], BitsFusion [69], ASCAN [32]) supply prior baselines, architectural inspiration, and data-pipeline details; none is invoked as a uniqueness theorem or as the sole justification of a predicted quantity. The abstract's "around 1.4 seconds" versus Sec. 3.2's "1.2 ∼ 2.3s" is a potential reporting/protocol inconsistency about text-encoder and CFG inclusion, but it is not a circular reduction of a prediction to a fitted input. Therefore no circular step is identified.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claims rest on hand-tuned architecture hyperparameters and the assumed transferability of the teacher's latent space and the internal dataset. No invented physical or conceptual entities are introduced beyond standard architectural components.

free parameters (6)
  • UNet width and depth [256,512,896] channels, [0,2,4] transformer blocks = chosen via ablations in Fig. 3
    Architecture capacity is selected by hand to balance FLOPs, latency, and ImageNet FID; the final model depends on this configuration.
  • SepConv expansion ratio = 2
    Set to recover FID after replacing standard convolutions with separable convolutions; ablation shown in Fig. 3.
  • FFN expansion ratio = 3 (down from 4x with gated units)
    Chosen to cut parameters and FLOPs by 12% with comparable FID, following MobileDiffusion.
  • Timestep shift for flow matching = 3
    Used for both training and inference (Sec. 4); follows SD3's recipe and affects high-resolution sample quality.
  • Text-encoder dropout probability = 46.4% per encoder
    Set so that roughly 10% of steps train unconditionally for classifier-free guidance (Sec. 4).
  • Logit-normal timestep sampling location and scale = (0,1)
    Standard SD3 hyperparameter; also defines the lambda(t) weighting in the timestep-aware scaling (Eq. 7).
assumptions (4)
  • domain assumption The SD3.5-Large and SD3.5-Large-Turbo teachers provide effective supervision in the shared SD3 autoencoder latent space.
    Sec. 3.3 relies on aligned flow-matching objectives and the shared AE latent space to justify direct output and feature distillation from a DiT teacher to a UNet student; if the latent spaces or objectives mismatched, the KD losses would not transfer cleanly.
  • domain assumption The internal training dataset built following ASCAN [32] is sufficient to train a T2I model from scratch at 256 to 1024 resolution.
    Sec. 4 and Sec. G state that the data pipeline follows Kag et al. [32]; the dataset itself is not released, so the model's quality depends on this unverifiable assumption.
  • domain assumption Rectified flow matching (Eqs. 1-2) is a valid generative training objective for a UNet backbone.
    The objective is inherited from prior work [47,50] and used throughout; if the v-prediction target is inappropriate for this architecture, the training recipe would not work.
  • domain assumption The fixed 8-channel SD3 autoencoder encoder is a suitable latent space, and the decoder can be trained independently with the encoder fixed.
    Sec. 3.2 fixes the SD3 encoder and trains a new tiny decoder on 256x256 patches; the assumption is that the encoder's latent is generic enough for both the teacher and student.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Training." pith.science (2026). https://pith.science/paper/D7I7HD7X

@misc{pith2026241209619,
  author       = {Pith},
  title        = {Pith review of: SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D7I7HD7X}},
  note         = {Machine review of arXiv:2412.09619}
}
read the original abstract

Existing text-to-image (T2I) diffusion models face several limitations, including large model sizes, slow runtime, and low-quality generation on mobile devices. This paper aims to address all of these challenges by developing an extremely small and fast T2I model that generates high-resolution and high-quality images on mobile platforms. We propose several techniques to achieve this goal. First, we systematically examine the design choices of the network architecture to reduce model parameters and latency, while ensuring high-quality generation. Second, to further improve generation quality, we employ cross-architecture knowledge distillation from a much larger model, using a multi-level approach to guide the training of our model from scratch. Third, we enable a few-step generation by integrating adversarial guidance with knowledge distillation. For the first time, our model SnapGen, demonstrates the generation of 1024x1024 px images on a mobile device around 1.4 seconds. On ImageNet-1K, our model, with only 372M parameters, achieves an FID of 2.06 for 256x256 px generation. On T2I benchmarks (i.e., GenEval and DPG-Bench), our model with merely 379M parameters, surpasses large-scale models with billions of parameters at a significantly smaller size (e.g., 7x smaller than SDXL, 14x smaller than IF-XL).

Figures

Figures reproduced from arXiv: 2412.09619 by the authors.

Figure 1
Figure 1. Comparison of various text-to-image models in terms of model size, mobile device compatibility, and visual output quality. Our [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Efficient UNet. Starting from a thinner and shorter version of the UNet from SDXL (as in (a)), we explore a series of architectural changes, i.e., (b)–(f), to develop a smaller and faster model while retaining high-quality generation performance, as evaluated in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Comparisons of Performance and Efficiency for various Design Choices of Efficient UNet. The generation quality is evaluated using FID calculated on ImageNet-1K for 2562 px generation. The efficiency metrics include model parameters, latency, and FLOPs. FLOPs and latency (on iPhone 15 Pro) are measured with a 128 × 128 latent, equivalent to a 1024 × 1024 decoded image, for one forward pass. We show the architecture e… view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Comparisons of Decoder Architecture between (a) SDXL/SD3 decoder and (b) our tiny decoder [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 7
Figure 7. Figure 7: Overview of Adversarial Step Dis￾tillation. Output distillation and distribution￾matching distillation are performed. 3.3. Training Recipe and Multi-Level Distillation To improve the generation quality of our efficient diffusion model, we propose a series of training t…
Figure 8
Figure 8. Figure 8: Human Evaluation. We conduct a user study to compare images generated by our model against baselines on three attributes: aesthetic quality, text-image alignment, and realistic generations. Our model surpasses the quality of SDXL and SD3 models, while performing compet…
Figure 9
Figure 9. Figure 9: Performance comparison of few-step generation for our [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 2
Figure 2. Figure 2: Comparisons of Decoder Reconstruction between SD3 decoder and our tiny decoder. Zoom in for better viewing. C. Detailed Results on Benchmarks We present detailed results for GenEval and DPG-Bench in Tab. 1 and Tab. 2, respectively. On GenEval, our model demonstrates ex…
Figure 1
Figure 1. Figure 1: Demo on iPhone 16 Pro-Max. We report the forward time for each 4-step generation, excluding the model loading time. B. Reconstruction by VAE Decoders We compare reconstruction results between the SD3 VAE decoder (49.55M parameters) and our tiny decoder (1.38M parameter…
Figure 3
Figure 3. Figure 3: Few-step generation qualitative comparison at 10242 resolution. The prompts used in these examples are from PixArt [13]. 16 [PITH_FULL_IMAGE:figures/full_fig_p016_3.png]
Figure 4
Figure 4. Figure 4: Additional Qualitative Comparison. Our model demonstrates competitive visual quality and superior prompt-following ability. Input text prompts are shown above each image grid; all images are generated at 10242 resolution. Zoom in for details. 17 [PITH_FULL_IMAGE:figur…
Figure 5
Figure 5. Figure 5: Additional qualitative comparison at 10242 resolution. Zoom in for details. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]
Figure 6
Figure 6. Figure 6: Additional T2I example visualization at 10242 resolution of our model. Zoom in for details. 19 [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]
Figure 7
Figure 7. Figure 7: Additional T2I example visualization at 10242 resolution of our model. Zoom in for details. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SnapGen++: Unleashing Diffusion Transformers for Efficient High-Fidelity Image Generation on Edge Devices

    cs.CV 2026-01 conditional novelty 6.0 of 10

    A compact elastic diffusion transformer with adaptive sparse attention and knowledge-guided distribution-matching distillation achieves 4-step 1K image generation on a phone in roughly 1.8 seconds.

  2. TurboVSR: Fantastic Video Upscalers and Where to Find Them

    cs.CV 2025-06 conditional novelty 6.0 of 10

    TurboVSR uses a high-compression video autoencoder with factorized conditioning and non-uniform shortcut sampling to achieve near-state-of-the-art perceptual video super-resolution at roughly 100x lower compute cost.

  3. JuZhou 1.0 Technical Report: The First Edge-Native Text-to-Image Foundation Model Trained Entirely on China-Developed AI Accelerators

    cs.CV 2026-06 unverdicted novelty 5.0 of 10

    JuZhou 1.0 is a 0.387B-parameter T2I diffusion model with 4-step inference achieving 0.69 GenEval, trained on 9M Chinese pairs using Sugon K100 accelerators and deployable on Android/iOS devices.

  4. SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer

    cs.CV 2025-01 conditional novelty 5.0 of 10

    SANA-1.5 combines layer growth, depth pruning, and VLM-judged best-of-N sampling to push GenEval text-to-image alignment from 0.81 to 0.96.

Reference graph

Works this paper leans on

88 extracted references · 31 canonical work pages · cited by 4 Pith papers

  1. [1]

    Stable diffusion 1.5

    Stability AI. Stable diffusion 1.5. https://huggingface.co/stable-diffusion-v1-5/stable- diffusion-v1-5, 2022. 8, 14

    2022

  2. [2]

    Stable diffusion 2.1

    Stability AI. Stable diffusion 2.1. https://huggingface.co/stabilityai/stable-diffusion-2-1,

  3. [3]

    Stable diffusion 3.5

    Stability AI. Stable diffusion 3.5. https://github.com/Stability-AI/sd3.5, 2024. 2

    2024

  4. [4]

    Stable diffusion 3.5 large

    Stability AI. Stable diffusion 3.5 large. https://huggingface.co/stabilityai/stable-diffusion-3.5-large,

  5. [5]

    Stable diffusion 3.5 large turbo

    Stability AI. Stable diffusion 3.5 large turbo. https://huggingface.co/stabilityai/stable-diffusion-3.5- large-turbo, 2024. 2, 7

    2024

  6. [6]

    Kandinsky 3: Text-to-image synthe- sis for multifunctional generative framework

    Vladimir Arkhipkin, Viacheslav Vasilev, Andrei Filatov, Igor Pavlov, Julia Agafonova, Nikolai Gerasimenko, Anna Averchenkova, Evelina Mironova, Anton Bukashkin, Kon- stantin Kulikov, et al. Kandinsky 3: Text-to-image synthe- sis for multifunctional generative framework. arXiv preprint arXiv:2410.21061, 2024. 3

    arXiv

  7. [7]

    Kandinsky 2.1, 2023

    Shakhmatov Arseni, Razzhigaev Anton, Nikolich Aleksandr, Arkhipkin Vladimir, Pavlov Igor, Kuznetsov Andrey, and Denis Dimitrov. Kandinsky 2.1, 2023. 3

    2023

  8. [8]

    All are worth words: A vit backbone for diffusion models

    Fan Bao, Shen Nie, Kaiwen Xue, Yue Cao, Chongxuan Li, Hang Su, and Jun Zhu. All are worth words: A vit backbone for diffusion models. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition , pages 22669–22679, 2023. 5

    2023

Show all 88 references
  1. [9]

    Efficientvit: Lightweight multi-scale attention for high- resolution dense prediction

    Han Cai, Junyan Li, Muyan Hu, Chuang Gan, and Song Han. Efficientvit: Lightweight multi-scale attention for high- resolution dense prediction. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 17302– 17313, 2023. 3

    2023

  2. [10]

    LD-Pruner: Efficient Pruning of Latent Dif- fusion Models using Task-Agnostic Insights

    Thibault Castells, Hyoung-Kyu Song, Bo-Kyeong Kim, and Shinkook Choi. LD-Pruner: Efficient Pruning of Latent Dif- fusion Models using Task-Agnostic Insights. InProceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 821–830, 2024. 2

    2024

  3. [11]

    EdgeFusion: On-Device Text-to-Image Generation

    Thibault Castells, Hyoung-Kyu Song, Tairen Piao, Shinkook Choi, Bo-Kyeong Kim, Hanyoung Yim, Changgwun Lee, Jae Gon Kim, and Tae-Ho Kim. EdgeFusion: On-Device Text-to-Image Generation. arXiv preprint arXiv:2404.11925, 2024. 3

    2024 arXiv

  4. [12]

    Muse: Text-to-image generation via masked generative transform- ers

    Huiwen Chang, Han Zhang, Jarred Barber, AJ Maschinot, Jose Lezama, Lu Jiang, Ming-Hsuan Yang, Kevin Mur- phy, William T Freeman, Michael Rubinstein, et al. Muse: Text-to-image generation via masked generative transform- ers. arXiv preprint arXiv:2301.00704, 2023. 15

    2023 arXiv

  5. [13]

    Pixart- α: Fast training of diffusion transformer for photorealistic text-to-image synthesis

    Junsong Chen, Jincheng Yu, Chongjian Ge, Lewei Yao, Enze Xie, Yue Wu, Zhongdao Wang, James Kwok, Ping Luo, Huchuan Lu, et al. Pixart- α: Fast training of diffusion transformer for photorealistic text-to-image synthesis. arXiv preprint arXiv:2310.00426, 2023. 2, 3, 7, 8, 14, 16

    arXiv

  6. [14]

    Pixart- \sigma: Weak-to-strong training of diffusion transformer for 4k text-to-image generation

    Junsong Chen, Chongjian Ge, Enze Xie, Yue Wu, Lewei Yao, Xiaozhe Ren, Zhongdao Wang, Ping Luo, Huchuan Lu, and Zhenguo Li. Pixart- \sigma: Weak-to-strong training of diffusion transformer for 4k text-to-image generation. arXiv preprint arXiv:2403.04692, 2024. 2, 3, 8, 14

    2024 arXiv

  7. [15]

    Transformers are ssms: General- ized models and efficient algorithms through structured state space duality

    Tri Dao and Albert Gu. Transformers are ssms: General- ized models and efficient algorithms through structured state space duality. arXiv preprint arXiv:2405.21060, 2024. 3

    2024 arXiv

  8. [16]

    Deepfloyd

    DeepFloyd. Deepfloyd. https://github.com/deep-floyd/IF,

  9. [17]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009. 2, 3, 7

    2009

  10. [18]

    Adam: A method for stochastic opti- mization

    P Kingma Diederik. Adam: A method for stochastic opti- mization. (No Title), 2014. 7

    2014

  11. [19]

    Scaling recti- fied flow transformers for high-resolution image synthesis

    Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas M ¨uller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, et al. Scaling recti- fied flow transformers for high-resolution image synthesis. In Forty-first International Conference on Mach...

    2024

  12. [20]

    Institutionum calculi integralis

    Leonhard Euler. Institutionum calculi integralis. imp. Acad. imp. Sa‘ent., 1768. 6

  13. [21]

    Lumina-T2X: Transforming Text into Any Modality, Resolution, and Duration via Flow-based Large Diffusion Transformers

    Peng Gao, Le Zhuo, Ziyi Lin, Chris Liu, Junsong Chen, Ruoyi Du, Enze Xie, Xu Luo, Longtian Qiu, Yuhang Zhang, et al. Lumina-T2X: Transforming Text into Any Modality, Resolution, and Duration via Flow-based Large Diffusion Transformers. arXiv preprint arXiv:2405.05945, 2024. 3

    2024 arXiv

  14. [22]

    Geneval: An object-focused framework for evaluating text- to-image alignment

    Dhruba Ghosh, Hannaneh Hajishirzi, and Ludwig Schmidt. Geneval: An object-focused framework for evaluating text- to-image alignment. Advances in Neural Information Pro- cessing Systems, 36, 2024. 7, 15

    2024

  15. [23]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. Advances in neural information processing systems, 27, 2014. 3, 5

    2014

  16. [24]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 3

    2016

  17. [25]

    Query-key normalization for transformers

    Alex Henry, Prudhvi Raj Dachapally, Shubham Pawar, and Yuxuan Chen. Query-key normalization for transformers. arXiv preprint arXiv:2010.04245, 2020. 5

    2010 arXiv

  18. [26]

    Gans trained by a two time-scale update rule converge to a local nash equilib- rium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilib- rium. Advances in neural information processing systems , 30, 2017. 3

    2017

  19. [28]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598, 2022. 7

    2022 arXiv

  20. [29]

    Denoising dif- fusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising dif- fusion probabilistic models. Advances in neural information processing systems, 33:6840–6851, 2020. 14

    2020

  21. [30]

    Mobilenets: Efficient convolutional neu- ral networks for mobile vision applications

    Andrew G Howard. Mobilenets: Efficient convolutional neu- ral networks for mobile vision applications. arXiv preprint arXiv:1704.04861, 2017. 3

    2017 arXiv

  22. [31]

    Ella: Equip diffusion models with llm for enhanced semantic alignment

    Xiwei Hu, Rui Wang, Yixiao Fang, Bin Fu, Pei Cheng, and Gang Yu. Ella: Equip diffusion models with llm for enhanced semantic alignment. arXiv preprint arXiv:2403.05135, 2024. 7, 15

    2024 arXiv

  23. [32]

    Ascan: Asymmetric convolution-attention networks for efficient recognition and generation

    Anil Kag, Huseyin Coskun, Jierun Chen, Junli Cao, Willi Menapace, Aliaksandr Siarohin, Sergey Tulyakov, and Jian Ren. Ascan: Asymmetric convolution-attention networks for efficient recognition and generation. arXiv preprint arXiv:2411.04967, 2024. 3, 7, 14

    2024 arXiv

  24. [33]

    Elucidating the design space of diffusion-based generative models

    Tero Karras, Miika Aittala, Timo Aila, and Samuli Laine. Elucidating the design space of diffusion-based generative models. Advances in neural information processing systems, 35:26565–26577, 2022. 15

    2022

  25. [34]

    Transformers are rnns: Fast autoregressive transformers with linear attention

    Angelos Katharopoulos, Apoorv Vyas, Nikolaos Pappas, and Franc ¸ois Fleuret. Transformers are rnns: Fast autoregressive transformers with linear attention. In International confer- ence on machine learning, pages 5156–5165. PMLR, 2020. 3

    2020

  26. [35]

    Bk-sdm: Architecturally Compressed Sta- ble Diffusion for Efficient Text-to-Image Generation

    Bo-Kyeong Kim, Hyoung-Kyu Song, Thibault Castells, and Shinkook Choi. Bk-sdm: Architecturally Compressed Sta- ble Diffusion for Efficient Text-to-Image Generation. In Workshop on Efficient Systems for Foundation Models@ ICML2023, 2023. 2, 3, 6

    2023

  27. [36]

    Auto-encoding variational bayes

    Diederik P Kingma. Auto-encoding variational bayes. 2nd International Conference on Learning Representations ,

  28. [37]

    Applying guidance in a limited interval improves sample and distribution quality in diffusion models

    Tuomas Kynk ¨a¨anniemi, Miika Aittala, Tero Karras, Samuli Laine, Timo Aila, and Jaakko Lehtinen. Applying guidance in a limited interval improves sample and distribution quality in diffusion models. arXiv preprint arXiv:2404.07724, 2024. 5, 15

    2024 arXiv

  29. [38]

    mplug: Effective and efficient vision-language learning by cross-modal skip-connections

    Chenliang Li, Haiyang Xu, Junfeng Tian, Wei Wang, Ming Yan, Bin Bi, Jiabo Ye, He Chen, Guohai Xu, Zheng Cao, et al. mplug: Effective and efficient vision-language learning by cross-modal skip-connections. InProceedings of the 2022 Conference on Empirical Methods in Natural Lan...

    2022

  30. [39]

    Playground v1,

    Daiqing Li, Aleks Kamko, Ali Sabet, Ehsan Akhgari, Lin- miao Xu, and Suhail Doshi. Playground v1, . 3

  31. [40]

    Playground v2,

    Daiqing Li, Aleks Kamko, Ali Sabet, Ehsan Akhgari, Lin- miao Xu, and Suhail Doshi. Playground v2, . 8

  32. [41]

    Playground V2

    Daiqing Li, Aleks Kamko, Ehsan Akhgari, Ali Sabet, Lin- miao Xu, and Suhail Doshi. Playground V2. 5: Three Insights towards Enhancing Aesthetic Quality in Text-to- Image Generation. arXiv preprint arXiv:2402.17245, 2024. 3, 8, 14

    2024 arXiv

  33. [42]

    On the scalability of diffusion-based text-to-image generation

    Hao Li, Yang Zou, Ying Wang, Orchid Majumder, Yusheng Xie, R Manmatha, Ashwin Swaminathan, Zhuowen Tu, Ste- fano Ermon, and Stefano Soatto. On the scalability of diffusion-based text-to-image generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern ...

    2024

  34. [43]

    Snap- fusion: Text-to-Image Diffusion Model on Mobile Devices within Two Seconds

    Yanyu Li, Huan Wang, Qing Jin, Ju Hu, Pavlo Chemerys, Yun Fu, Yanzhi Wang, Sergey Tulyakov, and Jian Ren. Snap- fusion: Text-to-Image Diffusion Model on Mobile Devices within Two Seconds. Advances in Neural Information Pro- cessing Systems, 36, 2024. 2, 3, 5

    2024

  35. [44]

    Hunyuan- DiT: A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding

    Zhimin Li, Jianwei Zhang, Qin Lin, Jiangfeng Xiong, Yanxin Long, Xinchi Deng, Yingfang Zhang, Xingchao Liu, Minbin Huang, Zedong Xiao, et al. Hunyuan- DiT: A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding. arXiv preprint arXiv:2405.0874...

    2024 arXiv

  36. [45]

    SDXL- Lightning: Progressive Adversarial Diffusion Distillation,

    Shanchuan Lin, Anran Wang, and Xiao Yang. SDXL- Lightning: Progressive Adversarial Diffusion Distillation,

  37. [46]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll´ar, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceeding...

    2014

  38. [47]

    Flow matching for generative mod- eling

    Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maximil- ian Nickel, and Matt Le. Flow matching for generative mod- eling. arXiv preprint arXiv:2210.02747, 2022. 2, 6

    2022 arXiv

  39. [48]

    Playground v3: Improving text-to- image alignment with deep-fusion large language models

    Bingchen Liu, Ehsan Akhgari, Alexander Visheratin, Aleks Kamko, Linmiao Xu, Shivam Shrirao, Joao Souza, Suhail Doshi, and Daiqing Li. Playground v3: Improving text-to- image alignment with deep-fusion large language models. arXiv preprint arXiv:2409.10695, 2024. 3

    2024 arXiv

  40. [49]

    Linfusion: 1 gpu, 1 minute, 16k image.arXiv preprint arXiv:2409.02097, 2024

    Songhua Liu, Weihao Yu, Zhenxiong Tan, and Xinchao Wang. Linfusion: 1 gpu, 1 minute, 16k image.arXiv preprint arXiv:2409.02097, 2024. 2, 3, 6

    2024 arXiv

  41. [50]

    Flow straight and fast: Learning to generate and transfer data with rectified flow

    Xingchao Liu, Chengyue Gong, and Qiang Liu. Flow straight and fast: Learning to generate and transfer data with rectified flow. arXiv preprint arXiv:2209.03003, 2022. 2, 6

    2022 arXiv

  42. [51]

    Hyperhuman: Hyper-realistic human gener- ation with latent structural diffusion

    Xian Liu, Jian Ren, Aliaksandr Siarohin, Ivan Skorokhodov, Yanyu Li, Dahua Lin, Xihui Liu, Ziwei Liu, and Sergey Tulyakov. Hyperhuman: Hyper-realistic human gener- ation with latent structural diffusion. arXiv preprint arXiv:2310.08579, 2023. 2, 3

    arXiv

  43. [52]

    Sit: Explor- ing flow and diffusion-based generative models with scalable interpolant transformers

    Nanye Ma, Mark Goldstein, Michael S Albergo, Nicholas M Boffi, Eric Vanden-Eijnden, and Saining Xie. Sit: Explor- ing flow and diffusion-based generative models with scalable interpolant transformers. arXiv preprint arXiv:2401.08740,

    arXiv

  44. [53]

    Snap video: Scaled spatiotemporal transformers for text-to-video synthesis

    Willi Menapace, Aliaksandr Siarohin, Ivan Skorokhodov, Ekaterina Deyneka, Tsai-Shien Chen, Anil Kag, Yuwei Fang, Aleksei Stoliar, Elisa Ricci, Jian Ren, et al. Snap video: Scaled spatiotemporal transformers for text-to-video synthesis. In Proceedings of the IEEE/CVF Conference...

    2024

  45. [55]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision, pages 4195–4205,

  46. [56]

    Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas M ¨uller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis. arXiv preprint arXiv:2307.01952, 2023. 2, 3, 5, 8, 14

    2023 arXiv

  47. [57]

    Movie gen: A cast of media foundation models

    Adam Polyak, Amit Zohar, Andrew Brown, Andros Tjandra, Animesh Sinha, Ann Lee, Apoorv Vyas, Bowen Shi, Chih- Yao Ma, Ching-Yao Chuang, et al. Movie gen: A cast of media foundation models. arXiv preprint arXiv:2410.13720,

    arXiv

  48. [58]

    Mobilenetv4- universal models for the mobile ecosystem

    Danfeng Qin, Chas Leichner, Manolis Delakis, Marco Fornoni, Shixin Luo, Fan Yang, Weijun Wang, Colby Ban- bury, Chengxi Ye, Berkin Akin, et al. Mobilenetv4- universal models for the mobile ecosystem. arXiv preprint arXiv:2404.10518, 2024. 3

    2024 arXiv

  49. [59]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

    2021

  50. [60]

    Zero-shot text-to-image generation

    Aditya Ramesh, Mikhail Pavlov, Gabriel Goh, Scott Gray, Chelsea V oss, Alec Radford, Mark Chen, and Ilya Sutskever. Zero-shot text-to-image generation. In Proceedings of the 38th International Conference on Machine Learning , pages 8821–8831. PMLR, 2021. 2

    2021

  51. [61]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022. 3, 5

    2022

  52. [62]

    Sara Mahdavi, Rapha Gontijo Lopes, Tim Salimans, Jonathan Ho, David J Fleet, and Mohammad Norouzi

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily Denton, Seyed Kamyar Seyed Ghasemipour, Burcu Karagol Ayan, S. Sara Mahdavi, Rapha Gontijo Lopes, Tim Salimans, Jonathan Ho, David J Fleet, and Mohammad Norouzi. Photorealistic text-to-image diffusion mod...

    2022

  53. [63]

    Fast high- resolution image synthesis with latent adversarial diffusion distillation

    Axel Sauer, Frederic Boesel, Tim Dockhorn, Andreas Blattmann, Patrick Esser, and Robin Rombach. Fast high- resolution image synthesis with latent adversarial diffusion distillation. arXiv preprint arXiv:2403.12015, 2024. 3, 7

    2024 arXiv

  54. [64]

    Adversarial diffusion distillation

    Axel Sauer, Dominik Lorenz, Andreas Blattmann, and Robin Rombach. Adversarial diffusion distillation. In European Conference on Computer Vision , pages 87–103. Springer,

  55. [65]

    Fast transformer decoding: One write-head is all you need

    Noam Shazeer. Fast transformer decoding: One write-head is all you need. arXiv preprint arXiv:1911.02150, 2019. 4

    1911 arXiv

  56. [66]

    Emu edit: Precise image editing via recognition and gen- eration tasks

    Shelly Sheynin, Adam Polyak, Uriel Singer, Yuval Kirstain, Amit Zohar, Oron Ashual, Devi Parikh, and Yaniv Taigman. Emu edit: Precise image editing via recognition and gen- eration tasks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pa...

    2024

  57. [67]

    Sdxs: Real- time one-step latent diffusion models with image conditions

    Yuda Song, Zehao Sun, and Xuanwu Yin. Sdxs: Real- time one-step latent diffusion models with image conditions. arXiv preprint arXiv:2403.16627, 2024. 2

    2024 arXiv

  58. [68]

    Roformer: Enhanced transformer with rotary position embedding

    Jianlin Su, Murtadha Ahmed, Yu Lu, Shengfeng Pan, Wen Bo, and Yunfeng Liu. Roformer: Enhanced transformer with rotary position embedding. Neurocomputing, 568:127063,

  59. [69]

    Bitsfusion: 1.99 bits weight quantization of diffusion model

    Yang Sui, Yanyu Li, Anil Kag, Yerlan Idelbayev, Junli Cao, Ju Hu, Dhritiman Sagar, Bo Yuan, Sergey Tulyakov, and Jian Ren. Bitsfusion: 1.99 bits weight quantization of diffusion model. arXiv preprint arXiv:2406.04333, 2024. 2, 3, 6

    2024 arXiv

  60. [70]

    Gemma 2: Improving open language models at a practical size

    Gemma Team, Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupati- raju, L´eonard Hussenot, Thomas Mesnard, Bobak Shahriari, Alexandre Ram´e, et al. Gemma 2: Improving open language models at a practical size. arXiv preprint arXiv:2408.00118,

    arXiv

  61. [71]

    Kolors: Effective Training of Diffusion Model for Photorealistic Text-to-Image Synthesis

    Kolors Team. Kolors: Effective Training of Diffusion Model for Photorealistic Text-to-Image Synthesis. arXiv preprint,

  62. [72]

    Attention is all you need

    A Vaswani. Attention is all you need. Advances in Neural Information Processing Systems, 2017. 3

    2017

  63. [73]

    Diffusion-gan: Training gans with diffusion

    Zhendong Wang, Huangjie Zheng, Pengcheng He, Weizhu Chen, and Mingyuan Zhou. Diffusion-gan: Training gans with diffusion. arXiv preprint arXiv:2206.02262, 2022. 3

    2022 arXiv

  64. [74]

    Omnigen: Unified image generation

    Shitao Xiao, Yueze Wang, Junjie Zhou, Huaying Yuan, Xin- grun Xing, Ruiran Yan, Shuting Wang, Tiejun Huang, and Zheng Liu. Omnigen: Unified image generation. arXiv preprint arXiv:2409.11340, 2024. 2

    2024 arXiv

  65. [75]

    Tack- ling the generative learning trilemma with denoising diffu- sion gans

    Zhisheng Xiao, Karsten Kreis, and Arash Vahdat. Tack- ling the generative learning trilemma with denoising diffu- sion gans. arXiv preprint arXiv:2112.07804, 2021. 3

    2021 arXiv

  66. [76]

    Sana: Ef- ficient high-resolution image synthesis with linear diffusion transformers

    Enze Xie, Junsong Chen, Junyu Chen, Han Cai, Yujun Lin, Zhekai Zhang, Muyang Li, Yao Lu, and Song Han. Sana: Ef- ficient high-resolution image synthesis with linear diffusion transformers. arXiv preprint arXiv:2410.10629, 2024. 2, 3, 8, 14

    arXiv

  67. [77]

    Imagere- ward: Learning and evaluating human preferences for text- to-image generation

    Jiazheng Xu, Xiao Liu, Yuchen Wu, Yuxuan Tong, Qinkai Li, Ming Ding, Jie Tang, and Yuxiao Dong. Imagere- ward: Learning and evaluating human preferences for text- to-image generation. Advances in Neural Information Pro- cessing Systems, 36, 2024. 7, 15

    2024

  68. [78]

    Semi-implicit denoising diffusion models (siddms)

    Yanwu Xu, Mingming Gong, Shaoan Xie, Wei Wei, Matthias Grundmann, Kayhan Batmanghelich, and Tingbo Hou. Semi-implicit denoising diffusion models (siddms). Ad- vances in neural information processing systems, 36:17383,

  69. [79]

    Ufogen: You forward once large scale text-to-image gener- ation via diffusion gans

    Yanwu Xu, Yang Zhao, Zhisheng Xiao, and Tingbo Hou. Ufogen: You forward once large scale text-to-image gener- ation via diffusion gans. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 8196–8206, 2024. 2, 3

    2024

  70. [80]

    Cogvideox: Text-to-video diffusion models with an expert transformer

    Zhuoyi Yang, Jiayan Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiao- han Zhang, Guanyu Feng, et al. Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint arXiv:2408.06072, 2024. 2

    2024 arXiv

  71. [81]

    Im- proved distribution matching distillation for fast image syn- thesis

    Tianwei Yin, Micha ¨el Gharbi, Taesung Park, Richard Zhang, Eli Shechtman, Fredo Durand, and William T Freeman. Im- proved distribution matching distillation for fast image syn- thesis. arXiv preprint arXiv:2405.14867, 2024. 3

    2024 arXiv

  72. [82]

    One-step diffusion with distribution matching distillation

    Tianwei Yin, Micha ¨el Gharbi, Richard Zhang, Eli Shecht- man, Fredo Durand, William T Freeman, and Taesung Park. One-step diffusion with distribution matching distillation. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 6613–6623...

    2024

  73. [83]

    Scaling autoregressive models for content-rich text-to-image generation

    Jiahui Yu, Yuanzhong Xu, Jing Yu Koh, Thang Luong, Gun- jan Baid, Zirui Wang, Vijay Vasudevan, Alexander Ku, Yin- fei Yang, Burcu Karagol Ayan, et al. Scaling autoregressive models for content-rich text-to-image generation. Transac- tions on Machine Learning Research. 8

  74. [84]

    Root mean square layer nor- malization

    Biao Zhang and Rico Sennrich. Root mean square layer nor- malization. Advances in Neural Information Processing Sys- tems, 32, 2019. 5

    2019

  75. [85]

    Laptop-Diff: Layer Pruning and Normal- ized Distillation for Compressing Diffusion Models

    Dingkun Zhang, Sijia Li, Chen Chen, Qingsong Xie, and Haonan Lu. Laptop-Diff: Layer Pruning and Normal- ized Distillation for Compressing Diffusion Models. arXiv preprint arXiv:2404.11098, 2024. 2

    2024 arXiv

  76. [86]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In CVPR, 2018. 5

    2018

  77. [87]

    Sine: Single image editing with text- to-image diffusion models

    Zhixing Zhang, Ligong Han, Arnab Ghosh, Dimitris N Metaxas, and Jian Ren. Sine: Single image editing with text- to-image diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 6027–6037, 2023. 2

    2023

  78. [88]

    Mobilediffusion: Subsecond text-to-image generation on mobile devices

    Yang Zhao, Yanwu Xu, Zhisheng Xiao, and Tingbo Hou. Mobilediffusion: Subsecond text-to-image generation on mobile devices. arXiv preprint arXiv:2311.16567, 2023. 2, 3, 4, 5

    2023 arXiv

  79. [89]

    Lumina-next: Making lumina-t2x stronger and faster with next-dit

    Le Zhuo, Ruoyi Du, Han Xiao, Yangguang Li, Dongyang Liu, Rongjie Huang, Wenze Liu, Lirui Zhao, Fu-Yun Wang, Zhanyu Ma, et al. Lumina-next: Making lumina-t2x stronger and faster with next-dit. arXiv preprint arXiv:2406.18583 ,

    arXiv

  80. [2024]

    Generate

    2, 8, 14 12 SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Training Supplementary Material A. Demo on Mobile Devices We present an on-device demo showcasing the capabili- ties of our efficient text-to-image model in gen...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.