Pith. sign in

REVIEW 3 major objections 5 minor 48 references

Prune Once: Retraining-Free Task-Agnostic Pruning for Vision-Language Models

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read PORTA prunes a vision-language model once using activation-variance importance and adaptive layer-wise sparsity, so the same compressed model transfers to classification, retrieval, and VQA tasks without retraining.

desk verdict A solid, well-ablated VLM pruning paper whose 'prune once' claim is undercut by an unreported per-sparsity alpha hyperparameter that appears to track CIFAR-10 tuned values. read the letter →

arxiv 2608.06901 v1 pith:LBNY5RKH submitted 2026-08-07 cs.CV cs.LG

Minseok Kang , Hyunwoo Kim , Chanyoung Kim , Minwoo Kim , Jaekoo Lee , Dahuin Jung This is my paper · ORCID
classification cs.CVcs.LG
keywords vision-languagemodelspost-trainingpruningtask-agnosticretraining-freecompressionactivationvariancelayer-wisesparsityallocationzero-shotevaluationmodel
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

PORTA is a pruning framework for vision-language models that aims to make the pruned model usable across many downstream tasks immediately after a single pruning pass, with no retraining or re-pruning. The paper argues that the usual magnitude-based importance measures, borrowed from large language models, are biased by the different activation scales of vision and text branches, and that the variance of activations along the token/patch dimension is a stable, modality-agnostic signal of a feature's representational utility. PORTA multiplies this per-feature variance by weight magnitude to score weights, and additionally assigns each layer a pruning ratio from the variability of its output features. Across CLIP, BLIP, Qwen2-VL, OpenFlamingo, and SDXL text encoders, the paper reports competitive zero-shot results on classification, retrieval, VQA, and captioning, with the clearest gains over Wanda and SparseGPT appearing at high sparsity.

What carries the argument

The object carrying the argument is the per-feature activation variance vector $v_j = \mathrm{Var}_{b,t}(X_{b,t,j})$ for input features of each linear layer, combined with the weight matrix as $S_{i,j} = v_j |W_{i,j}|$ for weight importance. The layer-wise allocation uses the normalized input covariance $\hat\Sigma_X = \Sigma_X / \mathrm{Tr}(\Sigma_X)$ and the score $s_{\mathrm{layer}} = \frac{1}{D_{\mathrm{out}}} \sum_o w_o \hat\Sigma_X w_o^\top$, the average output variance per channel, which is also written as a variance-weighted average of column norms $\|W_{:,j}\|_2^2$. Algorithm 1 converts these scores into pruning ratios $r_i = C (1 - p_i)^\alpha$ with a single strength parameter $\alpha$, after weighting by each layer's parameter count. The core work is to replace magnitude-based outlier reasoning with variance-based reasoning at both the weight level and the layer level.

What would settle it

Run PORTA on a held-out VLM with the $\alpha$ values used in the paper and a fixed generic calibration set, then measure zero-shot performance across several unseen tasks; if the same $\alpha$ does not reproduce the reported rankings, or if an alternative calibration set shifts performance by far more than 1%, the prune-once claim would be falsified. A sharper test is to find one VLM layer whose high-variance features are redundant and whose low-variance features are task-critical; pruning by variance should fail dramatically there.

Watch

Extended reading notes

Core claim

The central claim is that activation variance, not activation magnitude, is the right importance signal for pruning vision-language models. In CLIP-style contrastive models the vision and text encoders have very different activation scales and outlier patterns, so a magnitude-based score such as Wanda's prunes the two branches unevenly; variance, measured per input feature across tokens or patches, is shown to have similar dispersion in both branches and to track representation utility. Removing high-variance features from CLIP destroys retrieval accuracy, while removing low-variance features leaves it nearly intact. On top of that, PORTA computes a layer score from the normalized input covariance and the squared norms of output weight vectors, and uses it to allocate smaller pruning ratios to layers whose outputs use more of the feature space. The paper claims that together these two mechanisms let a VLM be pruned once with generic image-text calibration data and then applied zero-shot to heterogeneous tasks, outperforming Wanda, SparseGPT, ECoFLaP, and Multiflow at high sparsity.

Load-bearing premise

The load-bearing premise is that one strength parameter $\alpha$, combined with the layer output-variance scores, transfers across tasks, modalities, and sparsity levels without per-task recalibration; the paper's own sensitivity table shows the best $\alpha$ drifts as target sparsity changes, so if $\alpha$ must be retuned per deployment the 'prune once' advantage narrows.

Editorial extensions

If this is right

  • One pruned VLM can be stored once and served to classification, retrieval, VQA, and captioning workloads without re-pruning or fine-tuning, removing the per-task compression pipeline.
  • High-sparsity deployment becomes practical: at 65% sparsity on CLIP, PORTA reports a 12.6% mean improvement over SparseGPT and 21.5% over Wanda across the eight reported metrics.
  • Calibration becomes cheap and robust: PORTA reports roughly 1% variation across Flickr30K, MSCOCO, and Visual Genome calibration at 60% sparsity, and near-flat performance from 64 to 2048 calibration samples.
  • The same variance-based importance transfers to generative VLMs: Qwen2-VL on ScienceQA reaches 54.30 average accuracy at 50% sparsity, the best among pruned baselines.
  • Pruning overhead drops: PORTA prunes CLIP in about 195 seconds, 3.96x faster than SparseGPT and 2.06x faster than ECoFLaP.

Reading between the lines

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

  • If the variance-based signal generalizes beyond the tested families, the same scoring could be applied to other multimodal architectures such as audio-text, video-text, or interleaved generation models without redesigning the calibration step.
  • The reuse of variance statistics for both weight scoring and ratio allocation suggests the method could fold into quantization-aware compression pipelines, since both need activation statistics and the paper already shows a 2:4 structured-pruning extension.
  • A testable extension is to check whether the same $\alpha$ value transfers across architectures; the paper's sensitivity study shows the preferred $\alpha$ drifts with target sparsity, so an automatic rule for $\alpha$ from target sparsity and model depth would make the prune-once workflow fully parameter-free.
  • The axis-removal result implies variance could serve as a diagnostic for other structure-discovery tasks, such as identifying which feature dimensions a modality encoder actually uses.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes PORTA, a retraining-free, task-agnostic pruning framework for vision-language models. The importance score combines per-feature activation variance with weight magnitude (Eq. 2), and layer-wise sparsity is allocated through a score based on normalized output covariance (Eqs. 10-12) and a strength parameter alpha in Algorithm 1. Experiments cover CLIP, BLIP, and Qwen2-VL on classification, retrieval, and VQA, with comparisons to Wanda, SparseGPT, ECoFLaP, and Multiflow. The paper reports strong high-sparsity results and includes ablations on calibration source, sample count, the zero-mean assumption, and the diagonal covariance approximation.

Significance. If the central claim holds, PORTA is a practically useful step: a single pruned VLM that can be deployed across heterogeneous tasks without re-pruning or retraining, using only generic calibration data. The variance-based motivation is well supported by the axis-removal experiment in Table 2, the cross-combination study in Table 6, and the calibration-robustness experiments in Fig. 4 and Appendix E. The paper also provides a clean derivation of the layer score and explicitly ablates its two modeling assumptions. Code release and the multi-architecture evaluation are additional strengths. The main weakness is that the hyperparameter alpha, which controls the layer-wise ratio allocation, is tuned per sparsity level on CIFAR-10 in Appendix D but is not reported for the main results, leaving the task-agnostic 'prune once' claim incompletely audited.

major comments (3)
  1. [Sec. 4.3 / Appendix D / Table A4] The main tables do not report the value of alpha used for each sparsity level, and Appendix D shows that the CIFAR-10 numbers in Table 3 track per-sparsity best-alpha values: at 60% sparsity the reported CIFAR-10 accuracy of 95.06 is the best entry for alpha=10 in Table A4, and at 65% sparsity the reported 90.49 is the best entry for alpha=30. This is strong evidence that the layer-wise allocation strength was selected on CIFAR-10 labels, which conflicts with the paper's stated protocol of pruning without downstream task-specific samples. Please report alpha for every main-table result, provide a label-free rule or closed-form schedule for setting alpha, and clarify whether the retrieval and VQA results in Tables 3-5 were obtained with the same alpha as the classification column or with per-task selection.
  2. [Algorithm 1 / Sec. 3.3] Even if alpha is reported, the 'prune once' claim requires that a single alpha transfers across tasks and modalities. The current evidence is limited to CIFAR-10 classification on CLIP (Table A4) and one visualization at 60% sparsity (Fig. 3). Since Table A4 itself shows that the preferred alpha shifts with target sparsity (from roughly 5-10 at 50-55% to 30 at 65%), the paper should either give a principled, label-free schedule or demonstrate with a fixed alpha, held out from downstream validation, that the results in Tables 3-5 remain competitive across all task families. Without this, the central task-agnostic claim is not auditable.
  3. [Sec. 4.1 / 'zero-shot' protocol] The paper describes all evaluations as 'fully zero-shot,' but hyperparameter selection on CIFAR-10 in Appendix D means the CIFAR-10 column at least is a selected result rather than an independent zero-shot evaluation. The authors should state explicitly that alpha tuning was performed on a downstream dataset, or remove this apparent leak by reporting alpha values and a label-free selection mechanism.
minor comments (5)
  1. [Sec. 3.2 / Sec. 3.3] The notation for input activations is inconsistent: Sec. 3.2 uses X in R^{B x T x D_in} while Sec. 3.3 uses X in R^{T x D_in}; please unify the notation and clarify whether the variance in Eq. (1) is computed over both batch and token dimensions or over tokens only.
  2. [Algorithm 1] The returned ratios r_i are not explicitly constrained to [0,1]; if C u_i exceeds 1 for some layer, the allocation would be invalid. Please specify clipping or normalization and confirm that the final global sparsity exactly matches the target S.
  3. [Fig. 3 caption] The caption states alpha=10 for the 60% sparsity visualization, but Table 3 does not report which alpha was used for the corresponding 60% results; please align the visualization with the evaluated configuration.
  4. [Abstract / Sec. 1] The abstract and contribution statements describe the importance criterion as 'activation variation,' but Eq. (2) is Var * |W|; the weight-magnitude factor should be acknowledged in the summary of the method.
  5. [Table 1] Table 1 reports max-min dispersion for layers labeled L1, L20, L30 but does not state which model and which layers are used; please add the experimental details so the reader can interpret the modality-sensitivity claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: PORTA's importance and sparsity scores are defined from calibration activation statistics and weights, not from downstream predictions.

full rationale

PORTA's derivation chain is self-contained. Equation 1 defines v_j from calibration activation variance, Equation 2 multiplies it by pretrained weight magnitudes to form the importance score, and no downstream accuracy enters these definitions. The layer-wise ratio score s_layer (Eqs. 10-12) is an analytic variance-weighted norm of the weight matrix, and Algorithm 1 converts these scores into pruning ratios using a global sparsity target and the strength alpha. The only tunable external to the derivation is alpha, which is a validation-selected hyperparameter rather than a fitted parameter of the importance estimate. Appendix D sweeps alpha on CIFAR-10, and Table 3 does not state which alpha was used; this is a reporting and audit gap that weakens the 'no tuning' narrative, but it does not make the CIFAR-10 column equal to the method's input by construction, and no equation reduces output performance to the calibration statistics. There are no load-bearing self-citations in the derivation, and the calibration-robustness and cross-combination ablations compare against baselines under fixed scores. The central claim therefore has independent content; no significant circularity.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central method depends on one tuned hyperparameter (alpha), two stated statistical approximations, and the empirical premise that variance indicates representation utility. No new entities are postulated.

free parameters (1)
  • alpha (pruning ratio strength) = varies by sparsity; e.g., 5 at 50%, 30 at 65% (Appendix D Table A4)
    Controls the spread of per-layer pruning ratios in Algorithm 1. It is tuned on validation performance per sparsity level, and the values used in the main experiments are not listed.
assumptions (4)
  • domain assumption Input tokens are drawn from a zero-mean distribution when deriving output variance (Eq. 5).
    Invoked in Sec. 3.3 before Eq. 5; ablated in Sec. 4.6 (Table 7) with at most 0.1% change.
  • domain assumption The input covariance can be approximated by its diagonal entries v_j (Eq. 12).
    Used to make layer score computation cheap and to reuse importance variance statistics; ablated in Table 7 with at most 1% change.
  • domain assumption Feature variance along tokens measures representation utility, so low-variance features can be pruned safely.
    Empirical premise supported by axis-removal analysis (Table 2) and module-wise pruning (Table A8), but not derived from first principles.
  • domain assumption Layers with higher output-variance score s_layer deserve lower pruning ratios.
    The ratio mechanism in Algorithm 1 assigns lower sparsity to high s_layer layers; this is validated indirectly by the ablation in Table 6, not proven.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Prune Once: Retraining-Free Task-Agnostic Pruning for Vision-Language Models." pith.science (2026). https://pith.science/paper/LBNY5RKH

@misc{pith2026260806901,
  author       = {Pith},
  title        = {Pith review of: Prune Once: Retraining-Free Task-Agnostic Pruning for Vision-Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LBNY5RKH}},
  note         = {Machine review of arXiv:2608.06901}
}
read the original abstract

Vision-language models (VLMs) have achieved remarkable generalization across diverse multimodal tasks through large-scale pre-training, yet their rapidly increasing computational and memory requirements pose significant challenges for deployment in constrained environments. Existing pruning strategies often depend on task-specific criteria or LLM-oriented importance measures, making them unsuitable for task-agnostic pruning, where no task-specific samples are available at pruning time and the pruned model remains broadly applicable. We introduce a retraining-free VLM pruning framework called PORTA that derives a task- and modality-agnostic importance formulation based on activation variation, estimated from generic calibration data, which reliably captures feature-level representation utility across modalities. PORTA further incorporates an adaptive sparsity allocation mechanism that assigns layer-wise pruning ratios based on output feature variability, avoiding the limitations of uniform sparsity and reducing performance degradation at high compression levels. Extensive experiments across VLM architectures, such as CLIP, BLIP, and Qwen2-VL, demonstrate that PORTA achieves competitive downstream performance under high sparsity without requiring any retraining, supporting efficient VLM compression. Code is available at https://github.com/cau-hai-lab/PORTA.git.

Figures

Figures reproduced from arXiv: 2608.06901 by the authors.

Figure 1
Figure 1. Illustration of PORTA’s task- and modality-agnostic pruning paradigm: us￾ing only general data, a VLM is pruned once—without any re-pruning or retrain￾ing—and directly applied to diverse down￾stream tasks. Comparison across three tasks—classification (accuracy), retrieval (TR@5), and VQA (Avg)—under high sparsity. PORTA delivers uniformly strong results across all settings, outperforming Wanda and SparseGPT. Task-ag… view at source ↗
Figure 2
Figure 2. Modality-specific activation characteristics and PORTA overview. The left com￾ponent, (a) Text and vision branches exhibit clearly different activation patterns across layers. (b) PORTA overview: Var·|W| for weight importance and a pruning-ratio score for layer-wise sparsity allocation. To address this bias, we propose PORTA (Prune-Once: Retraining-free Task￾Agnostic pruning), which does not simply replace existing … view at source ↗
Figure 3
Figure 3. [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Effect of calibration data type, pruning method, and sample size on zero-shot performance at 60% sparsity with CLIP. strates that deeper layers exhibit increasing representational complexity and encode progressively more diverse concepts. 4.4 Calibration Data and Sampl…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 36 canonical work pages

  1. [1]

    In: Pro- ceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Agrawal, H., Desai, K., Wang, Y., Chen, X., Jain, R., Johnson, M., Batra, D., Parikh, D., Lee, S., Anderson, P.: noCaps: novel object captioning at scale. In: Pro- ceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 8948–8957 (2019)

    work page 2019

  2. [2]

    In: International Conference on Learning Representations (ICLR) (2022)

    Alizadeh, M., Tailor, S.A., Zintgraf, L.M., van Amersfoort, J., Farquhar, S., Lane, N.D., Gal, Y.: Prospect pruning: Finding trainable weights at initialization using meta-gradients. In: International Conference on Learning Representations (ICLR) (2022)

    work page 2022

  3. [3]

    Proceedings of the IEEE international conference on computer vision

    Antol, S., Agrawal, A., Lu, J., Mitchell, M., Batra, D., Zitnick, C.L., Parikh, D.: Vqa: Visual question answering. Proceedings of the IEEE international conference on computer vision. (2015)

    work page 2015

  4. [4]

    arXiv preprint arXiv:2308.01390 (2023)

    Awadalla, A., Gao, I., Gardner, J., Hessel, J., Hanafy, Y., Zhu, W., Marathe, K., Bitton, Y., Gadre, S., Sagawa, S., Jitsev, J., Kornblith, S., Koh, P.W., Ilharco, G., Wortsman, M., Schmidt, L.: OpenFlamingo: An open-source framework for train- ing large autoregressive vision-language models. arXiv preprint arXiv:2308.01390 (2023)

    arXiv 2023

  5. [5]

    Journal of Artificial Intel- ligence Research (JAIR)55, 409–442 (2016)

    Bernardi, R., Cakici, R., Elliott, D., Erdem, A., Erdem, E., Ikizler-Cinbis, N., Keller, F., Muscat, A., Plank, B.: Automatic description generation from images: A survey of models, datasets, and evaluation measures. Journal of Artificial Intel- ligence Research (JAIR)55, 409–442 (2016)

    work page 2016

  6. [6]

    Blalock, D., Ortiz, J.J.G., Frankle, J., Guttag, J.: What is the state of neural network pruning? In: Proceedings of Machine Learning and Systems (MLSys) (2020)

    work page 2020

  7. [7]

    arXiv preprint arXiv:2203.14713 (2022)

    Cao, M., Li, S., Li, J., Nie, L., Zhang, M.: Image-text retrieval: A survey on recent research and development. arXiv preprint arXiv:2203.14713 (2022)

    arXiv 2022

  8. [8]

    In: International Conference on Machine Learning (ICML) (2025)

    Chen, Y., Cheng, B., Han, J., Zhang, Y.y., Li, Y., Zhang, S.: Dlp: Dynamic layer- wise pruning in large language models. In: International Conference on Machine Learning (ICML) (2025)

    work page 2025

Show all 48 references
  1. [9]

    In: CVPR (2009)

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: A large-scale hierarchical image database. In: CVPR (2009)

    2009

  2. [10]

    In: Explainable Artificial Intelligence, pp

    Dorszewski, T., Tětková, L., Jenssen, R., Hansen, L.K., Wickstrøm, K.K.: From colors to classes: Emergence of concepts in vision transformers. In: Explainable Artificial Intelligence, pp. 28–47. Springer (2025)

    2025

  3. [11]

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

    Farina, M., Mancini, M., Cunegatti, E., Liu, G., Iacca, G., Ricci, E.: Multiflow: Shifting towards task-agnostic vision-language pruning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 16185– 16195 (2024)

    2024

  4. [12]

    In: ICML (2023)

    Frantar, E., Alistarh, D.: SparseGPT: Massive Language Models Can Be Accu- rately Pruned in One-Shot. In: ICML (2023)

    2023

  5. [13]

    arXiv preprint arXiv:2310.05916 (2023)

    Gandelsman, Y., Efros, A.A., Steinhardt, J.: Interpreting clip’s image representa- tion via text-based decomposition. arXiv preprint arXiv:2310.05916 (2023)

    2023 arXiv

  6. [14]

    NeurIPS (2015)

    Han, S., Mao, H., Dally, W.J.: Deep compression: Compressing deep neural net- works with pruning, trained quantization and huffman coding. NeurIPS (2015)

    2015

  7. [15]

    Advances in neural information processing systems28 (2015) PORTA 17

    Han, S., Pool, J., Tran, J., Dally, W.: Learning both weights and connections for efficient neural network. Advances in neural information processing systems28 (2015) PORTA 17

    2015

  8. [16]

    In: 4th Lifelong Machine Learning Workshop at ICML (2020)

    He, X., Sygnowski, J., Galashov, A., Rusu, A.A., Teh, Y.W., Pascanu, R.: Task agnostic continual learning via meta learning. In: 4th Lifelong Machine Learning Workshop at ICML (2020)

    2020

  9. [17]

    In: Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Jamal, M.A., Qi, G.J.: Task agnostic meta-learning for few-shot learning. In: Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 11719–11727 (2019)

    2019

  10. [18]

    In: International Conference on Learning Representations (ICLR) (2021)

    de Jorge, P., Sanyal, A., Behl, H.S., Torr, P.H.S., Rogez, G., Dokania, P.K.: Pro- gressive skeletonization: Trimming more fat from a network at initialization. In: International Conference on Learning Representations (ICLR) (2021)

    2021

  11. [19]

    International Journal of Computer Vision123(1), 32–73 (2017)

    Krishna, R., Zhu, Y., Groth, O., Johnson, J., Hata, K., Kravitz, J., Chen, S., Kalantidis, Y., Li, L.J., Shamma, D.A., Bernstein, M.S., Fei-Fei, L.: Visual genome: Connecting language and vision using crowdsourced dense image anno- tations. International Journal of Computer Vi...

    2017

  12. [20]

    Krizhevsky, A.: Learning multiple layers of features from tiny images. Tech. rep., University of Toronto (2009)

    2009

  13. [21]

    In: ICLR (2021)

    Lee, J., Park, S., Mo, S., Ahn, S., Shin, J.: Layer-adaptive Sparsity for the Magnitude-based Pruning. In: ICLR (2021)

    2021

  14. [22]

    In: ICLR (2019)

    Lee, N., Ajanthan, T., Torr, P.: SNIP: Single-Shot Network Pruning based on connection sensitivity. In: ICLR (2019)

    2019

  15. [23]

    In: ICML (2023)

    Li, J., Li, D., Savarese, S., Hoi, S.: BLIP-2: Bootstrapping language-image pre- training with frozen image encoders and large language models. In: ICML (2023)

    2023

  16. [24]

    In: ICML (2022)

    Li, J., Li, D., Xiong, C., Hoi, S.: BLIP: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In: ICML (2022)

    2022

  17. [25]

    In: Text Summarization Branches Out

    Lin, C.Y.: ROUGE: A package for automatic evaluation of summaries. In: Text Summarization Branches Out. pp. 74–81. Association for Computational Linguis- tics (2004)

    2004

  18. [26]

    In: ECCV

    Lin, T.Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., Zitnick, C.L.: Microsoft COCO: Common Objects in Context. In: ECCV. Springer (2014)

    2014

  19. [27]

    Liu,H.,Li,C.,Wu,Q.,Lee,Y.J.:Visualinstructiontuning.In:AdvancesinNeural Information Processing Systems (2023)

    2023

  20. [28]

    In: NeurIPS (2022)

    Lu, P., Mishra, S., Xia, T., Qiu, L., Chang, K.W., Zhu, S.C., Tafjord, O., Clark, P., Kalyan, A.: Learn to explain: Multimodal reasoning via thought chains for science question answering. In: NeurIPS (2022)

    2022

  21. [29]

    In: Indian Conference on Computer Vision, Graphics & Image Processing (2008)

    Nilsback, M.E., Zisserman, A.: Automated flower classification over a large num- ber of classes. In: Indian Conference on Computer Vision, Graphics & Image Processing (2008)

    2008

  22. [30]

    In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics (2002)

    Papineni, K., Roukos, S., Ward, T., Zhu, W.J.: BLEU: a Method for Automatic Evaluation of Machine Translation. In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics (2002)

    2002

  23. [31]

    In: Proceedings of the IEEE international con- ference on computer vision

    Plummer, B.A., Wang, L., Cervantes, C.M., Caicedo, J.C., Hockenmaier, J., Lazebnik, S.: Flickr30k entities: Collecting region-to-phrase correspondences for richer image-to-sentence models. In: Proceedings of the IEEE international con- ference on computer vision. pp. 2641–2649 (2015)

    2015

  24. [32]

    arXiv preprint arXiv:2307.01952 (2023)

    Podell, D., English, Z., Lacey, K., Blattmann, A., Dockhorn, T., Müller, J., Penna, J., Rombach, R.: Sdxl: Improving latent diffusion models for high-resolution image synthesis. arXiv preprint arXiv:2307.01952 (2023)

    2023 arXiv

  25. [33]

    In: ICML (2021) 18 Kang et al

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al.: Learning transferable visual models from natural language supervision. In: ICML (2021) 18 Kang et al

    2021

  26. [34]

    In: ICML (2023)

    Shi, D., Tao, C., Jin, Y., Yang, Z., Yuan, C., Wang, J.: UPop: Unified and pro- gressive pruning for compressing vision-language transformers. In: ICML (2023)

    2023

  27. [35]

    ICLR (2024)

    Sun, M., Liu, Z., Bair, A., Kolter, J.Z.: A Simple and Effective Pruning Approach for Large Language Models. ICLR (2024)

    2024

  28. [36]

    In: International Conference on Learning Repre- sentations (ICLR) (2024)

    Sung, Y.L., Yoon, J., Bansal, M.: Ecoflap: Efficient coarse-to-fine layer-wise prun- ing for vision-language models. In: International Conference on Learning Repre- sentations (ICLR) (2024)

    2024

  29. [37]

    In: Advances in Neural In- formation Processing Systems (NeurIPS) (2020)

    Tanaka, H., Kunin, D., Yamins, D.L., Ganguli, S.: Pruning neural networks with- out any data by iteratively conserving synaptic flow. In: Advances in Neural In- formation Processing Systems (NeurIPS) (2020)

    2020

  30. [38]

    In: CVPR (2015)

    Vedantam, R., Lawrence Zitnick, C., Parikh, D.: CIDEr: Consensus-based Image Description Evaluation. In: CVPR (2015)

    2015

  31. [39]

    In: International Conference on Learning Representations (ICLR) (2020)

    Wang, C., Zhang, G., Grosse, R.: Picking winning tickets before training by pre- serving gradient flow. In: International Conference on Learning Representations (ICLR) (2020)

    2020

  32. [40]

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

    Wang, H., Zhang, J., Ma, Q.: Exploring intrinsic dimension for vision-language model pruning. In: Forty-first International Conference on Machine Learning (2024)

    2024

  33. [41]

    In: Proc

    Wang, H., Wang, S., Zhang, W.Q., Suo, H., Wan, Y.: Task-agnostic structured pruning of speech representation models. In: Proc. Interspeech 2023. pp. 231–235 (2023)

    2023

  34. [42]

    arXiv preprint arXiv:2409.12191 (2024)

    Wang, P., Bai, S., Tan, S., Wang, S., Fan, Z., Bai, J., Chen, K., Liu, X., Wang, J., Ge, W., Fan, Y., Dang, K., Du, M., Ren, X., Men, R., Liu, D., Zhou, C., Zhou, J., Lin, J.: Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution. arXiv preprint...

    2024 arXiv

  35. [43]

    In: Find- ings of the Association for Computational Linguistics: ACL 2023 (2023)

    Wang, T., Zhou, W., Zeng, Y., Zhang, X.: EfficientVLM: Fast and accurate vision- language models via knowledge distillation and modal-adaptive pruning. In: Find- ings of the Association for Computational Linguistics: ACL 2023 (2023)

    2023

  36. [44]

    In: International Conference on Learning Representations (ICLR) (2023)

    Wang, Y., Li, D., Sun, R.: Ntk-sap: Improving neural network pruning by align- ing training dynamics. In: International Conference on Learning Representations (ICLR) (2023)

    2023

  37. [45]

    Williams, M., Aletras, N.: How does calibration data affect the post-training prun- ing and quantization of large language models? arXiv preprint arXiv:2311.09755 (2023)

    2023 arXiv

  38. [46]

    In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining

    Xu, J., Tan, X., Luo, R., Song, K., Li, J., Qin, T., Liu, T.Y.: Nas-bert: Task- agnostic and adaptive-size bert compression with neural architecture search. In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. pp. 1933–1943 (2021)

    2021

  39. [47]

    arXiv preprint arXiv:2310.05175 (2023)

    Yin, L., Wu, Y., Zhang, Z., Hsieh, C.Y., Wang, Y., Jia, Y., Li, G., Jaiswal, A., Pechenizkiy, M., Liang, Y., et al.: Outlier weighed layerwise sparsity (owl): A miss- ing secret sauce for pruning llms to high sparsity. arXiv preprint arXiv:2310.05175 (2023)

    2023 arXiv

  40. [48]

    silver zippers,

    Zhang,Z.,Gong,B.,Chen,Y.,Han,X.,Zeng,G.,Zhao,W.,Chen,Y.,Liu,Z.,Sun, M.: Bmcook: A task-agnostic compression toolkit for big models. In: Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. pp. 396–405 (2022) PORTA 19 Su...

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.