Pith. sign in

REVIEW 3 major objections 4 minor 54 references

PassionSR: Post-Training Quantization with Adaptive Scale in One-Step Diffusion based Image Super-Resolution

T0 review · 3 major / 4 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 post-training quantization pipeline quantizes one-step diffusion super-resolution models to 8-bit and 6-bit with visual quality comparable to full precision, while cutting parameters by up to 86% and operations by up to 82%.

desk verdict First PTQ pipeline for one-step diffusion SR with VAE quantization, but the 6-bit 'comparable to full precision' claim is contradicted by the paper's own Table 2. read the letter →

arxiv 2411.17106 v3 pith:4LZTW7IZ submitted 2024-11-26 cs.CV

Libo Zhu , Jianze Li , Haotong Qin , Wenbo Li , Yulun Zhang , Yong Guo , Xiaokang Yang This is my paper · ORCID
classification cs.CV
keywords imagesuper-resolutionone-stepdiffusionmodelspost-trainingquantizationlow-bitlearnableboundaryquantizerequivalenttransformationcalibrationefficientinference
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

One-step diffusion super-resolution models restore images in a single denoising step, but their storage and compute costs are still too high for phones and embedded devices. This paper tries to establish that such models can be post-training quantized to 8-bit or 6-bit weights and activations without a visible quality loss, by first simplifying the architecture to its UNet and VAE cores and then calibrating trainable quantizer parameters. The paper's central claim is that its three components—learnable clipping boundaries, per-channel activation rescaling, and a two-stage calibration schedule—recover most of the full-precision behavior while cutting parameters by up to 86% and operations by up to 82%. If true, single-step diffusion super-resolution becomes deployable on practical hardware instead of requiring large GPU clusters.

What carries the argument

The design is carried by a fake-quantization operator whose bounds are trainable: LBQ clips activations and weights to $[B_l, B_u]$, computes $\alpha=(B_u-B_l)/(2^N-1)$ and $\beta=B_l$, and produces $X_q=\alpha\lfloor (X_c-\beta)/\alpha\rceil+\beta$. LET then rescales each channel of the input as $\tilde{X}=(X-\delta)\oslash s$ while transforming weights and bias to $\tilde{W}=s\odot W$ and $\tilde{B}=B+\delta W$, leaving the full-precision output mathematically unchanged but making activations smoother; these factors merge into neighboring layers after calibration, so the quantized runtime has no extra cost. DQC separates calibration into two stages so that LET's scale factors and LBQ's boundaries are not optimized simultaneously, which stabilizes training and lowers GPU memory. The UNet loss is computed in latent space through the one-step transformation $I(Z_l,\varepsilon)=\sqrt{1/\hat{\alpha}}Z_l-\sqrt{(1-\hat{\alpha})/\hat{\alpha}}\,\varepsilon(Z_l)$, and the VAE encoder and decoder are each anchored to their full-precision outputs by MSE losses.

What would settle it

Apply the same LBQ, LET, and DQC pipeline directly to the original OSEDiff architecture without removing DAPE and the CLIPEncoder, using the same 500-pair calibration set; if the 8-bit model's quality no longer tracks the full-precision model, then the reported success comes from the architectural simplification rather than from the quantization components.

Watch

Extended reading notes

Core claim

Working from OSEDiff, the paper replaces the DAPE and CLIPEncoder conditioning branch with a constant empty-prompt embedding to obtain a simplified full-precision model, PassionSR-FP. The central claim is that this simplified model, when quantized with the proposed pipeline, delivers W8A8 and W6A6 outputs that are visually comparable to the 32-bit models while reducing parameters by 81.77% (8-bit) and 86.32% (6-bit) and operations by 76.56% and 82.42% in the UNet–VAE configuration. Quantizing the VAE is presented as essential because the VAE accounts for over 80% of the computational load once denoising is reduced to one step. The paper reports that prior diffusion-quantization baselines, which were designed for multi-step denoising, degrade sharply on one-step models, whereas PassionSR's learned quantizers adapt the distribution to the available bit widths.

Load-bearing premise

The load-bearing premise is that the simplified PassionSR-FP, with the DAPE and CLIPEncoder branches removed and replaced by a constant empty-prompt embedding, behaves closely enough like the original OSEDiff for quantization results to transfer; the paper's Table 2 itself shows a 25.57 to 26.70 PSNR shift on RealSR, so this premise is doing real work.

Editorial extensions

If this is right

  • At 8-bit precision, the quantized model is claimed to match or beat the full-precision reference on several metrics while using 81.77% fewer parameters and 76.56% fewer operations, so mobile inference becomes feasible.
  • At 6-bit precision, the pipeline remains usable with up to 86.32% parameter compression and 82.42% operation compression, extending low-bit post-training quantization to one-step diffusion models.
  • Quantizing the VAE rather than only the UNet is required for one-step models because the VAE dominates compute; including it raises the compression ratio from about 81% to 86% at 6-bit.
  • Existing multi-step diffusion quantization methods do not transfer directly to one-step models; the paper's comparisons show large drops for those baselines, supporting the need for one-step-specific calibration.

Reading between the lines

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

  • Beyond the paper, the same two-stage calibration idea could transfer to other single-step generative models, such as one-step text-to-image or video diffusion, wherever quantizer parameters and activation scales are trained jointly.
  • Because the conditioning branch is replaced by a constant embedding, the quantized model's behavior for prompt-dependent inputs remains untested; a direct extension would quantize the DAPE and CLIPEncoder branches instead of dropping them, at the cost of some compression.
  • A testable extension would be to vary the calibration dataset's content (e.g., faces, text, low-light scenes) and measure whether the learned boundaries and scales shift, indicating that task-specific calibration matters.
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 / 4 minor

Summary. The paper proposes PassionSR, a post-training quantization (PTQ) pipeline for one-step diffusion image super-resolution models. The method first simplifies OSEDiff into a UNet+VAE architecture (PassionSR-FP), then introduces a Learnable Boundary Quantizer (LBQ), a Learnable Equivalent Transformation (LET), and a Distributed Quantization Calibration (DQC) strategy. The authors evaluate on RealSR, DRealSR, and DIV2K validation at W8A8 and W6A6 settings, comparing against MaxMin, LSQ, Q-Diffusion, and EfficientDM. The central claims are that PassionSR achieves visual results comparable to full precision at both 8-bit and 6-bit, and that it significantly outperforms prior low-bit quantization methods for image super-resolution.

Significance. If the claims held, this would be a useful step toward deploying one-step diffusion super-resolution on resource-constrained hardware. The evaluation is broad in coverage: three datasets, eight metrics, and multiple bit widths, with all baselines re-run on the same backbone. The proposed components are clearly motivated, and the ablations show that LBQ and LET bring large gains over simple min-max quantization. However, the headline claim of FP-comparable visual quality at 6-bit is not supported by the paper's own main table, and the use of a simplified model as the full-precision reference introduces a transferability caveat. The method still appears competitive relative to the baselines, so the issues are fixable by reframing and additional evidence rather than by discarding the approach.

major comments (3)
  1. [§4.2, Table 2] The claim in the Abstract and Section 5 that '8-bit and 6-bit obtains comparable visual results with full-precision model' is not supported by Table 2. On RealSR at W6A6, compared with PassionSR-FP, LPIPS rises from 0.3339 to 0.4199 (+25.8%), DISTS from 0.1765 to 0.2592 (+46.9%), NIQE from 4.336 to 8.618 (+98.7%), MANIQA drops from 0.4686 to 0.2131 (-54.5%), and CLIP-IQA drops from 0.7520 to 0.4612 (-38.7%). Similar large degradations appear on DIV2K val at W6A6. Even at W8A8, NIQE worsens from 4.336 to 6.070 on RealSR and from 3.573 to 4.424 on DIV2K val. Since no threshold or user study is provided, the 'comparable visual results' claim is unsupported by the main experimental table. The method may still be useful as a strong PTQ baseline, but the headline claim must be substantially qualified or replaced with a measured perceptual-comparison study.
  2. [§3.2, Table 2] The simplification from OSEDiff to PassionSR-FP is not neutral on all datasets, and this affects the scope of the quantization claims. Table 2 shows that on RealSR, PassionSR-FP has PSNR 26.70 versus OSEDiff's 25.57, a gain of more than 1 dB, and also differs on LPIPS (0.3339 vs 0.3447). Since the full-precision reference for all quantization comparisons is PassionSR-FP, the reported 'comparable to full precision' results are only established for the simplified architecture, not for OSEDiff itself. Replacing the DAPE and CLIPEncoder branches with a constant empty-prompt embedding may change behavior on inputs where text conditioning matters; the paper should either evaluate quantization on the original OSEDiff or explicitly limit the claims to the simplified model.
  3. [§4.3, Table 4] The ablation text states that DQC produces 'slight performance enhancement,' but Table 4 shows the opposite on the main distortion metrics: adding DQC to LBQ+LET reduces PSNR from 25.40 to 24.41 and SSIM from 0.7529 to 0.7374 on RealSR at W6A6. Some perceptual and no-reference metrics improve (LPIPS, DISTS, NIQE, MUSIQ), so the effect is mixed, not a clear enhancement. The authors should either describe DQC as improving convergence and memory cost while having mixed metric effects, or provide a principled explanation of why the selected metrics favor the DQC configuration.
minor comments (4)
  1. [Table 3] The table has formatting errors: 'PassionSR-FTW32A32' should likely be 'PassionSR-FP W32A32', and the entry '3,732 ↓17.50%)' is missing an opening parenthesis.
  2. [Figure 6] The caption cites 'MaxMin [16]' but the method is otherwise referenced as [12]; please make the citation numbers consistent.
  3. [§4.4] The sentence 'LET playes an important role' contains a typo ('playes' should be 'plays').
  4. [§3.3.2] Equation (7) writes the equivalent transformation as 's ⊙ K = K̃' but does not define the corresponding transformation for the query matrix Q; the notation in the displayed equation should be made symmetric and self-contained.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: quantizer parameters are calibration-fitted on a small set and evaluated on held-out datasets; the 'comparable to full precision' claim is an empirical assessment, not a derivation from its own inputs.

full rationale

PassionSR is an empirical post-training quantization pipeline: LBQ boundaries and LET scale/offset parameters are optimized with a calibration loss over 500 DIV2K train crops, and the reported PSNR/SSIM/LPIPS/DISTS/NIQE/MUSIQ/MANIQA/CLIP-IQA numbers come from held-out RealSR, DRealSR, and DIV2K val. There is no derivational step whose conclusion is assumed as an input. The simplification of OSEDiff into PassionSR-FP is an architectural change evaluated directly against OSEDiff in Table 2, and the headline 'comparable to full precision' compares quantized PassionSR to its own full-precision backbone PassionSR-FP; that is a same-architecture baseline choice, not a definitional equivalence. LET is explicitly built on SmoothQuant/OmniQuant-style equivalent transformations and LBQ is a learnable-boundary quantizer; these are standard constructions adapted to one-step diffusion SR, not renamed known results. The self-references (DFOSD [17], QuantSR [26]) appear in related work and for compression-ratio calculation, and neither is load-bearing for the central quantization result. The strong 6-bit degradation visible in Table 2 (for example LPIPS rises from 0.3339 to 0.4199 and MANIQA falls from 0.4686 to 0.2131 on RealSR) is a correctness/support problem with the paper's headline claim, not circularity; the claim may be overstated, but it is not circular.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central claim rests on two sets of learned scalar parameters (LBQ boundaries and LET scales/offsets) fitted to a single 500-image calibration set, plus several empirical assumptions about model simplification, STE training, and dataset transfer. No new physical entities or formal axioms are introduced.

free parameters (2)
  • Learnable quantizer boundaries B_l, B_u = Not reported; learned per layer/tensor
    Trained via gradient descent on the DIV2K calibration set (Eq. 3); central to LBQ behavior.
  • LET scale factor s and offset delta = Not reported; learned per channel
    Learned to rescale activations before quantization (Eq. 4, 7); central to LET behavior.
assumptions (5)
  • domain assumption Removing DAPE and CLIPEncoder with a constant empty-prompt embedding preserves OSEDiff behavior closely enough for quantization transfer.
    Validated in Table 2, but the >1 dB PSNR change on RealSR (25.57 to 26.70) suggests the simplification is not neutral.
  • domain assumption The equivalent transformation in Eq. 4 and Eq. 7 keeps the FP output unchanged while making activations and weights easier to quantize.
    The identity is exact before quantization; the benefit after quantization is assumed and only empirically demonstrated (Fig. 7).
  • domain assumption Straight-through estimator (STE) gradients in Eq. 2 are sufficiently accurate to train LBQ and LET parameters.
    Standard in quantization literature, but no convergence guarantee; the paper relies on it for all learnable parameters.
  • domain assumption A calibration set of 500 LR-HR crops from DIV2K train generalizes to RealSR, DRealSR, and DIV2K val.
    PTQ standard practice; no sensitivity analysis is provided.
  • domain assumption The OSEDiff reconstruction operator I(Z_l, epsilon) in Eq. 9 is a valid target for the UNet calibration loss.
    Taken from OSEDiff as the one-step denoising transform; if this operator is not the correct one, the loss in Eq. 10 is mismatched.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PassionSR: Post-Training Quantization with Adaptive Scale in One-Step Diffusion based Image Super-Resolution." pith.science (2026). https://pith.science/paper/4LZTW7IZ

@misc{pith2026241117106,
  author       = {Pith},
  title        = {Pith review of: PassionSR: Post-Training Quantization with Adaptive Scale in One-Step Diffusion based Image Super-Resolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4LZTW7IZ}},
  note         = {Machine review of arXiv:2411.17106}
}
read the original abstract

Diffusion-based image super-resolution (SR) models have shown superior performance at the cost of multiple denoising steps. However, even though the denoising step has been reduced to one, they require high computational costs and storage requirements, making it difficult for deployment on hardware devices. To address these issues, we propose a novel post-training quantization approach with adaptive scale in one-step diffusion (OSD) image SR, PassionSR. First, we simplify OSD model to two core components, UNet and Variational Autoencoder (VAE) by removing the CLIPEncoder. Secondly, we propose Learnable Boundary Quantizer (LBQ) and Learnable Equivalent Transformation (LET) to optimize the quantization process and manipulate activation distributions for better quantization. Finally, we design a Distributed Quantization Calibration (DQC) strategy that stabilizes the training of quantized parameters for rapid convergence. Comprehensive experiments demonstrate that PassionSR with 8-bit and 6-bit obtains comparable visual results with full-precision model. Moreover, our PassionSR achieves significant advantages over recent leading low-bit quantization methods for image SR. Our code will be at https://github.com/libozhu03/PassionSR.

Figures

Figures reproduced from arXiv: 2411.17106 by the authors.

Figure 1
Figure 1. Visual comparison (×4) between full-precision (FP) multi-step and one-step diffusion SR models and our 8-bit quan￾tized PassionSR. Compared to FP models, PassionSR achieves about 81.77% params reduction and 4× speedup. Recently, diffusion-based image SR models have been attracting researchers’ attention. Diffusion models exhibit strong performance across various tasks, including image SR (see [PITH_FULL_IMAGE:figur… view at source ↗
Figure 2
Figure 2. Visual comparison (×4) of one-step diffusion SR mod￾els. We use OSEDiff as a 32-bit full-precision (FP) reference and provide 6-bit quantized version with different methods. inference speed. Firstly, we perform a pruning operation to simplify the model to its two core components, the UNet and VAE, with minimal or even no performance drop. We name the FP model structure after pruning as PassionSR-FP. It is easier for… view at source ↗
Figure 3
Figure 3. Diffusion-based image SR acceleration. bit quantization for multi-step diffusion models. How￾ever, there are few works specifically addressing the low￾bit quantization of one-step diffusion (OSD) models, which are significantly different from multi-step diffusion models. We follow the strategy in [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Overview of our PassionSR. Step 1: we simplify OSEDiff [ [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Loss comparison between w/ and w/o DQC 3.4.1 Distributed Quantization Calibration (DQC) Given the properties of the rounding function, the training process in model quantization tends to be unstable. This in￾stability will get worse when calibrating the boundaries in L…
Figure 6
Figure 6. Figure 6: Visual comparison (×4) with high-resolution image, full-precision model’s output and different quantization methods in some challenging cases at W8A8 and W6A6 UNet-VAE quantization. PassionSR gains significant visual advantages over other methods. Efficiency RealSR Met…
Figure 7
Figure 7. Figure 7: Distribution of scale factor and activation before and after smooth in the whole model. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 47 canonical work pages

  1. [1]

    Ntire 2017 chal- lenge on single image super-resolution: Dataset and study

    Eirikur Agustsson and Radu Timofte. Ntire 2017 chal- lenge on single image super-resolution: Dataset and study. In CVPRW, 2017. 6

    work page 2017

  2. [2]

    Cross aggregation transformer for image restoration

    Zheng Chen, Yulun Zhang, Jinjin Gu, Linghe Kong, Xin Yuan, et al. Cross aggregation transformer for image restoration. NeurIPS, 2022. 1

    work page 2022

  3. [3]

    Cross aggregation transformer for image restoration

    Zheng Chen, Yulun Zhang, Jinjin Gu, Linghe Kong, Xin Yuan, et al. Cross aggregation transformer for image restoration. NeurIPS, 2022. 3

    work page 2022

  4. [4]

    Dual aggregation transformer for image super-resolution

    Zheng Chen, Yulun Zhang, Jinjin Gu, Linghe Kong, Xiaokang Yang, and Fisher Yu. Dual aggregation transformer for image super-resolution. ICCV, 2023. 1

    work page 2023

  5. [5]

    Image quality assessment: Unifying struc- ture and texture similarity

    Keyan Ding, Kede Ma, Shiqi Wang, and Eero P Si- moncelli. Image quality assessment: Unifying struc- ture and texture similarity. TPAMI, 2020. 6

    work page 2020

  6. [6]

    Towards accurate post-training quantization for vision trans- former

    Yifu Ding, Haotong Qin, Qinghua Yan, Zhenhua Chai, Junjie Liu, Xiaolin Wei, and Xianglong Liu. Towards accurate post-training quantization for vision trans- former. ACM MM, 2022. 2

    work page 2022

  7. [7]

    Image super-resolution using deep convo- lutional networks

    Chao Dong, Chen Change Loy, Kaiming He, and Xi- aoou Tang. Image super-resolution using deep convo- lutional networks. TPAMI, 2015. 1

    work page 2015

  8. [8]

    Learned step size quantization

    Steven K Esser, Jeffrey L McKinstry, Deepika Bablani, Rathinakumar Appuswamy, and Dharmen- dra S Modha. Learned step size quantization. InICLR,

    work page

Show all 54 references
  1. [9]

    Efficientdm: Efficient quantization-aware fine-tuning of low-bit diffusion models.arXiv preprint arXiv:2310.03270, 2023

    Yefei He, Jing Liu, Weijia Wu, Hong Zhou, and Bohan Zhuang. Efficientdm: Efficient quantization-aware fine-tuning of low-bit diffusion models.arXiv preprint arXiv:2310.03270, 2023. 2, 3, 6, 7

    2023 arXiv

  2. [10]

    Ptqd: Accurate post- training quantization for diffusion models

    Yefei He, Luping Liu, Jing Liu, Weijia Wu, Hong Zhou, and Bohan Zhuang. Ptqd: Accurate post- training quantization for diffusion models. NeurIPS,

  3. [11]

    Accurate post training quantiza- tion with small calibration sets

    Itay Hubara, Yury Nahshan, Yair Hanani, Ron Banner, and Daniel Soudry. Accurate post training quantiza- tion with small calibration sets. ICML, 2021. 2

    2021

  4. [12]

    Quantization and train- ing of neural networks for efficient integer-arithmetic- only inference

    Benoit Jacob, Skirmantas Kligys, Bo Chen, Menglong Zhu, Matthew Tang, Andrew Howard, Hartwig Adam, and Dmitry Kalenichenko. Quantization and train- ing of neural networks for efficient integer-arithmetic- only inference. CVPR, 2018. 2, 4, 6

    2018

  5. [13]

    Real-world super- resolution via kernel estimation and noise injection

    Xiaozhong Ji, Yun Cao, Ying Tai, Chengjie Wang, Jilin Li, and Feiyue Huang. Real-world super- resolution via kernel estimation and noise injection. In CVPRW, 2020. 6

    2020

  6. [14]

    Musiq: Multi-scale image quality transformer

    Junjie Ke, Qifei Wang, Yilin Wang, Peyman Milan- far, and Feng Yang. Musiq: Multi-scale image quality transformer. In ICCV, 2021. 6

    2021

  7. [15]

    Ac- curate image super-resolution using very deep convo- lutional networks

    Jiwon Kim, Jung Kwon Lee, and Kyoung Mu Lee. Ac- curate image super-resolution using very deep convo- lutional networks. CVPR, 2016. 3

    2016

  8. [16]

    Low-bit quantization of neural networks for efficient inference

    Eli Kravchik, Fan Yang, Pavel Kisilev, and Yoni Choukroun. Low-bit quantization of neural networks for efficient inference. ICCVW, 2019. 2, 7

    2019

  9. [17]

    Distillation-free one-step diffusion for real-world image super-resolution

    Jianze Li, Jiezhang Cao, Zichen Zou, Xiongfei Su, Xin Yuan, Yulun Zhang, Yong Guo, and Xi- aokang Yang. Distillation-free one-step diffusion for real-world image super-resolution. arXiv preprint arXiv:2410.04224, 2024. 1

    2024 arXiv

  10. [18]

    Q-diffusion: Quantizing diffusion models

    Xiuyu Li, Yijiang Liu, Long Lian, Huanrui Yang, Zhen Dong, Daniel Kang, Shanghang Zhang, and Kurt Keutzer. Q-diffusion: Quantizing diffusion models. ICCV, 2023. 2, 3, 6, 7

    2023

  11. [19]

    Brecq: Pushing the limit of post-training quan- tization by block reconstruction

    Yuhang Li, Ruihao Gong, Xu Tan, Yang Yang, Peng Hu, Qi Zhang, Fengwei Yu, Wei Wang, and Shi Gu. Brecq: Pushing the limit of post-training quan- tization by block reconstruction. arXiv preprint arXiv:2102.05426, 2021. 3

    2021 arXiv

  12. [20]

    Q-dm: An efficient low-bit quan- tized diffusion model

    Yanjing Li, Sheng Xu, Xianbin Cao, Xiao Sun, and Baochang Zhang. Q-dm: An efficient low-bit quan- tized diffusion model. NeurIPS, 2024. 2, 3

    2024

  13. [21]

    Swinir: Image restoration using swin transformer

    Jingyun Liang, Jiezhang Cao, Guolei Sun, Kai Zhang, Luc Van Gool, and Radu Timofte. Swinir: Image restoration using swin transformer. ICCV, 2021. 1, 3

    2021

  14. [22]

    Awq: Activation-aware weight quantization for on-device llm compression and acceleration.Proceedings of Ma- chine Learning and Systems, 2024

    Ji Lin, Jiaming Tang, Haotian Tang, Shang Yang, Wei-Ming Chen, Wei-Chen Wang, Guangxuan Xiao, Xingyu Dang, Chuang Gan, and Song Han. Awq: Activation-aware weight quantization for on-device llm compression and acceleration.Proceedings of Ma- chine Learning and Systems, 2024. 5

    2024

  15. [23]

    Diffbir: Towards blind image restoration with generative diffusion prior

    Xinqi Lin, Jingwen He, Ziyan Chen, Zhaoyang Lyu, Bo Dai, Fanghua Yu, Wanli Ouyang, Yu Qiao, and Chao Dong. Diffbir: Towards blind image restoration with generative diffusion prior. ECCV, 2024. 1, 3, 7

    2024

  16. [24]

    Nonuniform-to-uniform quantization: Towards accurate quantization via gen- eralized straight-through estimation

    Zechun Liu, Kwang-Ting Cheng, Dong Huang, Eric Xing, and Zhiqiang Shen. Nonuniform-to-uniform quantization: Towards accurate quantization via gen- eralized straight-through estimation. 2022. 3

    2022

  17. [25]

    Smoothquant+: Ac- curate and efficient 4-bit post-training weightquantiza- tion for llm

    Jiayi Pan, Chengcan Wang, Kaifu Zheng, Yangguang Li, Zhenyu Wang, and Bin Feng. Smoothquant+: Ac- curate and efficient 4-bit post-training weightquantiza- tion for llm. arXiv preprint arXiv:2312.03788, 2023. 4, 6

    2023 arXiv

  18. [26]

    Quantsr: ac- curate low-bit quantization for efficient image super- resolution

    Haotong Qin, Yulun Zhang, Yifu Ding, Xianglong Liu, Martin Danelljan, Fisher Yu, et al. Quantsr: ac- curate low-bit quantization for efficient image super- resolution. NeurIPS, 2024. 8

    2024

  19. [27]

    High- 9 resolution image synthesis with latent diffusion mod- els

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High- 9 resolution image synthesis with latent diffusion mod- els. CVPR, 2022. 3

    2022

  20. [28]

    Post-training quantization on diffusion models

    Yuzhang Shang, Zhihang Yuan, Bin Xie, Bingzhe Wu, and Yan Yan. Post-training quantization on diffusion models. CVPR, 2023. 2, 3

    2023

  21. [29]

    Omniquant: Omnidi- rectionally calibrated quantization for large language models

    Wenqi Shao, Mengzhao Chen, Zhaoyang Zhang, Peng Xu, Lirui Zhao, Zhiqian Li, Kaipeng Zhang, Peng Gao, Yu Qiao, and Ping Luo. Omniquant: Omnidi- rectionally calibrated quantization for large language models. ICLR, 2024. 4

    2024

  22. [30]

    Temporal dynamic quanti- zation for diffusion models

    Junhyuk So, Jungwon Lee, Daehyun Ahn, Hyungjun Kim, and Eunhyeok Park. Temporal dynamic quanti- zation for diffusion models. NeurIPS, 2024. 3

    2024

  23. [31]

    Denoising diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. ICLR, 2021. 1

    2021

  24. [32]

    Consistency models

    Yang Song, Prafulla Dhariwal, Mark Chen, and Ilya Sutskever. Consistency models. ICML, 2023. 1

    2023

  25. [33]

    Towards accurate post- training quantization for diffusion models

    Changyuan Wang, Ziwei Wang, Xiuwei Xu, Yansong Tang, Jie Zhou, and Jiwen Lu. Towards accurate post- training quantization for diffusion models. CVPR,

  26. [34]

    Quest: Low-bit diffu- sion model quantization via efficient selective finetun- ing

    Haoxuan Wang, Yuzhang Shang, Zhihang Yuan, Junyi Wu, Junchi Yan, and Yan Yan. Quest: Low-bit diffu- sion model quantization via efficient selective finetun- ing. arXiv preprint arXiv:2402.03666, 2024. 3

    2024 arXiv

  27. [35]

    Exploring clip for assessing the look and feel of im- ages

    Jianyi Wang, Kelvin CK Chan, and Chen Change Loy. Exploring clip for assessing the look and feel of im- ages. In AAAI, 2023. 6

    2023

  28. [36]

    Exploiting diffusion prior for real-world image super-resolution

    Jianyi Wang, Zongsheng Yue, Shangchen Zhou, Kelvin CK Chan, and Chen Change Loy. Exploiting diffusion prior for real-world image super-resolution. IJCV, 2024. 1, 3

    2024

  29. [37]

    Real-esrgan: Training real-world blind super- resolution with pure synthetic data

    Xintao Wang, Liangbin Xie, Chao Dong, and Ying Shan. Real-esrgan: Training real-world blind super- resolution with pure synthetic data. ICCV, 2021. 1, 3

    2021

  30. [38]

    Sinsr: diffusion- based image super-resolution in a single step

    Yufei Wang, Wenhan Yang, Xinyuan Chen, Yaohui Wang, Lanqing Guo, Lap-Pui Chau, Ziwei Liu, Yu Qiao, Alex C Kot, and Bihan Wen. Sinsr: diffusion- based image super-resolution in a single step. CVPR,

  31. [39]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. TIP, 2004. 6

    2004

  32. [40]

    Prolific- dreamer: High-fidelity and diverse text-to-3d genera- tion with variational score distillation

    Zhengyi Wang, Cheng Lu, Yikai Wang, Fan Bao, Chongxuan Li, Hang Su, and Jun Zhu. Prolific- dreamer: High-fidelity and diverse text-to-3d genera- tion with variational score distillation. NeurIPS, 2024. 1

    2024

  33. [41]

    Compo- nent divide-and-conquer for real-world image super- resolution

    Pengxu Wei, Ziwei Xie, Hannan Lu, Zongyuan Zhan, Qixiang Ye, Wangmeng Zuo, and Liang Lin. Compo- nent divide-and-conquer for real-world image super- resolution. In ECCV, 2020. 6

    2020

  34. [42]

    One-step effective diffusion network for real- world image super-resolution

    Rongyuan Wu, Lingchen Sun, Zhiyuan Ma, and Lei Zhang. One-step effective diffusion network for real- world image super-resolution. NeurIPS, 2024. 1, 2, 3, 4, 6, 7

    2024

  35. [43]

    Seesr: Towards semantics-aware real-world image super-resolution

    Rongyuan Wu, Tao Yang, Lingchen Sun, Zhengqiang Zhang, Shuai Li, and Lei Zhang. Seesr: Towards semantics-aware real-world image super-resolution. CVPR, 2024. 1, 3

    2024

  36. [44]

    Qa-lora: Quantization-aware low-rank adaptation of large language models

    Yuhui Xu, Lingxi Xie, Xiaotao Gu, Xin Chen, Heng Chang, Hengheng Zhang, Zhengsu Chen, Xiaopeng Zhang, and Qi Tian. Qa-lora: Quantization-aware low-rank adaptation of large language models. ICLR,

  37. [45]

    Maniqa: Multi-dimension attention network for no-reference image quality assessment

    Sidi Yang, Tianhe Wu, Shuwei Shi, Shanshan Lao, Yuan Gong, Mingdeng Cao, Jiahao Wang, and Yu- jiu Yang. Maniqa: Multi-dimension attention network for no-reference image quality assessment. In CVPR,

  38. [46]

    Zero- quant: Efficient and affordable post-training quantiza- tion for large-scale transformers

    Zhewei Yao, Reza Yazdani Aminabadi, Minjia Zhang, Xiaoxia Wu, Conglong Li, and Yuxiong He. Zero- quant: Efficient and affordable post-training quantiza- tion for large-scale transformers. NeurIPS, 2022. 3

    2022

  39. [47]

    One-step diffusion with distribution matching distillation

    Tianwei Yin, Micha ¨el Gharbi, Richard Zhang, Eli Shechtman, Fredo Durand, William T Freeman, and Taesung Park. One-step diffusion with distribution matching distillation. CVPR, 2024. 1

    2024

  40. [48]

    Ptq4vit: Post-training quantization framework for vision transformers with twin uniform quantization

    Zhihang Yuan, Chenhao Xue, Yiqi Chen, Qiang Wu, and Guangyu Sun. Ptq4vit: Post-training quantization framework for vision transformers with twin uniform quantization. ECCV, 2022. 5

    2022

  41. [49]

    Designing a practical degradation model for deep blind image super-resolution

    Kai Zhang, Jingyun Liang, Luc Van Gool, and Radu Timofte. Designing a practical degradation model for deep blind image super-resolution. ICCV, 2021. 1, 3

    2021

  42. [50]

    Lin Zhang, Lei Zhang, and Alan C. Bovik. A feature- enriched completely blind image quality evaluator. TIP, 2015. 6

    2015

  43. [51]

    The unreasonable ef- fectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable ef- fectiveness of deep features as a perceptual metric. In CVPR, 2018. 6

    2018

  44. [52]

    Image super-resolution using very deep residual channel attention networks

    Yulun Zhang, Kunpeng Li, Kai Li, Lichen Wang, Bi- neng Zhong, and Yun Fu. Image super-resolution using very deep residual channel attention networks. ECCV, 2018. 3

    2018

  45. [53]

    Residual dense network for image super- resolution

    Yulun Zhang, Yapeng Tian, Yu Kong, Bineng Zhong, and Yun Fu. Residual dense network for image super- resolution. CVPR, 2018. 1

    2018

  46. [54]

    Unipc: A unified predictor-corrector framework for fast sampling of diffusion models

    Wenliang Zhao, Lujia Bai, Yongming Rao, Jie Zhou, and Jiwen Lu. Unipc: A unified predictor-corrector framework for fast sampling of diffusion models. NeurIPS, 2024. 1 10

    2024

Pith tools

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