Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Efficient Pruning of Text-to-Image Models: Insights from Pruning Stable Diffusion

T0 review · 4 major / 5 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 Simple magnitude pruning outperforms Wanda and SparseGPT for Stable Diffusion 2, which can be pruned to 38.5% sparsity with minimal quality loss.

desk verdict Useful empirical report on pruning Stable Diffusion 2, but the headline 38.5% claim rests on unreplicated FID/CLIP numbers and a post hoc optimum. read the letter →

arxiv 2411.15113 v1 pith:X42CA4KZ submitted 2024-11-22 cs.CV cs.AIcs.CLcs.LG

Samarth N Ramesh , Zhixue Zhao This is my paper · ORCID
classification cs.CVcs.AIcs.CLcs.LG
keywords ComputerVisionNaturalLanguageProcessingModelCompressionPruningText-to-ImageGenerationStableDiffusionPost-trainingMagnitude
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

This paper asks whether post-training pruning, a compression technique proven on large language models, transfers to text-to-image generation models like Stable Diffusion 2. It finds that the simple baseline, magnitude pruning, beats the more sophisticated Wanda and SparseGPT methods in this multimodal setting. The paper's main quantitative claim is that Stable Diffusion 2 can be pruned to 38.5% total sparsity with minimal quality loss, dropping FID from 18.07 to 18.15 and CLIP score from 0.314 to 0.311. It also reports sharp collapse thresholds in each component: the CLIP text encoder degrades suddenly beyond roughly 60-62.5% sparsity, while the U-Net generator declines gradually. If correct, this means a straightforward magnitude-based pruning recipe can make billion-parameter text-to-image models substantially smaller for edge deployment.

What carries the argument

The central objects are the two sub-models of Stable Diffusion 2: the CLIP text encoder and the U-Net diffusion generator, evaluated separately and together at sparsity levels stepping by 10% (and finer near thresholds). The load-bearing mechanism is magnitude pruning, which zeroes weights with the smallest absolute values, compared against Wanda, which weights magnitudes by activation norms, and OWL, which allocates layerwise sparsity based on outlier density. The argument is carried by the FID and CLIP Score metrics computed on 10,000 generated images, with the observed sharp thresholds used to set the optimal per-component sparsities.

What would settle it

Re-run the recommended 38.5% sparsity configuration (text 47.5%, image 35%) and the unpruned baseline multiple times with different random seeds, computing FID and CLIP Score over the same 10,000 MSCOCO prompts; if the spread across runs exceeds roughly 0.1 FID, the minimal-quality-loss claim is not established. For the magnitude-wins claim, a decisive test is to hold constant the evaluation protocol and swap only the pruning method at 50% text-encoder sparsity, comparing the resulting FID distributions rather than point estimates.

Watch

Extended reading notes

Core claim

Stable Diffusion 2, a 1.2-billion-parameter text-to-image model with a 340M-parameter CLIP text encoder and an 860M-parameter U-Net, can be post-training pruned with magnitude pruning to 38.5% total sparsity (47.5% text, 35% image) with negligible degradation: FID 18.15 vs 18.07 and CLIP Score 0.311 vs 0.314 on 10,000 MSCOCO captions. Contrary to the established trend in large language model pruning, activation-aware Wanda pruning and outlier-weighted OWL are consistently worse than or equal to plain magnitude pruning in this context. The paper interprets this as evidence that the information encoding in text-to-image models is different: both the text encoder and diffusion generator show sharp drop-off thresholds rather than graceful degradation, implying that specific weights carry critical semantic information whose removal collapses generation quality.

Load-bearing premise

The central quantitative claims depend on the stability of one FID and one CLIP Score measurement per configuration, with an observed gap as small as 0.08 FID and no repeated runs, seeds, or confidence intervals.

Editorial extensions

If this is right

  • If the central claim holds, a trivial magnitude-based pruning script, with no activation statistics and no retraining, is the practical choice for compressing Stable Diffusion 2-class models, and Wanda-style methods should not be transferred from LLMs without adaptation.
  • The 38.5% sparsity configuration would cut roughly 460 million parameters from the 1.2B model while keeping image quality on par with the full model, easing memory and latency constraints for on-device deployment.
  • Because the diffusion generator holds 72% of the parameters, allocating the majority of sparsity to the U-Net yields the best full-model quality, a direct corollary of the component-wise results.
  • Pruning either component beyond its threshold induces catastrophic degradation, so any practical pruning scheme must respect the text-encoder threshold near 60-62.5% and the generator's roughly 50% qualitative threshold.
  • The sharp thresholds suggest that saliency analysis of weights just below the collapse point could reveal which parameters encode prompt semantics, opening a window into interpretability and potential bias identification.

Reading between the lines

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

  • A direct consequence the paper leaves implicit: single-run FID on 10,000 images likely has non-negligible stochastic variance, so the advertised 0.08 FID gap between the pruned and unpruned model should be re-measured with repeated seeds and confidence intervals to be sure the 'minimal quality loss' conclusion is not measurement noise.
  • The claim that specific weights encode critical semantics could be tested directly: prune the weights just below and just above the collapse threshold, and probe how prompts alter failure modes; if failure is prompt-specific, it would support a semantic-locality interpretation, not just a generic capacity effect.
  • If the magnitude-over-Wanda result generalizes to other text-to-image models, such as SDXL or Stable Diffusion 3, then pruning research for generative vision should focus on conditional layers like cross-attention rather than on activation outlier statistics; a testable extension is to run the same component-wise sparsity sweep on those models.
  • The paper's optimal 47.5/35 split is computed for SD2's parameter mix; for architectures with a different text-encoder fraction, the same threshold-relative reasoning would shift the optimal split, and a simple algebraic rescaling of the thresholds could be checked on another model.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper presents a post-training pruning study of Stable Diffusion 2, comparing magnitude pruning, Wanda, SparseGPT (claimed but not tested), and OWL-based variants across text-encoder-only, U-Net-only, and full-model pruning. The authors report that magnitude pruning outperforms Wanda, that Stable Diffusion 2 can be pruned to 38.5% total sparsity (47.5% text encoder, 35% U-Net) with minimal quality loss, and that both components exhibit sharp performance drop-off thresholds. The central claims are evaluated with FID and CLIP Score on 10,000 MSCOCO captions, with qualitative image examples supporting the quantitative results.

Significance. If the findings hold, the paper would provide a practically useful result: for at least one text-to-image model, simple magnitude pruning is competitive or superior to activation-aware pruning, and substantial sparsity can be achieved with little degradation. The study is also valuable for separately examining text-encoder and U-Net pruning, and for releasing code. The paper has no fitted free parameters or circular derivation; the main methodological risk is the unreplicated evaluation, since the headline differences (e.g., 0.08 FID) are small relative to known stochastic variation in diffusion sampling and FID estimation. The sharp-threshold observation is interesting but similarly rests on single-run numbers.

major comments (4)
  1. [§3.4, Tables 5 and 6] The central quantitative claims are based on a single FID and a single CLIP Score per configuration, with no repeated runs, seeds, or confidence intervals. The headline result of 'minimal quality loss' at 38.5% sparsity rests on FID 18.15 vs. 18.07 and CLIP 0.311 vs. 0.314. Table 6 itself shows that magnitude-pruned configurations at 36% and 33.5% sparsity achieve FID 16.66 and 16.53, both below the unpruned baseline of 18.07; Table 10 shows the same 38.5% configuration with OWL at FID 17.97. This pattern strongly suggests that run-to-run or seed-dependent variation is comparable to or larger than the reported degradation. The authors should provide repeated evaluations with error bars, or at minimum a formal statistical test, before the 'minimal quality loss' and sharp-threshold claims can be considered established.
  2. [§4.4, Second Approach and Table 6] The 'optimal' configuration of 47.5% text sparsity and 35% image sparsity is selected post hoc from a sweep without a stated selection criterion. The text says the authors 'recommend' this configuration based on qualitative evaluations and improvements as sparsity is reduced, but Table 6 shows that lower-sparsity configurations (36% and 33.5%) have better or equal FID and CLIP Score than the 38.5% configuration. If the goal is maximal sparsity with minimal loss, the paper needs a predefined acceptable-degradation threshold; if the goal is simply the best quality-sparsity trade-off, the 36% or 33.5% configurations appear superior. A formal selection rule is needed to justify calling 38.5% 'optimal.'
  3. [§6, Future Work] The paper states that 'both Wanda and SparseGPT do not effectively extend to these models,' but SparseGPT is never tested in the experiments. The experimental comparison is limited to magnitude pruning, Wanda, and their OWL variants. The unsupported claim about SparseGPT weakens the general conclusion that magnitude pruning outperforms 'more advanced techniques.' Either implement SparseGPT, or restrict the claim to the methods actually evaluated.
  4. [§4.2, Figures 10 and 11] The U-Net-only pruning threshold used in the second full-model approach is identified as 50% sparsity, but the text says performance 'steadily decreases as sparsity exceeded 40%,' and Figure 11 states that performance 'greatly suffers beyond 40%.' The quantitative basis for choosing 50% as the threshold is not reported, and the inconsistency between 40% and 50% should be resolved. Since the second approach starts from this threshold, a clear and quantitative justification is needed.
minor comments (5)
  1. [§3.4] The text says 'a lower CLIP Score indicates a higher correlation,' but the CLIP Score is a similarity measure where higher values indicate better text-image alignment; this is inconsistent with the tables, where better configurations have higher CLIP scores.
  2. [Appendix A, Tables 7 and 8] The entries listed as '0' for FID and CLIP Score in presumably invalid configurations should be marked as 'N/A' rather than 0, since a FID of 0 would indicate perfect similarity and a CLIP Score of 0 is not meaningful.
  3. [References] References [32] and [33] appear to cite the same latent diffusion paper twice, once as an arXiv preprint and once as the CVPR version; this duplication should be consolidated.
  4. [Abstract] The abstract mentions 'interoperability' in the context of future research; this appears to be a typo for 'interpretability.'
  5. [§4.1 and Figure 9] The claim that Wanda 'consistently outperforms' magnitude at every sparsity level is clear from Figure 9, but the figure is not accompanied by the underlying numeric values; adding a table of text-encoder-only results would make the comparison reproducible.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an empirical benchmark study with no fitted parameters, no load-bearing self-citations, and an external FID anchor; the acknowledged CLIPScore caveat is an evaluation confound, not a circular derivation.

full rationale

This paper does not contain a derivation chain in which a prediction reduces to its inputs by construction. The central claims are empirical: magnitude pruning outperforms Wanda and OWL on Stable Diffusion 2, and pruning the text encoder to 47.5% and the diffusion generator to 35% yields 38.5% total sparsity with FID 18.15 vs 18.07 and CLIPScore 0.311 vs 0.314 (Tables 5, 6, 10). The pruning masks are produced by fixed, externally defined algorithms (magnitude, Wanda, OWL) and are not fitted to the reported FID or CLIP scores; the optimal configuration is selected by evaluating pre-specified sparsity levels near previously identified thresholds (Section 4.4, Second Approach), not by optimizing a parameter that is then renamed as a prediction. No load-bearing self-citations appear in the reference list, and no uniqueness theorem is imported from the authors' prior work. The only mild circularity-adjacent point is that CLIPScore uses a CLIP model and the pruned text encoder is also a CLIP model; the authors explicitly acknowledge this as a potential bias in Section 4.3. This is an evaluation confound rather than a circular step: CLIPScore is computed from generated images and prompts with a fixed reference model, and the pruning decisions are not made by optimizing CLIPScore. FID provides an independent Inception-based external anchor. The absence of repeated runs, seeds, and confidence intervals (Section 3.4) is a statistical robustness concern that could undermine the certainty of the 38.5% 'minimal quality loss' claim, but it is not a circularity failure under the definitions used here.

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

This is an empirical benchmark rather than a derivation, so there are no fitted free parameters and no invented entities. The claims rest on measurement assumptions: single-run FID/CLIP scores are stable, the 10k COCO subset is representative, zeroed weights translate into practical savings, and CLIPScore is not unduly biased by sharing a model family with the pruned text encoder.

assumptions (4)
  • domain assumption Single-run FID/CLIP evaluations on 10,000 generated images resolve differences as small as 0.08 FID, which is used to claim minimal quality loss.
    Section 3.4 and Tables 5-6 report one measurement per configuration, with no repeats, seeds, or confidence intervals.
  • domain assumption The random 10,000-image MSCOCO 2017 subset is representative of text-to-image evaluation.
    Section 3.3; no seed or subset file is provided, so exact replication is not possible.
  • domain assumption Unstructured weight zeroing yields the claimed reduction in model size and computational requirements.
    The abstract and intro claim size and compute savings, but no sparse inference, latency, or memory measurements are reported.
  • domain assumption CLIPScore computed with an external CLIP model is an unbiased judge of prompt-image alignment for a pruned CLIP text encoder.
    Section 4.3 notes a potential bias because the pruned encoder is itself CLIP; FID provides an independent check, but CLIPScore-based comparisons are partly confounded.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient Pruning of Text-to-Image Models: Insights from Pruning Stable Diffusion." pith.science (2026). https://pith.science/paper/X42CA4KZ

@misc{pith2026241115113,
  author       = {Pith},
  title        = {Pith review of: Efficient Pruning of Text-to-Image Models: Insights from Pruning Stable Diffusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X42CA4KZ}},
  note         = {Machine review of arXiv:2411.15113}
}
read the original abstract

As text-to-image models grow increasingly powerful and complex, their burgeoning size presents a significant obstacle to widespread adoption, especially on resource-constrained devices. This paper presents a pioneering study on post-training pruning of Stable Diffusion 2, addressing the critical need for model compression in text-to-image domain. Our study tackles the pruning techniques for the previously unexplored multi-modal generation models, and particularly examines the pruning impact on the textual component and the image generation component separately. We conduct a comprehensive comparison on pruning the model or the single component of the model in various sparsities. Our results yield previously undocumented findings. For example, contrary to established trends in language model pruning, we discover that simple magnitude pruning outperforms more advanced techniques in text-to-image context. Furthermore, our results show that Stable Diffusion 2 can be pruned to 38.5% sparsity with minimal quality loss, achieving a significant reduction in model size. We propose an optimal pruning configuration that prunes the text encoder to 47.5% and the diffusion generator to 35%. This configuration maintains image generation quality while substantially reducing computational requirements. In addition, our work uncovers intriguing questions about information encoding in text-to-image models: we observe that pruning beyond certain thresholds leads to sudden performance drops (unreadable images), suggesting that specific weights encode critical semantics information. This finding opens new avenues for future research in model compression, interoperability, and bias identification in text-to-image models. By providing crucial insights into the pruning behavior of text-to-image models, our study lays the groundwork for developing more efficient and accessible AI-driven image generation systems

Figures

Figures reproduced from arXiv: 2411.15113 by the authors.

Figure 1
Figure 1. Pruning only Text Encoder of Stable Diffusion 2 [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 4
Figure 4. Text Encoder Wanda Pruning Examples - Text En [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Pruning only Text Encoder of Stable Diffusion 2 [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figures from the paper (9 more)
Figure 8
Figure 8. Figure 8: Text Encoder Wanda Pruning with OWL Examples [PITH_FULL_IMAGE:figures/full_fig_p005_8.png]
Figure 6
Figure 6. Figure 6: Text Encoder Magnitude Pruning with OWL Exam [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Pruning only Text Encoder of Stable Diffusion 2 [PITH_FULL_IMAGE:figures/full_fig_p005_7.png]
Figure 9
Figure 9. Figure 9: Comparing Pruning Techniques for pruning only [PITH_FULL_IMAGE:figures/full_fig_p006_9.png]
Figure 10
Figure 10. Figure 10: Pruning only Image Diffusion Generator of Stable [PITH_FULL_IMAGE:figures/full_fig_p006_10.png]
Figure 11
Figure 11. Figure 11: Diffusion Generator Magnitude Pruning Examples [PITH_FULL_IMAGE:figures/full_fig_p006_11.png]
Figure 12
Figure 12. Figure 12: Model Improvement as sparsity is reduced from [PITH_FULL_IMAGE:figures/full_fig_p007_12.png]
Figure 13
Figure 13. Figure 13: Base Model vs Optimally Pruned Model at 38.5% [PITH_FULL_IMAGE:figures/full_fig_p008_13.png]
Figure 14
Figure 14. Figure 14: Magnitude Pruning Diffusion Generator causes [PITH_FULL_IMAGE:figures/full_fig_p009_14.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Bridging Diffusion Pruning and Step Distillation with Teacher-Aligned Repair

    cs.CV 2026-07 conditional novelty 6.0 of 10

    A short teacher-alignment repair stage between structured pruning and one-step distillation yields a 20% pruned one-step generator that improves FID from 3.53 to 3.12 on ImageNet-512 while reducing NFE from 63 to 1.

Reference graph

Works this paper leans on

42 extracted references · 18 canonical work pages · cited by 1 Pith paper

  1. [1]

    Andrew Brock. 2018. Large Scale GAN Training for High Fidelity Natural Image Synthesis. arXiv preprint arXiv:1809.11096 (2018)

    arXiv 2018

  2. [2]

    Tom B Brown. 2020. Language models are few-shot learners. arXiv preprint arXiv:2005.14165 (2020)

    arXiv 2020

  3. [3]

    Thibault Castells, Hyoung-Kyu Song, Bo-Kyeong Kim, and Shinkook Choi. 2024. LD-Pruner: Efficient Pruning of Latent Diffusion Models using Task-Agnostic Insights. http://arxiv.org/abs/2404.11936 arXiv:2404.11936 [cs]

    arXiv 2024

  4. [4]

    Tim Dettmers, Mike Lewis, Younes Belkada, and Luke Zettlemoyer. 2022. Gpt3. int8 (): 8-bit matrix multiplication for transformers at scale. Advances in Neural Information Processing Systems 35 (2022), 30318–30332

    2022

  5. [5]

    Prafulla Dhariwal and Alexander Nichol. 2021. Diffusion models beat gans on image synthesis. Advances in neural information processing systems 34 (2021), 8780–8794. Efficient Pruning of Text-to-Image Models: Insights from Pruning Stable Diffusion SAC’25, March 31 –April 4, 2025, Sicily, Italy Total Sparsity Text:Image Ratio Text Sparsity Image Sparsity FI...

    work page 2021

  6. [6]

    Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, and Robin Rombach. 2024. Scaling Rectified Flow Transformers for High-Resolution Image Synthesis

    work page 2024

  7. [7]

    Gongfan Fang, Xinyin Ma, and Xinchao Wang. 2023. Structural Pruning for Diffusion Models. In NeurIPS

    work page 2023

  8. [8]

    Elias Frantar and Dan Alistarh. 2023. SparseGPT: Massive Language Models Can be Accurately Pruned in One-Shot. In ICML (Proceedings of Machine Learning Research, Vol. 202). PMLR, 10323–10337

    work page 2023

Show all 42 references
  1. [9]

    Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde- Farley, Sherjil Ozair, Aaron C

    Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde- Farley, Sherjil Ozair, Aaron C. Courville, and Yoshua Bengio. 2020. Generative adversarial networks. Commun. ACM 63, 11 (2020), 139–144

    2020

  2. [10]

    Gray and D.L

    R.M. Gray and D.L. Neuhoff. 1998. Quantization. IEEE Transactions on Infor- mation Theory 44, 6 (Oct. 1998), 2325–2383. https://doi.org/10.1109/18.720541 Conference Name: IEEE Transactions on Information Theory

    1998 doi

  3. [11]

    Ramyad Hadidi, Jiashen Cao, Yilun Xie, Bahar Asgari, Tushar Krishna, and Hye- soon Kim. 2019. Characterizing the Deployment of Deep Neural Networks on Commercial Edge Devices. In 2019 IEEE International Symposium on Work- load Characterization (IISWC). 35–48. https://doi.org/1...

    2019

  4. [12]

    Song Han, Jeff Pool, John Tran, and William J. Dally. 2015. Learning both Weights and Connections for Efficient Neural Network. In NIPS. 1135–1143

    2015

  5. [13]

    Babak Hassibi, David G Stork, and Gregory J Wolff. 1993. Optimal brain surgeon and general network pruning. In IEEE international conference on neural networks . IEEE, 293–299

    1993

  6. [14]

    Benjamin Hawks, Javier Duarte, Nicholas J Fraser, Alessandro Pappalardo, Nhan Tran, and Yaman Umuroglu. 2021. Ps and qs: Quantization-aware pruning for efficient low latency neural network inference. Frontiers in Artificial Intelligence 4 (2021), 676564

    2021

  7. [15]

    Jack Hessel, Ari Holtzman, Maxwell Forbes, Ronan Le Bras, and Yejin Choi. 2021. CLIPScore: A Reference-free Evaluation Metric for Image Captioning. In EMNLP (1). Association for Computational Linguistics, 7514–7528

    2021

  8. [16]

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. 2017. GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium. In NIPS. 6626–6637

    2017

  9. [17]

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. 2020. Denoising Diffusion Probabilistic Models. In NeurIPS

    2020

  10. [18]

    Tero Karras, Samuli Laine, and Timo Aila. 2019. A style-based generator ar- chitecture for generative adversarial networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 4401–4410

    2019

  11. [19]

    Hyung-Kwon Ko, Gwanmo Park, Hyeon Jeon, Jaemin Jo, Juho Kim, and Jinwook Seo. 2023. Large-scale text-to-image generation models for visual artists’ cre- ative works. In Proceedings of the 28th international conference on intelligent user interfaces. 919–933

    2023

  12. [20]

    Olga Kovaleva, Saurabh Kulshreshtha, Anna Rogers, and Anna Rumshisky. 2021. BERT busters: Outlier dimensions that disrupt transformers. arXiv preprint arXiv:2105.06990 (2021)

    2021 arXiv

  13. [21]

    Yann LeCun, John Denker, and Sara Solla. 1989. Optimal Brain Damage. In Advances in Neural Information Processing Systems , D. Touretzky (Ed.), Vol. 2. Morgan-Kaufmann. https://proceedings.neurips.cc/paper_files/paper/1989/file/ 6c9882bbac1c7093bd25041881277658-Paper.pdf

    1989

  14. [22]

    Tailin Liang, John Glossner, Lei Wang, Shaobo Shi, and Xiaotong Zhang. 2021. Pruning and quantization for deep neural network acceleration: A survey. Neu- rocomputing 461 (2021), 370–403

    2021

  15. [23]

    Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C

    Tsung-Yi Lin, Michael Maire, Serge J. Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C. Lawrence Zitnick. 2014. Microsoft COCO: Common Objects in Context. In ECCV (5) (Lecture Notes in Computer Science, Vol. 8693) . Springer, 740–755

    2014

  16. [24]

    Alexandre Lopes, Fernando Pereira dos Santos, Diulhio de Oliveira, Mauricio Schiezaro, and Helio Pedrini. 2024. Computer Vision Model Compression Tech- niques for Embedded Systems:A Survey.Computers & Graphics 123 (2024), 104015. https://doi.org/10.1016/j.cag.2024.104015

    2024

  17. [25]

    Deepak Narayanan, Mohammad Shoeybi, Jared Casper, Patrick LeGresley, Mostofa Patwary, Vijay Korthikanti, Dmitri Vainbrand, Prethvi Kashinkunti, Julie Bernauer, Bryan Catanzaro, Amar Phanishayee, and Matei Zaharia. 2021. Efficient large-scale language model training on GPU clus...

    2021

  18. [26]

    Wei Niu, Xiaolong Ma, Sheng Lin, Shihao Wang, Xuehai Qian, Xue Lin, Yanzhi Wang, and Bin Ren. 2020. Patdnn: Achieving real-time dnn execution on mobile devices with pattern-based weight pruning. In Proceedings of the Twenty-Fifth International Conference on Architectural Suppo...

    2020

  19. [27]

    Pierre-Emmanuel Novac, Ghouthi Boukli Hacene, Alain Pegatoquet, Benoît Mira- mond, and Vincent Gripon. 2021. Quantization and Deployment of Deep Neural Networks on Microcontrollers. Sensors 21, 9 (2021). https://doi.org/10.3390/ s21092984 SAC’25, March 31 –April 4, 2025, Sicil...

    2021

  20. [28]

    Pavana Prakash, Jiahao Ding, Rui Chen, Xiaoqi Qin, Minglei Shu, Qimei Cui, Yuanxiong Guo, and Miao Pan. 2022. IoT device friendly and communication- efficient federated learning via joint model pruning and quantization. IEEE Internet of Things Journal 9, 15 (2022), 13638–13650

    2022

  21. [29]

    Giovanni Puccetti, Anna Rogers, Aleksandr Drozd, and Felice Dell’Orletta. 2022. Outliers dimensions that disrupt transformers are driven by frequency. arXiv preprint arXiv:2205.11380 (2022)

    2022 arXiv

  22. [30]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. Learning Transferable Visual Models From Natural Language Supervision. In ICML (Proce...

    2021

  23. [31]

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen

  24. [32]

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2021. High-Resolution Image Synthesis with Latent Diffusion Models. arXiv:2112.10752 [cs.CV]

    2021 arXiv

  25. [33]

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2022. High-Resolution Image Synthesis with Latent Diffusion Models. In CVPR. IEEE, 10674–10685

    2022

  26. [34]

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. 2015. U-net: Convolu- tional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceeding...

    2015

  27. [35]

    Weiss, Niru Maheswaranathan, and Surya Ganguli

    Jascha Sohl-Dickstein, Eric A. Weiss, Niru Maheswaranathan, and Surya Ganguli

  28. [36]

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. 2020. Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456 (2020)

    2020 arXiv

  29. [37]

    Zico Kolter

    Mingjie Sun, Zhuang Liu, Anna Bair, and J. Zico Kolter. 2024. A Simple and Effec- tive Pruning Approach for Large Language Models. In ICLR. OpenReview.net

    2024

  30. [38]

    William Timkey and Marten Van Schijndel. 2021. All bark and no bite: Rogue dimensions in transformer language models obscure representational quality. arXiv preprint arXiv:2109.04404 (2021)

    2021 arXiv

  31. [39]

    Ling Yang, Zhilong Zhang, Yang Song, Shenda Hong, Runsheng Xu, Yue Zhao, Wentao Zhang, Bin Cui, and Ming-Hsuan Yang. 2023. Diffusion Models: A Comprehensive Survey of Methods and Applications. ACM Comput. Surv. 56, 4, Article 105 (Nov. 2023), 39 pages. https://doi.org/10.1145/3626235

    2023 doi

  32. [40]

    Lu Yin, You Wu, Zhenyu Zhang, Cheng-Yu Hsieh, Yaqing Wang, Yiling Jia, Gen Li, Ajay Kumar Jaiswal, Mykola Pechenizkiy, Yi Liang, Michael Bendersky, Zhangyang Wang, and Shiwei Liu. 2024. Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High S...

    2024

  33. [2015]

    In ICML (JMLR Workshop and Conference Proceedings, Vol

    Deep Unsupervised Learning using Nonequilibrium Thermodynamics. In ICML (JMLR Workshop and Conference Proceedings, Vol. 37). JMLR.org, 2256–2265

  34. [2022]

    arXiv preprint arXiv:2204.06125 1, 2 (2022), 3

    Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125 1, 2 (2022), 3

    2022 arXiv

Pith tools

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