Pith. sign in

REVIEW 3 major objections 6 minor 54 references

Sparsity Outperforms Low-Rank Projections in Few-Shot Adaptation

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A sparse optimizer that randomly updates a tiny parameter subset outperforms low-rank projection methods in few-shot CLIP adaptation.

desk verdict Useful sparse optimizer with excellent ablations, but the SOTA claim rests on an unfair baseline comparison and missing error bars. read the letter →

arxiv 2504.12436 v2 pith:CA2MS7I4 submitted 2025-04-16 cs.CV cs.AI

Nairouz Mrabah , Nicolas Richet , Ismail Ben Ayed , Éric Granger This is my paper · ORCID
classification cs.CVcs.AI
keywords few-shotadaptationvision-languagemodelssparseoptimizationlow-rankCLIPgradientsparsitymomentpruningparameter-efficientfine-tuning
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

Few-shot adaptation of vision-language models overfits when all parameters are updated, and low-rank projections constrain updates to a fixed subspace that still overfits and is hard to tune. This paper proposes a Sparse Optimization (SO) optimizer, an Adam-style method that keeps a very small density ratio (0.05%) of gradient entries per step, resamples the sparsity support every few iterations, and prunes the first moment by importance. On 11 datasets with 1, 2, and 4 shots, SO reports the highest average accuracy, surpassing LoRA, ReLoRA, DoRA, VeRA, PiSSA, and GaLore while storing only a fraction of their gradient and optimizer state. The point is that carefully chosen sparsity, not low-rank structure, is the better protective mechanism against few-shot overfitting.

What carries the argument

The engine is a sparse variant of Adam with two selection rules. At each step the optimizer retains $M = \lfloor \kappa d \rfloor$ gradient entries chosen uniformly at random (Random-M), then combines them with the previous sparse first moment; the union can hold up to $2M$ values, from which the Top-$M$ largest-magnitude entries are kept for the first moment, and the second moment is kept on the same indices. The sparsity support is refreshed every $T$ iterations. These mechanisms implement the paper's two paradigms, local sparsity with global density and local randomness with global importance, and they are what prevent the optimizer from rushing into an overfit subspace while keeping the update path full-rank over time.

What would settle it

Run LoRA or ReLoRA on the same 11 datasets with a per-dataset early-stopping budget chosen on a held-out split (or with 2000 iterations and a tuned rank), and check whether the average accuracy still trails SO; if the low-rank methods match or exceed 73.8, 76.6, and 78.9 at 1, 2, and 4 shots, the paper's central claim would no longer hold.

Watch

Extended reading notes

Core claim

The central claim is that restricting updates to a low-rank subspace is the wrong inductive bias for few-shot vision-language adaptation: the rank is a discrete hyperparameter that must be tuned per dataset, and fixed-subspace updates still overfit. The paper replaces this with a sparse optimizer, SO, that enforces local sparsity (only $M = \lfloor \kappa d \rfloor$ entries are updated at a time) and global density (the support resamples every $T$ iterations, so many parameters receive updates over training), plus local randomness (gradient entries are selected uniformly at random) and global importance (the first moment is pruned by Top-$M$ magnitude, with the second moment aligned to the same indices). With $\kappa = 0.05\%$ and $T = 10$, SO achieves the highest average accuracy at every shot count, including 73.8, 76.6, and 78.9 for 1, 2, and 4 shots on the 11 datasets, with the largest gains on EuroSAT, Stanford-Cars, Flowers, and UCF101, while keeping gradient and optimizer memory near 0.47 MB and 0.70 MB for the full CLIP model.

Load-bearing premise

The comparison assumes that stopping every method at the same training-loss threshold (0.01), with 500 iterations for low-rank baselines and 2000 for SO, is fair; if a longer or per-dataset-tuned budget for the baselines lifts their accuracy, SO's reported advantage could shrink.

Editorial extensions

If this is right

  • Extreme sparsity ($\kappa = 0.05\%$) is sufficient to adapt a full CLIP backbone competitively, with 1, 2, and 4-shot averages of 73.8, 76.6, and 78.9 across 11 datasets.
  • The performance ranking no longer hinges on a discrete rank choice: the ablations show dynamic sparsity support beats a fixed support, and random gradient selection beats importance-based selection at every shot count.
  • Memory overhead drops to about 0.47 MB for gradients and 0.70 MB for optimizer states on CLIP, compared with 468 MB for full Adam gradients and 936 MB for its optimizer states.
  • Low-rank baselines need an early-stopping threshold to avoid overfitting, while SO trains to the same loss threshold without rank tuning, reducing the number of hyperparameters that must be set in few-shot settings.
  • Because SO reduces to plain Adam when the density ratio is 1, it can serve as a drop-in optimizer on top of existing architectures rather than a reparameterization method.

Reading between the lines

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

  • If the support-resampling mechanism is the active ingredient, the principle should transfer to full fine-tuning of large language models, where memory is a bottleneck; the paper lists this as future work, and one direct test is to replace Adam in low-rank LLM training with SO and compare convergence.
  • The supplementary rank-evolution plots suggest random gradient selection keeps the gradient rank near full while importance-based selection collapses it, implying that SO's sparsity acts as a stochastic exploration mechanism; one could test whether SO's advantage disappears on nearly convex loss landscapes where random updates are pure noise.
  • The two paradigms are separable: importance-based moment pruning could be bolted onto existing adapters such as LoRA or VeRA, producing a hybrid low-rank-plus-sparse method that might outperform either family alone.
  • A testable extension is to vary the refresh interval $T$ as a schedule rather than a constant, since the paper's sensitivity study already shows smaller $T$ values improve accuracy at the cost of slower convergence.
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

3 major / 6 minor

Summary. The paper proposes Sparse Optimization (SO), an Adam-type optimizer that randomly selects a small fraction of gradient entries at each iteration, refreshes the support every T iterations, and prunes the first and second moments by magnitude, and applies it to few-shot adaptation of CLIP. The central claim is that this sparsity-based approach outperforms low-rank projection methods such as LoRA, ReLoRA, VeRA, DoRA, PiSSA, and GaLore on 11 datasets at 1, 2, and 4 shots, while reducing memory overhead. The paper also presents extensive ablations supporting the internal design choices: random gradient pruning, importance-based moment pruning, dynamic sparsity support, and the use of moments.

Significance. If the comparative claim were established, the result would be a useful contribution to parameter-efficient transfer learning, showing that a dynamically supported sparse optimizer can be a competitive alternative to low-rank reparameterization for few-shot VLM adaptation. The paper is strong in its internal validation: the ablations (Tables 2-5) are systematic, the method is clearly specified in Algorithm 1, code is provided, and the memory analysis in Appendix C is a useful reference. The main weakness is that the headline state-of-the-art claim rests on a comparison that is not yet fair: the low-rank baselines are run with fixed rank and a smaller iteration budget than SO, and the main results table contains no variance information.

major comments (3)
  1. [§4(a), Table 1, Fig. 2, Appendix F] The central claim that SO achieves state-of-the-art performance is not yet supported because the comparison in Table 1 is asymmetric. Low-rank baselines are run with rank fixed to 2 and a maximum of 500 iterations, while SO is run with κ=0.05%, T=10, and up to 2000 iterations. The paper's own Fig. 2 and Appendix F show that LoRA's accuracy depends strongly on rank, with the best rank varying across DTD, Pets, and UCF101, so rank=2 systematically understates at least some baselines on some datasets. Similarly, the 500-iteration cap truncates a baseline whenever its training loss has not reached 0.01 by iteration 500, while SO is permitted to continue to 2000 iterations. Please rerun the baselines under an equal budget or with per-method rank/budget tuning on a validation split, and report best-over-rank results or per-rank curves for all methods.
  2. [Table 1] Table 1 reports averages over only 3 random seeds with no standard deviations, error bars, or significance tests. The observed average advantages of SO over the strongest baseline (ReLoRA) are 1.3 points at 1 shot, 1.5 points at 2 shots, and 1.8 points at 4 shots; with three seeds these gaps could plausibly be within run-to-run noise. Please provide per-seed numbers or confidence intervals, and state whether the differences are statistically meaningful.
  3. [§4(a), Appendix D] The hyperparameters κ and T appear to be selected using the same benchmark datasets (Pets and DTD are shown in Fig. 3), and no validation split is described. Because the paper motivates the method by arguing that hyperparameter tuning is impractical in few-shot settings, the selection protocol must be clarified. If κ and T were chosen by inspecting test accuracy on the evaluation datasets, the comparison is optimistic; if they were chosen on a held-out validation split, that split should be described. This point is load-bearing for the generality claim that SO 'remains robust' without validation-based tuning.
minor comments (6)
  1. [Figure 1] The caption of Figure 1 states that LoRA uses rank fixed at 3, while Section 4(a) states that rank is set to 2 for all low-rank baselines; please reconcile this discrepancy.
  2. [Algorithm 1] In Algorithm 1, the while condition uses |L(Θ_{t-1})| > τ, but for a cross-entropy loss the absolute value is not meaningful and the iteration index appears off by one; please correct the stopping condition and indexing.
  3. [Section 4] The results subsections are labeled (a), (b), (d), (e), (f), (g); subsection (c) is missing and should be renumbered.
  4. [Table 6] The header of Table 6 concatenates method names without clear separators (e.g., 'SO GaLoRE LoRA'); please format the header as separate columns for readability.
  5. [References] References [20] and [54] are duplicates of the same VeRA paper; please cite it once.
  6. [Appendix G] Some standard deviations in Appendix G appear to be typos, such as '64.74±77.8' in Table 19 and '29.86±30.8' in Table 20; please check all reported variances.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: SO is an independently specified sparse optimizer; the SOTA claim is an empirical benchmark comparison, with only a minor protocol self-citation that is not load-bearing.

full rationale

The paper contains no derivation chain whose output is equivalent to its input. SO is specified independently (Eqs. 1-5) as Adam with random gradient pruning and Top-M moment pruning; its generalization claims are supported by ablations (Tables 2-5) and by external benchmark comparisons (Table 1) against published low-rank methods. Hyperparameters kappa=0.05% and T=10 are hand-chosen and explored in Appendix D; this is selection on benchmarks, not a fitted parameter renamed as a prediction. The only self-citation is the adoption of CLIP-LoRA [46] settings for low-rank baselines (rank=2, 500 iterations) in Section 4(a). [46] is a published paper with external code and is not used to derive the SO algorithm, so this is protocol citation, not load-bearing circularity. The paper's own Fig. 2 and Appendix F show LoRA's rank and iteration sensitivity, and Section 4(a) admits the early-stopping rule is 'necessary to obtain competitive results' for low-rank methods; these are legitimate fairness concerns about the SOTA comparison (asymmetric iteration budgets, fixed rank), but they are experimental-design issues, not circular reasoning. The central claim is an empirical result self-contained against external benchmarks.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The method introduces two hyperparameters (kappa and T) that are fixed across datasets but selected using benchmark datasets. No new physical entities are introduced. The main assumptions are standard transfer-learning assumptions and the fairness of the early-stopping protocol.

free parameters (3)
  • density ratio kappa = 0.05%
    Controls the fraction of gradient entries retained each update; chosen based on sensitivity experiments on Pets and DTD, which are part of the 11-dataset benchmark, and fixed across datasets.
  • support refresh interval T = 10 iterations
    Determines how often the random sparsity mask is re-sampled; chosen alongside kappa on the same benchmark datasets.
  • early stopping loss threshold = 0.01
    Common to all methods, but the threshold interacts with the differing iteration caps; not tuned on a held-out set.
assumptions (3)
  • domain assumption CLIP's pretrained representations are a valid starting point for few-shot adaptation via full-model fine-tuning.
    The method fine-tunes all layers of CLIP, assuming this does not destroy the pretrained features.
  • domain assumption Top-1 accuracy on the 11 selected datasets is a sufficient measure of few-shot generalization.
    The state-of-the-art claim is evaluated solely on these datasets.
  • ad hoc to paper The loss < 0.01 stopping rule and iteration caps yield comparable training effort across methods.
    Baselines are capped at 500 iterations while SO runs to 2000; the threshold is reached at different iteration counts per dataset.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sparsity Outperforms Low-Rank Projections in Few-Shot Adaptation." pith.science (2026). https://pith.science/paper/CA2MS7I4

@misc{pith2026250412436,
  author       = {Pith},
  title        = {Pith review of: Sparsity Outperforms Low-Rank Projections in Few-Shot Adaptation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CA2MS7I4}},
  note         = {Machine review of arXiv:2504.12436}
}
read the original abstract

Adapting Vision-Language Models (VLMs) to new domains with few labeled samples remains a significant challenge due to severe overfitting and computational constraints. State-of-the-art solutions, such as low-rank reparameterization, mitigate these issues but often struggle with generalization and require extensive hyperparameter tuning. In this paper, a novel Sparse Optimization (SO) framework is proposed. Unlike low-rank approaches that typically constrain updates to a fixed subspace, our SO method leverages high sparsity to dynamically adjust very few parameters. We introduce two key paradigms. First, we advocate for \textit{local sparsity and global density}, which updates a minimal subset of parameters per iteration while maintaining overall model expressiveness. As a second paradigm, we advocate for \textit{local randomness and global importance}, which sparsifies the gradient using random selection while pruning the first moment based on importance. This combination significantly mitigates overfitting and ensures stable adaptation in low-data regimes. Extensive experiments on 11 diverse datasets show that SO achieves state-of-the-art few-shot adaptation performance while reducing memory overhead.

Figures

Figures reproduced from arXiv: 2504.12436 by the authors.

Figure 1
Figure 1. Test accuracy over training iterations for LoRA (with rank fixed at 3) and SO (with a sparsity ratio equal to 99.95%) in a 1-shot [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. LoRA performance in a 1-shot setting on three datasets (DTD, Oxford Pets, and UCF101), using a pretrained CLIP (ViT-B/16) [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Test accuracy of SO for different density ratios [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: LoRA performance in a 2-shot setting on three datasets—DTD, Oxford Pets, and UCF101— using a pretrained CLIP with [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]
Figure 5
Figure 5. Figure 5: LoRA performance in a 4-shot setting on three datasets—DTD, Oxford Pets, and UCF101— using a pretrained CLIP with [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 6
Figure 6. Figure 6: Test accuracy of SO (importance-based gradient pruning) with density ratio [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 43 canonical work pages

  1. [1]

    Bit- Fit: Simple parameter-efficient fine-tuning for transformer- based masked language-models

    Elad Ben Zaken, Yoav Goldberg, and Shauli Ravfogel. Bit- Fit: Simple parameter-efficient fine-tuning for transformer- based masked language-models. In Annual Meeting of the Association for Computational Linguistics (ACL) , pages 1– 9, 2022. 4

    work page 2022

  2. [2]

    Food-101–mining discriminative components with random forests

    Lukas Bossard, Matthieu Guillaumin, and Luc Van Gool. Food-101–mining discriminative components with random forests. In European Conference on Computer Vision (ECCV), pages 446–461, 2014. 7

    work page 2014

  3. [3]

    Lorashear: Efficient large language model structured pruning and knowledge recovery

    Tianyi Chen, Tianyu Ding, Badal Yadav, Ilya Zharkov, and Luming Liang. Lorashear: Efficient large language model structured pruning and knowledge recovery. arXiv preprint arXiv:2310.18356, 2023. 4

    arXiv 2023

  4. [4]

    Fira: Can we achieve full-rank training of llms under low-rank constraint? arXiv preprint arXiv:2410.01623, 2024

    Xi Chen, Kaituo Feng, Changsheng Li, Xunhao Lai, Xi- angyu Yue, Ye Yuan, and Guoren Wang. Fira: Can we achieve full-rank training of llms under low-rank constraint? arXiv preprint arXiv:2410.01623, 2024

    arXiv 2024

  5. [5]

    A survey on deep neural network pruning: Taxonomy, compar- ison, analysis, and recommendations

    Hongrong Cheng, Miao Zhang, and Javen Qinfeng Shi. A survey on deep neural network pruning: Taxonomy, compar- ison, analysis, and recommendations. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) , pages 10558–10578, 2024. 4

    work page 2024

  6. [6]

    Reproducible scaling laws for contrastive language-image learning

    Mehdi Cherti, Romain Beaumont, Ross Wightman, Mitchell Wortsman, Gabriel Ilharco, Cade Gordon, Christoph Schuh- mann, Ludwig Schmidt, and Jenia Jitsev. Reproducible scaling laws for contrastive language-image learning. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 2818–2829, 2023. 1

    work page 2023

  7. [7]

    Describing textures in the wild

    Mircea Cimpoi, Subhransu Maji, Iasonas Kokkinos, Sammy Mohamed, and Andrea Vedaldi. Describing textures in the wild. In IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 3606–3613, 2014. 7

    work page 2014

  8. [8]

    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 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 248–255, 2009. 6

    work page 2009

Show all 54 references
  1. [9]

    Everybody prune now: Structured pruning of llms with only forward passes

    Lucio Dery, Steven Kolawole, Jean-Franc ¸ois Kagy, Virginia Smith, Graham Neubig, and Ameet Talwalkar. Everybody prune now: Structured pruning of llms with only forward passes. arXiv preprint arXiv:2402.05406, 2024. 4

    2024

  2. [10]

    Learning gener- ative visual models from few training examples: An incre- mental bayesian approach tested on 101 object categories

    Li Fei-Fei, Rob Fergus, and Pietro Perona. Learning gener- ative visual models from few training examples: An incre- mental bayesian approach tested on 101 object categories. In IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (WCVPR), pages 178–178, 2004. 7

    2004

  3. [11]

    Rosa: Random sub- space adaptation for efficient fine-tuning

    Marawan Gamal Abdel Hameed, Aristides Milios, Siva Reddy, and Guillaume Rabusseau. Rosa: Random sub- space adaptation for efficient fine-tuning. arXiv preprint arXiv:2407.07802, 2024

    2024 arXiv

  4. [12]

    SLTrain: a sparse plus low-rank approach for parameter and memory efficient pretraining

    Andi Han, Jiaxiang Li, Wei Huang, Mingyi Hong, Akiko Takeda, Pratik Jawanpuria, and Bamdev Mishra. SLTrain: a sparse plus low-rank approach for parameter and memory efficient pretraining. In Advances in Neural Information Pro- cessing Systems (NeurIPS), 2024. 4

    2024

  5. [13]

    Towards a unified view of parameter-efficient transfer learning

    Junxian He, Chunting Zhou, Xuezhe Ma, Taylor Berg- Kirkpatrick, and Graham Neubig. Towards a unified view of parameter-efficient transfer learning. In International Con- ference on Learning Representations (ICLR), 2021. 3

    2021

  6. [14]

    Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification

    Patrick Helber, Benjamin Bischke, Andreas Dengel, and Damian Borth. Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 12(7):2217–2226, 2019. 7

    2019

  7. [15]

    Parameter-efficient transfer learning for nlp

    Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin De Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. Parameter-efficient transfer learning for nlp. In International Conference on Machine Learning (ICML), pages 2790–2799, 2019. 1

    2019

  8. [16]

    LoRA: Low-rank adaptation of large language models

    Edward J Hu, yelong shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-rank adaptation of large language models. InIn- ternational Conference on Learning Representations (ICLR),

  9. [17]

    Do sparse brain activ- ity patterns underlie human cognition? NeuroImage, 263: 119633, 2022

    Iiro P J ¨a¨askel¨ainen, Enrico Glerean, Vasily Klucharev, Anna Shestakova, and Jyrki Ahveninen. Do sparse brain activ- ity patterns underlie human cognition? NeuroImage, 263: 119633, 2022. 1

    2022

  10. [18]

    Vi- sual prompt tuning

    Menglin Jia, Luming Tang, Bor-Chun Chen, Claire Cardie, Serge Belongie, Bharath Hariharan, and Ser-Nam Lim. Vi- sual prompt tuning. In European Conference on Computer Vision (ECCV), pages 709–727, 2022. 1

    2022

  11. [19]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In International Conference on Learning Representations (ICLR), 2015. 3

    2015

  12. [20]

    VeRA: Vector-based random matrix adaptation

    Dawid Jan Kopiczko, Tijmen Blankevoort, and Yuki M Asano. VeRA: Vector-based random matrix adaptation. In International Conference on Learning Representations (ICLR), 2024. 1, 4

    2024

  13. [21]

    3d object representations for fine-grained categorization

    Jonathan Krause, Michael Stark, Jia Deng, and Li Fei-Fei. 3d object representations for fine-grained categorization. In IEEE International Conference on Computer Vision Work- shops (WICCV), pages 554–561, 2013. 7

    2013

  14. [22]

    Block pruning for faster transformers

    Franc ¸ois Lagunas, Ella Charlaix, Victor Sanh, and Alexander Rush. Block pruning for faster transformers. In Empirical Methods in Natural Language Processing (EMNLP) , pages 10619–10629, 2021. 4

    2021

  15. [23]

    Relora: High-rank training through low-rank updates

    Vladislav Lialin, Sherin Muckatira, Namrata Shivagunde, and Anna Rumshisky. Relora: High-rank training through low-rank updates. In International Conference on Learning Representations (ICLR), 2023. 3, 4, 6

    2023

  16. [24]

    Nora: Nested low-rank adaptation for efficient fine-tuning large models

    Cheng Lin, Lujun Li, Dezhi Li, Jie Zou, Wei Xue, and Yike Guo. Nora: Nested low-rank adaptation for efficient fine-tuning large models. arXiv preprint arXiv:2408.10280,

    arXiv

  17. [25]

    DoRA: Weight-decomposed low-rank adaptation

    Shih-Yang Liu, Chien-Yi Wang, Hongxu Yin, Pavlo Molchanov, Yu-Chiang Frank Wang, Kwang-Ting Cheng, and Min-Hung Chen. DoRA: Weight-decomposed low-rank adaptation. In International Conference on Machine Learn- ing (ICML), pages 32100–32121, 2024. 3, 4

    2024

  18. [26]

    Segclip: Patch aggregation with learnable centers for open-vocabulary semantic segmentation

    Huaishao Luo, Junwei Bao, Youzheng Wu, Xiaodong He, and Tianrui Li. Segclip: Patch aggregation with learnable centers for open-vocabulary semantic segmentation. In In- ternational Conference on Machine Learning (ICML), pages 23033–23044, 2023. 1

    2023

  19. [27]

    Llm-pruner: On the structural pruning of large language models

    Xinyin Ma, Gongfan Fang, and Xinchao Wang. Llm-pruner: On the structural pruning of large language models. In Ad- vances in Neural Information Processing Systems (NeurIPS),

  20. [28]

    Fine-grained visual classi- fication of aircraft

    Subhransu Maji, Esa Rahtu, Juho Kannala, Matthew B Blaschko, and Andrea Vedaldi. Fine-grained visual classi- fication of aircraft. In British Machine Vision Conference (BMVC), 2013. 7

    2013

  21. [29]

    PiSSA: Principal singular values and singular vectors adaptation of large language models

    Fanxu Meng, Zhaohui Wang, and Muhan Zhang. PiSSA: Principal singular values and singular vectors adaptation of large language models. In Advances in Neural Information Processing Systems (NeurIPS), 2024. 3, 4

    2024

  22. [30]

    Automated flower classification over a large number of classes

    Maria-Elena Nilsback and Andrew Zisserman. Automated flower classification over a large number of classes. InIndian Conference on Computer Vision, Graphics & Image Process- ing (ICVGIP), pages 722–729, 2008. 7

    2008

  23. [31]

    Lisa: Layerwise importance sampling for memory-efficient large language model fine- tuning

    Rui Pan, Xiang Liu, Shizhe Diao, Renjie Pi, Jipeng Zhang, Chi Han, and Tong Zhang. Lisa: Layerwise importance sampling for memory-efficient large language model fine- tuning. In Advances in Neural Information Processing Sys- tems (NeurIPS), 2024

    2024

  24. [32]

    Cats and dogs

    Omkar M Parkhi, Andrea Vedaldi, Andrew Zisserman, and CV Jawahar. Cats and dogs. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3498–3505, 2012. 7

    2012

  25. [33]

    Learn- ing transferable visual models from natural language super- vision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learn- ing transferable visual models from natural language super- vision. In International Conference on Machine Learning...

    2021

  26. [34]

    A closer look at the few-shot adaptation of large vision-language models

    Julio Silva-Rodriguez, Sina Hajimiri, Ismail Ben Ayed, and Jose Dolz. A closer look at the few-shot adaptation of large vision-language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 23681–23690, 2024. 3

    2024

  27. [35]

    Ucf101: A dataset of 101 human actions classes from videos in the wild

    Khurram Soomro, Amir Roshan Zamir, and Mubarak Shah. Ucf101: A dataset of 101 human actions classes from videos in the wild. arXiv preprint arXiv:1212.0402, 2012. 7

    2012 arXiv

  28. [36]

    A simple and effective pruning approach for large language models

    Mingjie Sun, Zhuang Liu, Anna Bair, and J Zico Kolter. A simple and effective pruning approach for large language models. In International Conference on Learning Represen- tations (ICLR), 2024. 4

    2024

  29. [37]

    A comparative study between full-parameter and lora-based fine-tuning on chinese instruction data for instruction follow- ing large language model

    Xianghui Sun, Yunjie Ji, Baochang Ma, and Xiangang Li. A comparative study between full-parameter and lora-based fine-tuning on chinese instruction data for instruction follow- ing large language model. arXiv preprint arXiv:2304.08109,

    arXiv

  30. [38]

    Training neu- ral networks with fixed sparse masks

    Yi-Lin Sung, Varun Nair, and Colin A Raffel. Training neu- ral networks with fixed sparse masks. In Advances in Neural Information Processing Systems (NeurIPS) , pages 24193– 24205, 2021. 4

    2021

  31. [39]

    Cora: Adapting clip for open-vocabulary detection with region prompting and anchor pre-matching

    Xiaoshi Wu, Feng Zhu, Rui Zhao, and Hongsheng Li. Cora: Adapting clip for open-vocabulary detection with region prompting and anchor pre-matching. In IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) , pages 7031–7040, 2023. 1

    2023

  32. [40]

    A simple model for behav- ioral time scale synaptic plasticity (btsp) provides content ad- dressable memory with binary synapses and one-shot learn- ing

    Yujie Wu and Wolfgang Maass. A simple model for behav- ioral time scale synaptic plasticity (btsp) provides content ad- dressable memory with binary synapses and one-shot learn- ing. Nature communications, 16(1):342, 2025

    2025

  33. [41]

    Struc- tured pruning learns compact and accurate models

    Mengzhou Xia, Zexuan Zhong, and Danqi Chen. Struc- tured pruning learns compact and accurate models. In An- nual Meeting of the Association for Computational Linguis- tics (ACL), 2022. 4

    2022

  34. [42]

    Sun database: Large-scale scene recognition from abbey to zoo

    Jianxiong Xiao, James Hays, Krista A Ehinger, Aude Oliva, and Antonio Torralba. Sun database: Large-scale scene recognition from abbey to zoo. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3485–3492, 2010. 6

    2010

  35. [43]

    Raise a child in large language model: Towards effective and generalizable fine-tuning

    Runxin Xu, Fuli Luo, Zhiyuan Zhang, Chuanqi Tan, Baobao Chang, Songfang Huang, and Fei Huang. Raise a child in large language model: Towards effective and generalizable fine-tuning. In Empirical Methods in Natural Language Pro- cessing (EMNLP), pages 9514–9528, 2021. 4

    2021

  36. [44]

    CorDA: Context-oriented decomposition adaptation of large language models for task-aware parameter-efficient fine- tuning

    Yibo Yang, Xiaojie Li, Zhongzhu Zhou, Shuaiwen Leon Song, Jianlong Wu, Liqiang Nie, and Bernard Ghanem. CorDA: Context-oriented decomposition adaptation of large language models for task-aware parameter-efficient fine- tuning. In Advances in Neural Information Processing Sys- ...

    2024

  37. [45]

    Task residual for tuning vision-language models

    Tao Yu, Zhihe Lu, Xin Jin, Zhibo Chen, and Xinchao Wang. Task residual for tuning vision-language models. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10899–10909, 2023. 3

    2023

  38. [46]

    Low-rank few-shot adaptation of vision-language models

    Maxime Zanella and Ismail Ben Ayed. Low-rank few-shot adaptation of vision-language models. In IEEE/CVF Con- ference on Computer Vision and Pattern Recognition Work- shops (WCVPR), pages 1593–1603, 2024. 3, 6, 7

    2024

  39. [47]

    Gradient- based parameter selection for efficient fine-tuning

    Zhi Zhang, Qizhe Zhang, Zijun Gao, Renrui Zhang, Ekate- rina Shutova, Shiji Zhou, and Shanghang Zhang. Gradient- based parameter selection for efficient fine-tuning. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 28566–28577, 2024. 4

    2024

  40. [48]

    Galore: Memory- efficient llm training by gradient low-rank projection

    Jiawei Zhao, Zhenyu Zhang, Beidi Chen, Zhangyang Wang, Anima Anandkumar, and Yuandong Tian. Galore: Memory- efficient llm training by gradient low-rank projection. In In- ternational Conference on Machine Learning (ICML), 2024. 3, 4, 6

    2024

  41. [49]

    Conditional prompt learning for vision-language models

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Zi- wei Liu. Conditional prompt learning for vision-language models. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 16816–16825, 2022. 1

    2022

  42. [50]

    Learning to prompt for vision-language models

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. Learning to prompt for vision-language models. Inter- national Journal of Computer Vision (IJCV) , 130(9):2337– 2348, 2022. 1, 6

    2022

  43. [51]

    Not all features mat- ter: Enhancing few-shot clip with adaptive prior refinement

    Xiangyang Zhu, Renrui Zhang, Bowei He, Aojun Zhou, Dong Wang, Bin Zhao, and Peng Gao. Not all features mat- ter: Enhancing few-shot clip with adaptive prior refinement. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 2605–2615, 2023. 1

    2023

  44. [52]

    The effects of regularization and data augmentation are class de- pendent

    Randall Balestriero, Leon Bottou, and Yann LeCun. The effects of regularization and data augmentation are class de- pendent. In Advances in Neural Information Processing Sys- tems (NeurIPS), volume 35, pages 37878–37891, 2022. 1

    2022

  45. [53]

    CLIP-Adapter: Better vision-language models with feature adapters

    Peng Gao, Shijie Geng, Renrui Zhang, Teli Ma, Rongyao Fang, Yongfeng Zhang, Hongsheng Li, and Yu Qiao. CLIP-Adapter: Better vision-language models with feature adapters. International Journal of Computer Vision (IJCV), 132(2):581–595, 2024. 1

    2024

  46. [54]

    Sparsity Outperforms Low-Rank Projections in Few-Shot Adaptation

    Dawid Jan Kopiczko, Tijmen Blankevoort, and Yuki M Asano. VeRA: Vector-based random matrix adaptation. arXiv preprint arXiv:2310.11454, 2023. 3 Supplementary Material for: “Sparsity Outperforms Low-Rank Projections in Few-Shot Adaptation” A. Few-shot Adaptation of VLMs VLMs ar...

    2023 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.