Pith. sign in

REVIEW 3 major objections 5 minor 44 references

MambaMap: Online Vectorized HD Map Construction using State Space Model

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read MambaMap shows that fusing past BEV features and instance queries through a gated state space model improves online vectorized HD map construction, outperforming state-of-the-art methods on nuScenes and Argoverse2.

desk verdict A plausible, well-ablated SSM temporal-fusion system for online HD map construction, with a real SOTA story on nuScenes but an internally inconsistent gain claim and no variance estimates yet. read the letter →

arxiv 2507.20224 v1 pith:4RZY5KFY submitted 2025-07-27 cs.CV

Ruizi Yang , Xiaolu Liu , Junbo Chen , Jianke Zhu This is my paper
classification cs.CV
keywords onlineHDmapconstructionvectorizedstatespacemodeltemporalfusionmemorybankBEVfeaturesinstancequeriesautonomousdriving
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 tries to establish that online vectorized HD map construction benefits from long-range temporal fusion carried out by a state space model instead of a single recurrent hidden state or stacked attention over many frames. The proposed framework, MambaMap, keeps a small memory bank of historical BEV features and instance queries, warps them to the current frame, and fuses them through a gated state space block with multi-directional scanning. It also reorders historical instance queries to match the current frame before scanning them in instance-first and temporal-first orders. The paper reports that this consistently beats previous state-of-the-art methods on nuScenes and Argoverse2, including geographically disjoint splits, at both 60x30m and 100x50m perception ranges.

What carries the argument

The carrying device is the Gated State Space block: each 1D sequence is projected into a lower-dimensional state-space path (using the DSS diagonal state-space layer) and a higher-dimensional gating signal, then the gated output is element-wise multiplied and added to the input as a residual. Applied to four scanning directions over the BEV feature map and to the two query-order sequences, it gives linear-time receptive fields over the warped history. The other load-bearing piece is the memory bank of size N=4 with Hungarian query-order alignment, which supplies the temporal context without storing long redundant sequences.

What would settle it

A decisive test: on a sequence with closely spaced or crossing lanes, compare MambaMap's Hungarian assignments with ground-truth instance correspondences over frames. If L2 query matching often swaps identities, the fused queries should show visible errors at crossings, and replacing the L2 cost with the ground-truth assignment should raise mAP; if instead the model stays accurate and the permutation matches truth, the assumption holds.

Watch

Extended reading notes

Core claim

The central claim is that selective temporal fusion in state space is the right mechanism for streaming HD map construction. Concretely, MambaMap stores the last N refined BEV features and N refined instance query sets, aligns historical BEV features to the current frame with a 4x4 ego-motion warp, fuses them via a Gated State Space block (a diagonal state-space layer with GELU gating and residual connection), and scans the fused BEV map in left, right, up, and down directions before averaging. At the instance level, historical queries are reordered by Hungarian matching on L2 distance and scanned both instance-first (all instances of a frame together) and temporal-first (same instance slot across frames) to produce fused queries. The author's claim is that this design captures long-range temporal dependencies with linear complexity and yields state-of-the-art mAP: 67.3/66.5 on nuScenes small/large range, 64.9/60.7 on Argoverse2, with a 3.9 mAP gain over the StreamMapNet baseline on the small nuScenes range.

Load-bearing premise

The load-bearing premise is that the Euclidean distance between two learned instance query vectors reliably identifies whether they represent the same road element across frames; if that fails, the ordering step can mix up different map elements and corrupt the temporal fusion.

Editorial extensions

If this is right

  • On the evidence reported, SSM-based fusion is a direct alternative to recurrent hidden-state streaming for vectorized map construction, with linear complexity in the number of historical frames.
  • A memory bank of four frames plus gated selection is enough to recover most of the temporal benefit; growing the bank to six frames slightly hurts accuracy, suggesting redundancy and noise start to dominate.
  • Because the gains hold on geographically disjoint splits, the temporal mechanism appears to generalize to unseen city layouts rather than memorizing training locations.
  • Instance matching by plain L2 distance between query vectors is sufficient for temporal consistency, avoiding extra supervision or ground-truth correspondences.

Reading between the lines

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

  • One testable extension is to make the instance matching cost part of training (e.g., via a learned matching head) and see whether the 3.9 mAP gain grows or shrinks; the paper's ablations compare costs but do not analyze when L2 matching fails.
  • The same memory-bank-plus-gated-SSM recipe could be transferred to other streaming BEV tasks, such as 3D object detection or motion prediction, where per-instance temporal identity is also important.
  • The saturation at N=4 suggests the model learns which historical frames matter; an explicit learned importance weighting over the memory bank might push the operating point past N=4, a possibility the paper does not explore.
  • Because the method effectively tracks instances through query matching, it could produce stable instance IDs over time from the same machinery, which would be useful for downstream planning.
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 MambaMap, an online vectorized HD map construction method that fuses temporal information at both BEV and instance-query levels using a state-space model. A memory bank stores recent BEV features and instance queries; BEV Mamba Fusion applies gated state-space blocks with multi-directional scanning, while Instance Mamba Fusion matches queries across frames by L2 distance and scans spatial-temporal sequences. Experiments on nuScenes and Argoverse2 report improved mAP over prior methods, including on geographically disjoint splits, with ablations over fusion levels, scanning strategies, matching costs, SSM variants, and memory-bank size.

Significance. If the reported results are reproducible, the paper makes a useful contribution: it demonstrates that state-space temporal fusion can improve long-range temporal consistency in online vectorized HD map construction, and the ablations are unusually complete, covering fusion levels, scanning strategies, matching costs, SSM variants, and memory-bank size. The fixed-size memory bank with gated SSM is a clean way to avoid quadratic sequence cost. However, the central SOTA claim is currently supported only by single-run point estimates and is weakened by an internal contradiction in the text, so the empirical evidence needs strengthening before the claim is established.

major comments (3)
  1. [§IV-B, Table III] In the paragraph after Table III, the text states that 'our MambaMap outperforms all competing methods with a minimum gain of 3.1 mAP,' but the table's own numbers contradict this: on the Argoverse2 geographically disjoint split, MambaMap (61.0 mAP) is only 0.4 mAP ahead of SQD-MapNet and MemFusionMap (both 60.6), and on the nuScenes disjoint split the smallest gain is 2.1 mAP over MemFusionMap (40.1 vs 38.0), not 3.1. This internal inconsistency directly undermines the 'substantial margins' claim and must be corrected, with the reported margins restated consistently.
  2. [§IV-B, Tables I-III] The SOTA claim rests entirely on point estimates from what appear to be single training runs; no error bars, standard deviations, or repeated-seed statistics are reported for any table. The closest reported margin (0.4 mAP on the Argoverse2 new split) is within typical run-to-run variation for this benchmark, so without variance estimates the superiority claim is not statistically established. In addition, the paper mixes published baseline numbers (marked '*') with baselines 'reproduced using official public codes' but gives no reproduction configuration or code release, making the comparison hard to audit. Please report multiple-seed statistics and release code/checkpoints or detailed reproduction settings.
  3. [§III-D, Eqs. (9)–(11)] The instance-level temporal fusion assumes that L2 distance between query vectors is a reliable proxy for the identity of the same map element across frames, and the Hungarian matching in Eq. (10) reorders history under that proxy. Table VI compares three matching costs by final mAP but does not analyze how often the matching is correct or whether query vectors carry stable instance identity. This is a correctness-risk concern: if the L2 cost frequently mismatches elements, the temporal fusion could swap features across instances and corrupt predictions. A concrete test would be to measure the ground-truth-instance agreement of the Hungarian matches on a sample of frames, or to compare against matching by predicted geometry; at minimum, the paper should discuss this limitation.
minor comments (5)
  1. [Table IV] The checkmark-only rows should be explicitly labeled; the first row has no checkmarks, and the rows with BMF or IMF alone are ambiguous without a legend.
  2. [Figure 2] The text label 'Fused Quries' contains a typo and should read 'Fused Queries'.
  3. [§IV-A] The hyperparameter list uses both 'Nnum = 100' and the symbol 'N_q' from Section III-D without defining their relationship; please unify the notation.
  4. [Tables I and II] The FPS column shows MambaMap is slower than StreamMapNet (12.7 vs 14.9 on nuScenes; 13.6 vs 15.9 on Argoverse2), but the text emphasizes computational efficiency and does not comment on this overhead; please add a sentence discussing the speed-accuracy trade-off.
  5. [§IV-B and Table III] The abstract and conclusion claim superiority 'across various splits and perception ranges,' but the geographically disjoint-split experiments in Table III are reported only at the 30 m range; the 50 m range on the new splits is missing. Please either add those results or restate the claim to match the experiments.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: MambaMap's performance claims come from external benchmark experiments, and its self-citations are not load-bearing premises.

full rationale

The paper's central claim is an empirical one: that MambaMap achieves state-of-the-art mAP on nuScenes and Argoverse2. These numbers are produced by training and evaluating on held-out validation splits of external datasets; they are not obtained by fitting a parameter to the claimed result or by substituting one equation into another. The architecture defined in Section III relies on standard components, including a memory bank, gated DSS blocks, and scanning strategies, and the effectiveness of each design choice is tested through ablations in Tables IV-VIII against the same external evaluation metric. Choosing the best variant from validation results is normal model selection, not circular prediction. The paper cites its own prior works, MGMap [40] and uncertainty-instructed structure injection [24], but only as related or comparison items; no theorem, fitted constant, or premise from those works is required for MambaMap's derivation. The SSM machinery is attributed to external works such as DSS [29] and GSS [30], not to the authors' own prior results. The instance matching cost in Section III-D is an assumption about query similarity, tested against alternatives in Table VI, and is not presented as a consequence of the final performance. The one notable defect is an internal inconsistency: Section IV-B states a 'minimum gain of 3.1 mAP' on the new split, whereas Table III shows a gain of 2.1 mAP on the nuScenes new split (38.0 to 40.1) and only 0.4 mAP on the Argoverse2 new split (60.6 to 61.0). This, along with the absence of error bars, weakens the strength of the SOTA claim, but it is a numerical/textual discrepancy in reported empirical results rather than a circular argument. No step in the paper's derivation chain reduces to its own inputs, so the circularity score is low.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No physical entities or new formal objects are introduced. MambaMap is a composition of existing neural network components; its free parameters are ordinary hyperparameters selected by hand or by light ablation. The main new assumptions are the L2 query matching rule and the reliability of ego-motion warping.

free parameters (5)
  • memory bank size N = 4
    Number of historical frames kept in the memory bank, chosen via ablation (Table VIII). The central temporal fusion depends on this value.
  • SSM dimensionality reduction factor alpha = 0.5
    Controls the lower-dimensional projection in Eq. 7; set by hand and not ablated.
  • SSM expansion factor beta = 4
    Controls the gating projection width in Eq. 6; set by hand and not ablated.
  • loss weight lambda1 (point loss) = 5.0
    Balances polyline point loss in Eq. 12; chosen without reported sensitivity analysis.
  • loss weight lambda2 (classification loss) = 50.0
    Balances focal classification loss in Eq. 12; chosen without reported sensitivity analysis.
assumptions (4)
  • domain assumption Ego-motion transforms used to warp historical BEV features are accurate enough for feature alignment
    Equation (3) aligns historical BEV features using Tt-k->t; inaccurate localization would misalign the memory bank and reduce fusion quality.
  • domain assumption L2 distance between queries is a reliable proxy for instance identity across frames
    Section III-D, Eq. 9, is the basis for Hungarian matching in Instance Mamba Fusion. The paper does not prove that query ordering is stable across time.
  • domain assumption Baselines marked as reproduced are run under identical settings to MambaMap
    Tables I-III mix published numbers (marked with *) and reproductions from official code; any mismatch in image size, epochs, or hardware could change the claimed margins.
  • standard math DSS layer is a valid state space model for BEV and query sequences
    The DSS layer from [29] is used in Eq. 7 with no modification; the authors' own Table VII shows it outperforms S4, Mamba, and Mamba-2 in this setting.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MambaMap: Online Vectorized HD Map Construction using State Space Model." pith.science (2026). https://pith.science/paper/4RZY5KFY

@misc{pith2026250720224,
  author       = {Pith},
  title        = {Pith review of: MambaMap: Online Vectorized HD Map Construction using State Space Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4RZY5KFY}},
  note         = {Machine review of arXiv:2507.20224}
}
read the original abstract

High-definition (HD) maps are essential for autonomous driving, as they provide precise road information for downstream tasks. Recent advances highlight the potential of temporal modeling in addressing challenges like occlusions and extended perception range. However, existing methods either fail to fully exploit temporal information or incur substantial computational overhead in handling extended sequences. To tackle these challenges, we propose MambaMap, a novel framework that efficiently fuses long-range temporal features in the state space to construct online vectorized HD maps. Specifically, MambaMap incorporates a memory bank to store and utilize information from historical frames, dynamically updating BEV features and instance queries to improve robustness against noise and occlusions. Moreover, we introduce a gating mechanism in the state space, selectively integrating dependencies of map elements in high computational efficiency. In addition, we design innovative multi-directional and spatial-temporal scanning strategies to enhance feature extraction at both BEV and instance levels. These strategies significantly boost the prediction accuracy of our approach while ensuring robust temporal consistency. Extensive experiments on the nuScenes and Argoverse2 datasets demonstrate that our proposed MambaMap approach outperforms state-of-the-art methods across various splits and perception ranges. Source code will be available at https://github.com/ZiziAmy/MambaMap.

Figures

Figures reproduced from arXiv: 2507.20224 by the authors.

Figure 1
Figure 1. (a) MambaMap framework for online HD map construction with [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) Overview of our proposed MambaMap approach. Temporal fusion is performed at both BEV and instance levels using a Memory Bank for [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. (a) BEV Scanning Strategy: BEV features are processed using a [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Qualitative visualization comparison with StreamMapNet [10] across different scenarios. In the HD maps, green lines represent road boundaries, [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

44 extracted references · 28 canonical work pages

  1. [1]

    Tnt: Target-driven trajectory prediction

    Hang Zhao, Jiyang Gao, Tian Lan, Chen Sun, Ben Sapp, Balakrishnan Varadarajan, Yue Shen, Yi Shen, Yuning Chai, Cordelia Schmid, et al. Tnt: Target-driven trajectory prediction. In CoRL, pages 895–904. PMLR, 2021

    work page 2021

  2. [2]

    Vectornet: Encoding hd maps and agent dynamics from vectorized representation

    Jiyang Gao, Chen Sun, Hang Zhao, Yi Shen, Dragomir Anguelov, Congcong Li, and Cordelia Schmid. Vectornet: Encoding hd maps and agent dynamics from vectorized representation. In CVPR, pages 11525–11533, 2020

    work page 2020

  3. [3]

    Loam: Lidar odometry and mapping in real-time

    Ji Zhang, Sanjiv Singh, et al. Loam: Lidar odometry and mapping in real-time. In RSS, volume 2, pages 1–9. Berkeley, CA, 2014

    work page 2014

  4. [4]

    Lego-loam: Lightweight and ground- optimized lidar odometry and mapping on variable terrain

    Tixiao Shan and Brendan Englot. Lego-loam: Lightweight and ground- optimized lidar odometry and mapping on variable terrain. In IROS, pages 4758–4765. IEEE, 2018

    work page 2018

  5. [5]

    Hdmapnet: An online hd map construction and evaluation framework

    Qi Li, Yue Wang, Yilun Wang, and Hang Zhao. Hdmapnet: An online hd map construction and evaluation framework. In ICRA, pages 4628–

    work page

  6. [6]

    Vectormapnet: End-to-end vectorized hd map learning

    Yicheng Liu, Tianyuan Yuan, Yue Wang, Yilun Wang, and Hang Zhao. Vectormapnet: End-to-end vectorized hd map learning. InICML, pages 22352–22369. PMLR, 2023

    work page 2023

  7. [7]

    Maptr: Structured modeling and learning for online vectorized hd map construction

    Bencheng Liao, Shaoyu Chen, Xinggang Wang, Tianheng Cheng, Qian Zhang, Wenyu Liu, and Chang Huang. Maptr: Structured modeling and learning for online vectorized hd map construction. In ICLR, 2022

    work page 2022

  8. [8]

    End-to-end vectorized hd-map construction with piecewise bezier curve

    Limeng Qiao, Wenjie Ding, Xi Qiu, and Chi Zhang. End-to-end vectorized hd-map construction with piecewise bezier curve. In CVPR, pages 13218–13228, 2023

    work page 2023

Show all 44 references
  1. [9]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. NeurIPS, 30, 2017

    2017

  2. [10]

    Streammapnet: Streaming mapping network for vectorized online hd map construction

    Tianyuan Yuan, Yicheng Liu, Yue Wang, Yilun Wang, and Hang Zhao. Streammapnet: Streaming mapping network for vectorized online hd map construction. In WACV, pages 7356–7365, 2024

    2024

  3. [11]

    Bevformer: Learning bird’s- eye-view representation from multi-camera images via spatiotemporal transformers

    Zhiqi Li, Wenhai Wang, Hongyang Li, Enze Xie, Chonghao Sima, Tong Lu, Yu Qiao, and Jifeng Dai. Bevformer: Learning bird’s- eye-view representation from multi-camera images via spatiotemporal transformers. In ECCV, pages 1–18. Springer, 2022

    2022

  4. [12]

    Sparse4d v2: Recurrent temporal fusion with sparse model

    Xuewu Lin, Tianwei Lin, Zixiang Pei, Lichao Huang, and Zhizhong Su. Sparse4d v2: Recurrent temporal fusion with sparse model. arXiv preprint arXiv:2305.14018, 2023

    2023 arXiv

  5. [13]

    Bevdet4d: Exploit temporal cues in multi-camera 3d object detection

    Junjie Huang and Guan Huang. Bevdet4d: Exploit temporal cues in multi-camera 3d object detection. arXiv preprint arXiv:2203.17054 , 2022

    2022 arXiv

  6. [14]

    Bevformer v2: Adapting modern image backbones to bird’s- eye-view recognition via perspective supervision

    Chenyu Yang, Yuntao Chen, Hao Tian, Chenxin Tao, Xizhou Zhu, Zhaoxiang Zhang, Gao Huang, Hongyang Li, Yu Qiao, Lewei Lu, et al. Bevformer v2: Adapting modern image backbones to bird’s- eye-view recognition via perspective supervision. In CVPR, pages 17830–17839, 2023

    2023

  7. [15]

    Stream query denoising for vectorized hd-map construction

    Shuo Wang, Fan Jia, Weixin Mao, Yingfei Liu, Yucheng Zhao, Zehui Chen, Tiancai Wang, Chi Zhang, Xiangyu Zhang, and Feng Zhao. Stream query denoising for vectorized hd-map construction. In ECCV, pages 203–220. Springer, 2024

    2024

  8. [16]

    Maptracker: Tracking with strided memory fusion for consistent vector hd mapping

    Jiacheng Chen, Yuefan Wu, Jiaqi Tan, Hang Ma, and Yasutaka Furukawa. Maptracker: Tracking with strided memory fusion for consistent vector hd mapping. In ECCV, pages 90–107. Springer, 2025

    2025

  9. [17]

    Memfusionmap: Working memory fusion for online vectorized hd map construction

    Jingyu Song, Xudong Chen, Liupei Lu, Jie Li, and Katherine A Skin- ner. Memfusionmap: Working memory fusion for online vectorized hd map construction. arXiv preprint arXiv:2409.18737 , 2024

    2024 arXiv

  10. [18]

    Rmem: Restricted memory banks improve video object segmentation

    Junbao Zhou, Ziqi Pang, and Yu-Xiong Wang. Rmem: Restricted memory banks improve video object segmentation. In CVPR, pages 18602–18611, 2024

    2024

  11. [19]

    Combining recurrent, convolutional, and continuous-time models with linear state space layers

    Albert Gu, Isys Johnson, Karan Goel, Khaled Saab, Tri Dao, Atri Rudra, and Christopher R ´e. Combining recurrent, convolutional, and continuous-time models with linear state space layers. NeurIPS, 34:572–585, 2021

    2021

  12. [20]

    Efficiently model- ing long sequences with structured state spaces

    Albert Gu, Karan Goel, and Christopher R ´e. Efficiently model- ing long sequences with structured state spaces. arXiv preprint arXiv:2111.00396, 2021

    2021 arXiv

  13. [21]

    Mamba: Linear-time sequence modeling with selective state spaces

    Albert Gu and Tri Dao. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752 , 2023

    2023 arXiv

  14. [22]

    Bevformer: learning bird’s-eye- view representation from lidar-camera via spatiotemporal transformers

    Zhiqi Li, Wenhai Wang, Hongyang Li, Enze Xie, Chonghao Sima, Tong Lu, Qiao Yu, and Jifeng Dai. Bevformer: learning bird’s-eye- view representation from lidar-camera via spatiotemporal transformers. PAMI, 2024

    2024

  15. [23]

    Detr4d: Direct multi-view 3d object detection with sparse attention

    Zhipeng Luo, Changqing Zhou, Gongjie Zhang, and Shijian Lu. Detr4d: Direct multi-view 3d object detection with sparse attention. arXiv preprint arXiv:2212.07849 , 2022

    2022 arXiv

  16. [24]

    Uncertainty-instructed structure injection for generalizable hd map construction

    Xiaolu Liu, Ruizi Yang, Song Wang, Wentong Li, Junbo Chen, and Jianke Zhu. Uncertainty-instructed structure injection for generalizable hd map construction. In CVPR, pages 22359–22368, 2025

    2025

  17. [25]

    End-to-end object detec- tion with transformers

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to-end object detec- tion with transformers. In ECCV, pages 213–229. Springer, 2020

    2020

  18. [26]

    Maptrv2: An end- to-end framework for online vectorized hd map construction

    Bencheng Liao, Shaoyu Chen, Yunchi Zhang, Bo Jiang, Qian Zhang, Wenyu Liu, Chang Huang, and Xinggang Wang. Maptrv2: An end- to-end framework for online vectorized hd map construction. IJCV, pages 1–23, 2024

    2024

  19. [27]

    Pivotnet: Vectorized pivot learning for end-to-end hd map construction

    Wenjie Ding, Limeng Qiao, Xi Qiu, and Chi Zhang. Pivotnet: Vectorized pivot learning for end-to-end hd map construction. In ICCV, pages 3672–3682, 2023

    2023

  20. [28]

    Dn-detr: Accelerate detr training by introducing query denoising

    Feng Li, Hao Zhang, Shilong Liu, Jian Guo, Lionel M Ni, and Lei Zhang. Dn-detr: Accelerate detr training by introducing query denoising. In CVPR, pages 13619–13627, 2022

    2022

  21. [29]

    Diagonal state spaces are as effective as structured state spaces

    Ankit Gupta, Albert Gu, and Jonathan Berant. Diagonal state spaces are as effective as structured state spaces. NeurIPS, 35:22982–22994, 2022

    2022

  22. [30]

    Long range language modeling via gated state spaces

    Harsh Mehta, Ankit Gupta, Ashok Cutkosky, and Behnam Neyshabur. Long range language modeling via gated state spaces. arXiv preprint arXiv:2206.13947, 2022

    2022 arXiv

  23. [31]

    Vision mamba: Efficient visual represen- tation learning with bidirectional state space model

    Lianghui Zhu, Bencheng Liao, Qian Zhang, Xinlong Wang, Wenyu Liu, and Xinggang Wang. Vision mamba: Efficient visual represen- tation learning with bidirectional state space model. arXiv preprint arXiv:2401.09417, 2024

    2024 arXiv

  24. [32]

    Vmamba: Visual state space model

    Yue Liu, Yunjie Tian, Yuzhong Zhao, Hongtian Yu, Lingxi Xie, Yaowei Wang, Qixiang Ye, Jianbin Jiao, and Yunfan Liu. Vmamba: Visual state space model. NeurIPS, 37:103031–103063, 2025

    2025

  25. [33]

    Videomamba: State space model for efficient video understanding

    Kunchang Li, Xinhao Li, Yi Wang, Yinan He, Yali Wang, Limin Wang, and Yu Qiao. Videomamba: State space model for efficient video understanding. In ECCV, pages 237–255. Springer, 2025

    2025

  26. [34]

    Feature pyramid networks for object detection

    Tsung-Yi Lin, Piotr Doll ´ar, Ross Girshick, Kaiming He, Bharath Hariharan, and Serge Belongie. Feature pyramid networks for object detection. In CVPR, pages 2117–2125, 2017

    2017

  27. [35]

    Deformable detr: Deformable transformers for end-to-end object detection

    Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, and Jifeng Dai. Deformable detr: Deformable transformers for end-to-end object detection. arXiv preprint arXiv:2010.04159 , 2020

    2010 arXiv

  28. [36]

    Gaussian error linear units (gelus)

    Dan Hendrycks and Kevin Gimpel. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415 , 2016

    2016 arXiv

  29. [37]

    Efficientnet: Rethinking model scaling for convolutional neural networks

    Mingxing Tan and Quoc Le. Efficientnet: Rethinking model scaling for convolutional neural networks. In ICML, pages 6105–6114. PMLR, 2019

    2019

  30. [38]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR, pages 770–778, 2016

    2016

  31. [39]

    Online vectorized hd map construction using geometry

    Zhixin Zhang, Yiyuan Zhang, Xiaohan Ding, Fusheng Jin, and Xi- angyu Yue. Online vectorized hd map construction using geometry. arXiv preprint arXiv:2312.03341 , 2023

    2023 arXiv

  32. [40]

    Mgmap: Mask-guided learning for online vectorized hd map construction

    Xiaolu Liu, Song Wang, Wentong Li, Ruizi Yang, Junbo Chen, and Jianke Zhu. Mgmap: Mask-guided learning for online vectorized hd map construction. In CVPR, pages 14812–14821, 2024

    2024

  33. [41]

    nuscenes: A multimodal dataset for autonomous driving

    Holger Caesar, Varun Bankiti, Alex H Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom. nuscenes: A multimodal dataset for autonomous driving. In CVPR, pages 11621–11631, 2020

    2020

  34. [42]

    Argoverse 2: Next generation datasets for self- driving perception and forecasting

    Benjamin Wilson, William Qi, Tanmay Agarwal, John Lambert, Jagjeet Singh, Siddhesh Khandelwal, Bowen Pan, Ratnesh Kumar, Andrew Hartnett, Jhony Kaesemodel Pontes, Deva Ramanan, Peter Carr, and James Hays. Argoverse 2: Next generation datasets for self- driving perception and f...

    2021

  35. [43]

    Fixing weight decay regularization in adam

    Ilya Loshchilov, Frank Hutter, et al. Fixing weight decay regularization in adam. arXiv preprint arXiv:1711.05101 , 5:5, 2017

    2017 arXiv

  36. [44]

    Transformers are ssms: Generalized models and efficient algorithms through structured state space duality

    Tri Dao and Albert Gu. Transformers are ssms: Generalized models and efficient algorithms through structured state space duality. arXiv preprint arXiv:2405.21060, 2024

    2024 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.