Pith. sign in

REVIEW 5 major objections 5 minor 27 references

LLaVA-Zip: Adaptive Visual Token Compression with Intrinsic Image Information

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

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

Pith's one-line read A standard-deviation statistic computed from each image's visual features decides how many tokens LLaVA retains, and the paper reports that this beats random compression on eight benchmarks.

desk verdict The DFMR module is a cheap, sensible idea, but Table 1 never evaluates the adaptive rule it proposes, so the headline result is an artifact of the training distribution, not a validated improvement. read the letter →

arxiv 2412.08771 v1 pith:QVEHIHNE submitted 2024-12-11 cs.CV cs.AIcs.LG

Ke Wang , Hong Xuan This is my paper · ORCID
classification cs.CVcs.AIcs.LG
keywords visualtokencompressionmultimodallargelanguagemodelsLLaVAdynamicfeaturemapreductionaveragepoolingstandarddeviationmetricadaptivebudgetsefficientvision-language
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 proposes a plug-in module, Dynamic Feature Map Reduction (DFMR), for the LLaVA-1.5 multimodal model. DFMR decides how aggressively to average-pool visual tokens by measuring the variability of each image's visual feature map: images with high patch-to-patch variance keep more tokens, while flat, repetitive images are compressed more. The paper reports that training LLaVA-1.5 with this adaptive scheme improves average scores across eight visual question-answering and multimodal benchmarks compared with both the uncompressed model and a baseline that compresses tokens by a fixed random factor. The motivation is practical: if visual tokens can be cut without losing accuracy, the same model can handle multiple images or video within a fixed token limit and with less memory.

What carries the argument

The load-bearing object is the mean patch standard deviation $\sigma$ of the visual feature map. For each window of the feature map, DFMR computes the standard deviation of the token values, averages these over all windows, and compares the average to a manually set threshold $\tau = 5\times10^{-2}$. A low $\sigma$ signals repetitive or uniform content and licenses a larger pooling factor; a high $\sigma$ signals fine detail and forces a smaller pooling factor. The threshold is the single control knob that trades token savings against fidelity, and the paper shows its effect on the distribution of chosen compression factors.

What would settle it

Count how often DFMR selects each compression factor on a held-out image set under $\tau = 5\times10^{-2}$, compute the average number of visual tokens DFMR and the random baseline actually use, and re-run the eight benchmarks with DFMR's per-image choices matched by token count to the random baseline. If the benchmark averages converge once token budgets are equalized, the content metric is not the cause of the improvement.

Watch

Extended reading notes

Core claim

The central claim is that intrinsic image statistics alone can guide how many visual tokens a multimodal language model needs. DFMR computes, for each candidate compression factor $s$, the mean of the per-patch standard deviations over the visual feature map, and increases $s$ only while this value stays below a fixed threshold. The resulting per-image choice of 576, 144, or 64 tokens is used during both pretraining and fine-tuning. On the eight benchmarks reported, LLaVA-1.5 with DFMR outperforms the random-compression baseline at each evaluated token length and on average, with the largest gains appearing at heavier compression. The paper therefore claims that compression guided by image content is better than compression guided by chance.

Load-bearing premise

The comparison treats DFMR and the random baseline as if they use the same average number of tokens, but the paper never reports how often DFMR picks each compression level; if DFMR mostly keeps 576 tokens, its higher benchmark scores may come from more tokens rather than from the content metric.

Editorial extensions

If this is right

  • A single LLaVA-style model can process one image, several images, or video frames without exceeding the LLM's token limit, by assigning each image its own compression level.
  • Training with adaptive compression acts as a form of data augmentation: the same image contributes tokens of several lengths, which the paper argues can ease the scarcity of open-domain image-text pairs in continued pretraining.
  • The threshold $\tau$ gives practitioners a tunable knob to shift the operating point between accuracy and memory, with lower thresholds favoring more compression.
  • Because the decision rule is a simple statistic computed before the projector, DFMR adds only constant-time overhead during data loading and training.

Reading between the lines

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

  • If DFMR's gains come from content-aware selection rather than extra tokens, then a matched-budget comparison would isolate the contribution of the variance metric; the paper does not report how often each compression factor is chosen.
  • The same standard-deviation heuristic could be ported to other vision encoders and pooling schemes, with the threshold re-calibrated per encoder.
  • A testable extension is to use the metric to choose compression per region or per frame in video, rather than per whole image.
  • Images with low $\sigma$ are likely compressible before tokenization, so the metric could also guide storage or transmission savings, not just in-model token counts.
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

5 major / 5 minor

Summary. The paper proposes Dynamic Feature Map Reduction (DFMR), a module inserted between the vision encoder and projector of LLaVA-1.5. DFMR computes the mean within-patch standard deviation of CLIP visual features and uses a manually chosen threshold tau=5e-2 to select a compression factor s in {1,2,3}, then applies s x s average pooling. The authors train three models: the standard LLaVA-1.5, LLaVA-1.5 Random in which s is sampled uniformly per image, and LLaVA-1.5 DFMR in which s is selected by the proposed rule. They evaluate all models at fixed token counts of 576, 144, and 64 on eight benchmarks and report that DFMR improves over the Random baseline on average and across most rows. The central claim is that using intrinsic image information to choose per-image compression improves performance under varied visual token budgets.

Significance. The idea of using a cheap, intrinsic image statistic to guide visual token compression is appealing and potentially useful for resource-constrained multimodal models. The comparison against a stochastic random-compression baseline is a reasonable experimental design and gives the paper some independent grounding. The COCO distribution analysis in Section 4.4 also shows that the proposed standard-deviation metric separates images in a plausible way. However, the main empirical claim is not supported by the evaluation as reported: Table 1 fixes s for all images at evaluation time and never exercises the dynamic per-image selection rule, the training-time distribution of s for DFMR is not reported, and no variance or significance information is given. The threshold tau is hand-picked without ablation. As a result, the reported gains could plausibly stem from an uncontrolled difference in average token budgets during training rather than from the information-content metric. With a corrected evaluation protocol, the method could still be of interest, but the current evidence is insufficient.

major comments (5)
  1. [Section 4.1 (Evaluation Details) and Table 1] The evaluation protocol fixes the compression factor for every image in a given row: “we evaluate its performance on images with compression ratios of 1, 2, and 3.” This means Table 1 never tests DFMR’s claimed dynamic, per-image selection of s; it tests only how well each model performs when forced to use a fixed token budget. The differences between LLaVA-1.5 Random and LLaVA-1.5 DFMR in each row can be fully explained by the different distributions of s seen during training, independent of whether the standard-deviation metric identifies informative images. The paper must report how often DFMR selects each s during training and, ideally, evaluate DFMR’s actual adaptive rule on the test images, rather than only fixed-s rows.
  2. [Section 4.1 (Training Details) and Section 4.3 (Performance)] The comparison between LLaVA-1.5 Random and LLaVA-1.5 DFMR is confounded by an unreported and likely unequal distribution of training token budgets. Random samples s uniformly from {1,2,3}, so its average training budget is 576*(1/3)+144*(1/3)+64*(1/3) = 261.3 tokens. If DFMR’s threshold rule selects s=1 for a large fraction of images, DFMR is trained predominantly at 576 tokens, which would explain why DFMR is ahead at evaluation token counts of 576 and 144 but often behind Random at 64 tokens. The paper must report the empirical distribution of s selected by DFMR over the training data and show the average training token budget; otherwise the average row of Table 1 does not support the claim that DFMR “enhances model performance across all tasks.”
  3. [Section 4.3 (Performance) and Abstract] The abstract states that DFMR “significantly improves” performance, and Section 4.3 repeats “significantly better results,” but no error bars, standard deviations across seeds, or statistical significance tests are reported anywhere. Each benchmark is a single number per configuration, so the reader cannot assess whether the observed differences of a few points are meaningful. The authors should provide at least three training runs or per-benchmark bootstrap confidence intervals, and should temper the word “significantly” unless such evidence is supplied.
  4. [Section 3.2 (DFMR) and Section 4.1 (Implementation Details)] The threshold tau=5e-2 is a manually defined hyperparameter, and the compression factor set {1,2,3} is fixed in advance. The paper gives no ablation or sensitivity analysis for tau, and no principled argument that 5e-2 is the right value. Because tau controls the distribution of s during training, the reported gains could partly reflect a threshold that was tuned with the evaluation benchmarks in mind. The authors should report an ablation over tau, or justify the choice independently of the test benchmarks, and should show that the qualitative conclusions are stable across reasonable threshold values.
  5. [Section 5 (Discussion)] The statement “Our proposed method operates with a computational cost of O(1)” is incorrect for the DFMR module itself: Equations (1)–(3) require computing patch standard deviations over all spatial positions and embedding dimensions, which is O(N_v D_v) in the number of visual tokens and embedding dimension. The actual cost may be small relative to the LLM, but it is not O(1). This should be corrected to an accurate complexity statement, and the claimed absence of dataloader burden should be supported by the relevant measurement if it is meant to be a practical efficiency result.
minor comments (5)
  1. [Section 4.4 (Analysis)] The method is called Dynamic Feature Map Reduction in the abstract and Section 3.2, but Section 4.4 refers to “Dynamic Feature Mixing Ratio (DFMR)”; the acronym expansion should be made consistent throughout.
  2. [Figure 2 caption and Section 4.4] The caption labels the panels as “Bottom 10 Samples” and “Top 10 Samples” with the top being high-standard-deviation images, but the body text says Figure 2(b) shows images “composed of background content or repetitive patterns”, which would correspond to the bottom 10. The panel-to-description mapping is unclear and should be clarified.
  3. [Title and Abstract] The paper title is “LLaVA-Zip” but the proposed module is DFMR and “LLaVA-Zip” is not defined or used in the text; either the method name should be aligned with the title or the title should be changed to avoid confusion.
  4. [Section 4.1 (Implementation Details)] The threshold is described as “fixed threshold of 5e-2”, but Section 3.2 and Figure 2 also discuss thresholds 7e-2 and 9e-2; the relationship between these values and the experiments in Table 1 should be stated explicitly.
  5. [References] Some references appear incomplete or informally formatted, such as [14], where author names are run together; please format all entries according to the journal style.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the sigma metric and threshold rule are input-side heuristics, and the benchmark gains are external empirical claims, not consequences of the definitions.

full rationale

The paper's derivation chain is a heuristic preprocessing rule, not a claim that reduces to its own inputs by construction. DFMR computes per-patch standard deviations of CLIP visual features (Eqs. 1-3), compares their mean to a manually fixed threshold tau, and uses the result to select a pooling factor s in {1,2,3}; the pooled tokens then feed an otherwise standard LLaVA-1.5 model. The metric is defined on the input image features, not on benchmark labels or model outputs, so the benchmark improvements in Table 1 are external empirical findings rather than logical consequences of the definitions. No parameter is fitted to a subset of data and then relabeled as a prediction; tau is stated as a hand-set constant, not derived from the evaluation benchmarks. There are no load-bearing self-citations: the authors cite no prior work of their own, and no uniqueness theorem is imported to force the method. The 'information' language is a rhetorical interpretation of variance, and the use of a manually defined threshold may limit generalizability, but that is an empirical-validity concern, not circularity. A substantive experimental caveat is that Section 4.1 evaluates each model at fixed token counts instead of exercising DFMR's per-image dynamic choice, and the training-time distribution over s for DFMR is never reported; this is a possible confound in the empirical comparison, but it is not an equation-level equivalence between inputs and outputs, so it does not make the derivation circular.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim rests on a hand-set threshold, a predefined set of compression factors, and an untested assumption that feature variance measures information content. No new entities are introduced.

free parameters (2)
  • Threshold tau = 5e-2
    Manually defined threshold that determines the compression factor s for each image; no sensitivity analysis or principled selection method is provided. The choice directly controls the trade-off between token count and fidelity.
  • Compression factor set = {1, 2, 3}
    The set of allowed pooling factors (yielding 576, 144, and 64 tokens). Using a different set changes the reachable token counts and would likely change results.
assumptions (3)
  • ad hoc to paper Mean within-patch standard deviation of CLIP features is a valid proxy for image information content
    The method's compression decision is entirely driven by this metric; the paper provides no empirical validation that the metric correlates with task-relevant information, only qualitative examples.
  • ad hoc to paper The decision rule that stops when sigma exceeds tau yields the intended behavior that lower sigma leads to more compression
    The rule as written is ambiguous and the paper does not clearly specify the monotonic relationship between sigma and s or the exact stopping condition, yet the whole method depends on this behavior.
  • domain assumption Training LLaVA-1.5 with compressed visual tokens generalizes to the evaluated token counts
    The method assumes that a model trained with dynamic token counts can be evaluated fairly at fixed counts of 576, 144, and 64; the paper does not analyze distribution shift in training versus evaluation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLaVA-Zip: Adaptive Visual Token Compression with Intrinsic Image Information." pith.science (2026). https://pith.science/paper/QVEHIHNE

@misc{pith2026241208771,
  author       = {Pith},
  title        = {Pith review of: LLaVA-Zip: Adaptive Visual Token Compression with Intrinsic Image Information},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QVEHIHNE}},
  note         = {Machine review of arXiv:2412.08771}
}
read the original abstract

Multi-modal large language models (MLLMs) utilizing instruction-following data, such as LLaVA, have achieved great progress in the industry. A major limitation in these models is that visual tokens consume a substantial portion of the maximum token limit in large language models (LLMs), leading to increased computational demands and decreased performance when prompts include multiple images or videos. Industry solutions often mitigate this issue by increasing computational power, but this approach is less feasible in academic environments with limited resources. In this study, we propose Dynamic Feature Map Reduction (DFMR) based on LLaVA-1.5 to address the challenge of visual token overload. DFMR dynamically compresses the visual tokens, freeing up token capacity. Our experimental results demonstrate that integrating DFMR into LLaVA-1.5 significantly improves the performance of LLaVA in varied visual token lengths, offering a promising solution for extending LLaVA to handle multi-image and video scenarios in resource-constrained academic environments and it can also be applied in industry settings for data augmentation to help mitigate the scarcity of open-domain image-text pair datasets in the continued pretraining stage.

Figures

Figures reproduced from arXiv: 2412.08771 by the authors.

Figure 1
Figure 1. Overall Structure of LLaVA-Dynamics. The core module is the Dynamic Feature Map Reduction (DFMR), which dynamically compresses the visual tokens from the visual encoder based on the calculated image information. Results Summary. The proposed DFMR module demonstrates clear improvements in all benchmark tasks over the baseline and the original LLaVA-1.5, see [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Data Distribution. The figure illustrates the distribution of the calculated standard deviation metric under different compression ratios s ∈ {1, 2, 3} and using the proposed DFMR with manually defined threshold τ ∈ {5e −2 , 7e −2 , 9e −2}. Bottom 10 Samples. These samples correspond to the 10 images with the smallest calculated mean standard deviation metric σ¯, typically consisting of repetitive or uniform content… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 5 canonical work pages

  1. [1]

    Llava-onevision: Easy visual task transfer

    Bo Li, Yuanhan Zhang, Dong Guo, Renrui Zhang, Feng Li, Hao Zhang, Kaichen Zhang, Peiyuan Zhang, Yanwei Li, Ziwei Liu, et al. Llava-onevision: Easy visual task transfer. arXiv preprint arXiv:2408.03326, 2024

    arXiv 2024

  2. [2]

    Improved baselines with visual instruction tuning

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. Improved baselines with visual instruction tuning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 26296–26306, 2024

    2024

  3. [3]

    Visual instruction tuning

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning. Advances in neural information processing systems, 36, 2024

    2024

  4. [4]

    Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International conference on machine learning , pages 19730–19742. PMLR, 2023. 6 A PREPRINT - D ECEMBER 13, 2024

    work page 2023

  5. [5]

    Less is more: A simple yet effective token reduction method for efficient multi-modal llms

    Dingjie Song, Wenjun Wang, Shunian Chen, Xidong Wang, Michael Guan, and Benyou Wang. Less is more: A simple yet effective token reduction method for efficient multi-modal llms. arXiv preprint arXiv:2409.10994, 2024

    arXiv 2024

  6. [6]

    Beyond llava-hd: Diving into high-resolution large multimodal models

    Yi-Fan Zhang, Qingsong Wen, Chaoyou Fu, Xue Wang, Zhang Zhang, Liang Wang, and Rong Jin. Beyond llava-hd: Diving into high-resolution large multimodal models. arXiv preprint arXiv:2406.08487, 2024

    arXiv 2024

  7. [7]

    Hired: Attention-guided token dropping for efficient inference of high-resolution vision-language models in resource-constrained environments

    Kazi Hasan Ibn Arif, JinYi Yoon, Dimitrios S Nikolopoulos, Hans Vandierendonck, Deepu John, and Bo Ji. Hired: Attention-guided token dropping for efficient inference of high-resolution vision-language models in resource-constrained environments. arXiv preprint arXiv:2408.10945, 2024

    arXiv 2024

  8. [8]

    Matryoshka multimodal models

    Mu Cai, Jianwei Yang, Jianfeng Gao, and Yong Jae Lee. Matryoshka multimodal models. arXiv preprint arXiv:2405.17430, 2024

    arXiv 2024

Show all 27 references
  1. [9]

    Flamingo: a visual language model for few-shot learning

    Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katherine Millican, Malcolm Reynolds, et al. Flamingo: a visual language model for few-shot learning. Advances in neural information processing systems , 35:2371...

    2022

  2. [10]

    Focusllava: A coarse-to-fine approach for efficient and effective visual token compression

    Yuke Zhu, Chi Xie, Shuang Liang, Bo Zheng, and Sheng Guo. Focusllava: A coarse-to-fine approach for efficient and effective visual token compression. arXiv preprint arXiv:2411.14228, 2024

    2024 arXiv

  3. [11]

    Qwen-vl: A frontier large vision-language model with versatile abilities

    Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Junyang Lin, Chang Zhou, and Jingren Zhou. Qwen-vl: A frontier large vision-language model with versatile abilities. arXiv preprint arXiv:2308.12966, 2023

    2023 arXiv

  4. [12]

    Learning transferable visual models from natural language supervision

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

    2021

  5. [13]

    Judging llm-as-a-judge with mt-bench and chatbot arena

    Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric Xing, et al. Judging llm-as-a-judge with mt-bench and chatbot arena. Advances in Neural Information Processing Systems, 36:46595–46623, 2023

    2023

  6. [14]

    Lmms-eval: Accelerating the development of large multimoal models, March 2024

    Kaichen Zhang* Fanyi Pu* Xinrun Du Yuhao Dong Haotian Liu Yuanhan Zhang Ge Zhang Chunyuan Li Bo Li*, Peiyuan Zhang* and Ziwei Liu. Lmms-eval: Accelerating the development of large multimoal models, March 2024

    2024

  7. [15]

    Gqa: A new dataset for real-world visual reasoning and com- positional question answering

    Drew A Hudson and Christopher D Manning. Gqa: A new dataset for real-world visual reasoning and com- positional question answering. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6700–6709, 2019

    2019

  8. [16]

    Towards vqa models that can read

    Amanpreet Singh, Vivek Natarajan, Meet Shah, Yu Jiang, Xinlei Chen, Dhruv Batra, Devi Parikh, and Marcus Rohrbach. Towards vqa models that can read. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8317–8326, 2019

    2019

  9. [17]

    Making the v in vqa matter: Elevating the role of image understanding in visual question answering

    Yash Goyal, Tejas Khot, Douglas Summers-Stay, Dhruv Batra, and Devi Parikh. Making the v in vqa matter: Elevating the role of image understanding in visual question answering. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 6904–6913, 2017

    2017

  10. [18]

    Evaluating object hallucination in large vision-language models

    Yifan Li, Yifan Du, Kun Zhou, Jinpeng Wang, Wayne Xin Zhao, and Ji-Rong Wen. Evaluating object hallucination in large vision-language models. arXiv preprint arXiv:2305.10355, 2023

    2023 arXiv

  11. [19]

    Mm-vet: Evaluating large multimodal models for integrated capabilities

    Weihao Yu, Zhengyuan Yang, Linjie Li, Jianfeng Wang, Kevin Lin, Zicheng Liu, Xinchao Wang, and Lijuan Wang. Mm-vet: Evaluating large multimodal models for integrated capabilities. arXiv preprint arXiv:2308.02490, 2023

    2023 arXiv

  12. [20]

    Mme: A comprehensive evaluation benchmark for multimodal large language models

    Chaoyou Fu, Peixian Chen, Yunhang Shen, Yulei Qin, Mengdan Zhang, Xu Lin, Zhenyu Qiu, Wei Lin, Jinrui Yang, Xiawu Zheng, Ke Li, Xing Sun, and Rongrong Ji. Mme: A comprehensive evaluation benchmark for multimodal large language models. ArXiv, abs/2306.13394, 2023

    2023 arXiv

  13. [21]

    Seed-bench: Benchmarking multimodal llms with generative comprehension

    Bohao Li, Rui Wang, Guangzhi Wang, Yuying Ge, Yixiao Ge, and Ying Shan. Seed-bench: Benchmarking multimodal llms with generative comprehension. arXiv preprint arXiv:2307.16125, 2023

    2023 arXiv

  14. [22]

    Microsoft coco: Common objects in context

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

    2014

  15. [23]

    Megatron-lm: Training multi-billion parameter language models using model parallelism

    Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, and Bryan Catanzaro. Megatron-lm: Training multi-billion parameter language models using model parallelism. arXiv preprint arXiv:1909.08053, 2019. 7 A PREPRINT - D ECEMBER 13, 2024

    1909 arXiv

  16. [24]

    Ray: A distributed framework for emerging {AI} applications

    Philipp Moritz, Robert Nishihara, Stephanie Wang, Alexey Tumanov, Richard Liaw, Eric Liang, Melih Elibol, Zongheng Yang, William Paul, Michael I Jordan, et al. Ray: A distributed framework for emerging {AI} applications. In 13th USENIX symposium on operating systems design and...

    2018

  17. [25]

    Ffcv: Accelerating training by removing data bottlenecks

    Guillaume Leclerc, Andrew Ilyas, Logan Engstrom, Sung Min Park, Hadi Salman, and Aleksander M ˛ adry. Ffcv: Accelerating training by removing data bottlenecks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12011–12020, 2023

    2023

  18. [26]

    The llama 3 herd of models

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. The llama 3 herd of models. arXiv preprint arXiv:2407.21783, 2024

    2024 arXiv

  19. [27]

    Video-llava: Learning united visual representation by alignment before projection

    Bin Lin, Yang Ye, Bin Zhu, Jiaxi Cui, Munan Ning, Peng Jin, and Li Yuan. Video-llava: Learning united visual representation by alignment before projection. arXiv preprint arXiv:2311.10122, 2023. 8

    2023 arXiv

Pith tools

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