Pith. sign in

REVIEW 3 major objections 5 minor 65 references

LIRA: Local Cross-Layer Information Routing for Vision-Language-Action Decoding

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims that conditioning each action-decoder block on a small centered window of adjacent VLM layers, rather than one matched layer or all layers, improves VLM-to-action routing and raises zero-shot LIBERO-Plus success from…

desk verdict LIRA is a clean, well-controlled empirical paper on a small but real architectural change—routing each action-decoder block a local window of adjacent VLM layers—and its main unresolved issue is interpretive, not evidentiary. read the letter →

arxiv 2608.07596 v1 pith:5BNVWMZZ submitted 2026-08-06 cs.RO cs.CV

Zhewei Zhang , Puyue Wang , Guanren Qiao , Yijie Weng , Jiawei Hu , Guo Li , Lujia Wang , Junyan Wang
show 5 more authors
Tao Gu Hongliang Lu Guiliang Liu Hong Jia Xinhu Zheng
This is my paper · ORCID
classification cs.ROcs.CV
keywords vision-language-actionmodelsVLM-to-actioninterfacecross-layerroutingdepth-alignedwindowactiondecodingrobotmanipulationzero-shottransferLIBERO-Plus
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

LIRA proposes a specific answer to a previously under-studied question: which intermediate layers of a pretrained vision-language model should feed the action decoder of a robot policy. The paper argues that neither a single matched layer nor the full VLM hierarchy is the right source: each decoder block should receive LIRA Query features from a small window of neighboring VLM layers centered on its matched layer. This depth-aware local routing adds no trainable parameters to the VLA-Adapter baseline, and the paper reports consistent success-rate gains on LIBERO, CALVIN ABC→D, LIBERO-Plus, and a physical robot. The headline result is a zero-shot LIBERO-Plus success improvement from 59.1% to 78.0% at the same 0.5B-parameter scale. The paper's discovery, if correct, is that local cross-layer structure, not layer count or token budget, is what the VLM-to-action interface is missing.

What carries the argument

The load-bearing object is the depth-aligned local routing window of Eq. (1): for Parallel Fusion Block $i$, the window is $W_r(i)=\{\ell\in\{1,\dots,L\}: |\ell-i|\le r\}$, and the LIRA Query features $\mathbf{R}_t^{(\ell)}$ from those layers are concatenated along the token dimension to form $\tilde{\mathbf{R}}_t^{(i)}$. With $r=1$ this is a centered three-layer neighborhood, and the concatenation keeps the mechanism parameter-free because the decoder's cross-attention already consumes variable-length token sequences. The matched task-token features $\mathbf{T}_t^{(i)}$ remain the depth-aligned semantic anchor, while $\tilde{\mathbf{R}}_t^{(i)}$ supplies complementary nearby-depth context. This machinery carries the paper's locality prior over VLM depth and defines the ablation space (window width, routing topology, and query budget).

What would settle it

A concrete check: retrain the same LIRA protocol with the routing window center deliberately offset from the matched layer (for example, centering block $i$ on VLM layer $i+3$) and compare LIBERO-Plus zero-shot success; if the offset preserves the 78.0% level, then depth alignment is not load-bearing, while a sharp drop would confirm that exact layer identity matters. A second check is to permute VLM layer indices while keeping the visual features fixed: if the gains persist after permutation, the locality prior is not about true depth ordering.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that VLM-to-action conditioning is best understood as depth-aware information routing, and that the optimal routing policy is a depth-aligned local window: for the i-th Parallel Fusion Block, the routing window selects all VLM layers within distance r of layer i, with r=1 by default, giving a centered three-layer window (two layers at the boundaries). The LIRA Query features from the selected layers are concatenated along the token dimension and fed through the block's existing cross-attention, so no new projection layers or trainable parameters are introduced. The paper asserts that this centered three-layer window outperforms matched-layer-only conditioning, two- and four-layer neighborhoods, and global aggregation on LIBERO-Long and CALVIN-5, and that it improves the controlled VLA-Adapter baseline on every principal aggregate metric. The strongest reported outcome is zero-shot LIBERO-Plus success rising to 78.0% from the baseline's 59.1%.

Load-bearing premise

The load-bearing premise is that VLM layer $i$ and decoder block $i$ are meaningfully aligned in depth, so a window centered on layer $i$ has a principled center; if the VLM's layer ordering is not semantically monotonic, the center is arbitrary and the reported gains could be an artifact of this matching.

Editorial extensions

If this is right

  • At a fixed 0.5B parameter budget, replacing one matched VLM layer with a centered three-layer window raises LIBERO-Plus zero-shot success from 59.1% to 78.0%, implying that the routing interface itself, not model capacity, drives the robustness gain.
  • Long-horizon tasks inherit the largest benefit: LIBERO-Long success rises from 95.0% to 97.6% and CALVIN five-task completion from 76.5% to 80.5%, so local cross-layer context matters most when actions depend on sustained task evidence.
  • Query-token count alone cannot explain the gain: even 256 last-layer tokens reach only 92.8% on LIBERO-Long, below LIRA's 64-token, three-layer result of 97.6%.
  • Window-width ablations show a centered three-layer window outperforms one-, two-, four-layer, and global aggregation, so the optimal routing is neither single-layer nor full-hierarchy mixing.
  • On the physical robot, LIRA improves pick-and-place from 7/10 to 9/10 and two-stage task completion from 6/10 to 8/10, suggesting the benefit carries across subgoals, though the paper calls these results preliminary.

Reading between the lines

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

  • Editorial inference: the depth-locality prior likely transfers to any multi-layer backbone feeding a depthwise decoder (for example, diffusion policies with per-block conditioning), not only VLM-based action decoders; a direct test would be to route local windows of a frozen vision encoder into a diffusion action head.
  • Editorial inference: the LIBERO-Plus gains may reflect a cheap ensembling effect, since neighboring VLM layers are partially decorrelated; this could be tested by measuring whether the success gain tracks inter-layer feature similarity.
  • Editorial inference: the design effectively restricts the VLM-to-decoder connectivity matrix to a diagonal band of width $2r+1$; a testable extension is to tune $r$ per task or per benchmark rather than fixing $r=1$.
  • Editorial inference: because all headline comparisons use one 0.5B backbone, the optimal window width may depend on model depth; sweeping $r$ on a larger VLM would show whether the locality prior is scale-invariant.
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 LIRA, a local cross-layer routing mechanism for vision-language-action (VLA) models. Rather than conditioning each action-decoder block on a single matched VLM layer (as in VLA-Adapter), LIRA aggregates query-token features from a small window of adjacent VLM layers centered on the matched depth, while retaining the layer-aligned task-token branch. The mechanism concatenates features along the token dimension and therefore introduces no new trainable parameters. The authors evaluate LIRA against VLA-Adapter and other systems on LIBERO, CALVIN ABC→D, LIBERO-Plus (zero-shot and fine-tuned), and a Franka real-robot setup, reporting consistent gains, most notably a 59.1% to 78.0% improvement in zero-shot LIBERO-Plus success. Ablations cover conditioning source, window width, routing topology, and query-token budget.

Significance. If the results hold, LIRA is a simple, parameter-free architectural improvement that exposes complementary intermediate VLM features to a compact action decoder, with a particularly large robustness gain under distribution shift. The controlled comparison at the same 0.5B scale, the ablation that rules out query-token budget as the sole cause of the gain, and the explicit no-additional-parameters property are clear strengths. However, the central depth-locality claim—that adjacency in VLM depth, rather than merely having a small set of distinct VLM states, explains the improvement—is not directly tested, and the absence of error bars makes the smaller gains difficult to assess. The paper's appended limitation statement is admirably candid about the single-backbone and single-robot scope.

major comments (3)
  1. [§3.2, Eq. (1)] The central claim is that local depth structure explains the gains, but the paper never varies the identity of the VLM layer matched to each decoder block. Since W_r(i) is always centered on the matched layer i, the ablations in Tables 5–7 cannot distinguish 'adjacent VLM layers are semantically complementary' from 'any small set of nearby-but-distinct VLM states helps.' Please add a permutation or reversal test: for example, route W'_r(i)={i-1,i+1,i+2}, random triples of layers, or a reversed layer ordering, and compare on LIBERO-Long and CALVIN-5. If these variants perform as well as the centered window, the depth-locality prior is not supported; if they degrade, the claim is strengthened.
  2. [§4.1, §4.2] No error bars or multiple-seed runs are reported. The LIBERO average gain is 1.6 points (97.3% to 98.9%) and the CALVIN five-task gain is 4.0 points (76.5% to 80.5%); without training-seed variance it is impossible to tell whether these improvements are stable. Please report mean±std over at least three seeds for the main LIRA vs. VLA-Adapter comparisons, or explicitly justify why the evaluation protocol has negligible variance.
  3. [§4.1] VLA-Adapter is called a 'controlled architectural baseline' and the paper states that the same objective and training recipe are used, but it is not explicitly stated whether the VLA-Adapter numbers were obtained by retraining in the same codebase or taken from the original publication. Please state this clearly. If the numbers are from the original paper, the 'controlled' claim is not supported and the comparison should be re-run under identical training conditions.
minor comments (5)
  1. [§4.3, Table 6] The routing topologies Flashback, Column-wise, and Q-Former are not defined in the main text; please define them in the text or refer to the appendix in the table caption.
  2. [§4.1] The sentence 'All ablations follow the same compact training and evaluation protocol unless stated otherwise' is vague; please specify which ablations deviate, such as the query-budget ablation in Table 5.
  3. [§4.5, Table 9] With 10 trials per task, differences of one success are well within sampling noise; please add per-task confidence intervals or explicitly frame these results as feasibility demonstrations.
  4. [§5] The limitation statement is appropriate, but the single-backbone and 10-trial real-world caveats should also appear in the abstract or introduction so that readers do not overinterpret the headline numbers.
  5. [§3.2, Eq. (1)] The boundary clipping of the window is described only in prose; please write the full definition of W_r(i) with explicit boundary handling.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: LIRA is an empirical architecture study with measured benchmark comparisons; no claimed derivation reduces to its inputs.

full rationale

LIRA is an empirical architecture paper; there is no claimed derivation from first principles that could reduce to its inputs. Equation (1) defines the routing window W_r(i) = {ℓ : |ℓ−i| ≤ r}, and Eq. (2) defines how PFBs consume the routed features; these are architectural definitions, not predictions fitted to benchmark outcomes. The controlled baseline VLA-Adapter is external work, the ablation comparisons are measured configurations, and the headline zero-shot LIBERO-Plus result is evaluated on a held-out benchmark rather than recovered from a fitted parameter. Choosing r=1 after ablations on LIBERO-Long and CALVIN-5 is standard model selection, not a circular prediction, and the paper does not rename a fitted value as a prediction. No load-bearing self-citation is present: the closest baseline is an independent prior system, and no uniqueness or correctness claim is imported from the authors' own prior work. The stated limitations in Section 5 (single compact backbone, one PFB decoder, one robot, 10 trials per task) concern external validity and statistical strength, not circularity. Accordingly, the score is 0 with no circular steps.

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

The central claim depends on two tuned hyperparameters (window radius and query budget) and on the hypothesis that adjacent VLM layers carry complementary information. No new physical entities are introduced; the architectural components are evaluated only within the paper.

free parameters (2)
  • Routing window radius r = 1
    Chosen by ablation on LIBERO-Long and CALVIN-5 (Table 7). The central claim depends on the three-layer centered window being optimal; this is a design choice fitted to the evaluation benchmarks.
  • LIRA Query token count M = 64
    Chosen by ablation on LIBERO-Long (Table 5). The paper shows that 256 tokens actually hurt LIRA, so the default budget is a tuned hyperparameter, not a universal setting.
assumptions (3)
  • domain assumption PFB index i matches VLM layer i (depth-aligned matching); each decoder block is associated with the VLM layer of the same index.
    Invoked in Section 3.1 and Eq. (1). The local window is centered on the matched layer i; if this one-to-one depth correspondence is arbitrary or not semantically meaningful, the centering has no principled basis.
  • ad hoc to paper Adjacent VLM layers contain complementary task-relevant information for action prediction within a radius of r=1 layer.
    Hypothesis stated in Section 3.2. This locality prior is the core design assumption and is not independently derived; it is validated only through the experiments.
  • domain assumption The supervised continuous-action objective and training recipe are identical for LIRA and the VLA-Adapter baseline, so performance differences isolate the routing mechanism.
    Stated in Section 3.3 and 4.1. If the baseline is not trained with identical data and hyperparameters, the comparison is confounded.
invented entities (2)
  • LIRA Query tokens
    purpose: Learnable query tokens appended to the VLM input; their layerwise features are routed to the action decoder as a conditioning interface.
    Mechanistically an architectural component, not a physical entity. It is an adaptation of VLA-Adapter's VLM-side query tokens; its utility is evaluated empirically within this paper, but there is no evidence outside the paper yet.
  • Depth-aligned local routing window
    purpose: Selects a neighborhood of VLM layers per PFB to aggregate query features.
    The new mechanism; supported only by the paper's own ablations and benchmarks, not by external replication.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LIRA: Local Cross-Layer Information Routing for Vision-Language-Action Decoding." pith.science (2026). https://pith.science/paper/5BNVWMZZ

@misc{pith2026260807596,
  author       = {Pith},
  title        = {Pith review of: LIRA: Local Cross-Layer Information Routing for Vision-Language-Action Decoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5BNVWMZZ}},
  note         = {Machine review of arXiv:2608.07596}
}
abstract

Vision-Language-Action (VLA) models transform representations from pretrained vision-language models (VLMs) into robot actions, yet the interface that routes intermediate VLM features into action decoders remains underexplored. Existing designs either expose only a narrow part of the representation hierarchy or rigidly match each decoder block to one VLM layer, restricting access to complementary task evidence across depths. We introduce LIRA, a local cross-layer action-conditioning mechanism that formulates VLM-to-action conditioning as depth-aware information routing. LIRA operates on task-token features and LIRA Query features derived from intermediate VLM states, then assigns each Parallel Fusion Block a depth-aligned local window centered on its corresponding VLM layer. Parallel Fusion Blocks aggregate neighboring LIRA Query features and integrate them with task-token features and proprioceptive inputs before action prediction. This routing interface leaves the backbone architecture, action decoder, and supervised training recipe unchanged. Across LIBERO, LIBERO-Plus, CALVIN ABC$\rightarrow$D, and real-world manipulation, LIRA improves the principal aggregate metrics over the VLA-Adapter baseline under the same 0.5B-parameter configuration. In zero-shot transfer to LIBERO-Plus, LIRA increases average success from 59.1% to 78.0%, an 18.9-point gain indicating improved robustness under controlled distribution shifts.

Figures

Figures reproduced from arXiv: 2608.07596 by the authors.

Figure 1
Figure 1. Zero-shot transfer to LIBERO-Plus. Left: represen [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of LIRA. The Prismatic-style VLM processes visual-language task tokens together with learnable LIRA Query tokens, while Action Query tokens initialize an action decoder composed of L Parallel Fusion Blocks (PFBs). Task-token features remain aligned with the corresponding VLM depth, while each PFB receives LIRA Query features aggregated from a centered local window of adjacent VLM layers. Each PFB uses Actio… view at source ↗
Figure 3
Figure 3. Conditioning-source, VLM-layer, and LIRA Query-token ablations on LIBERO-Long. Combining task-token features [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Real-world experimental platform with a Franka [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 42 canonical work pages

  1. [1]

    Robotics: Science and Systems, (RSS) , year=

    Rt-1: Robotics transformer for real-world control at scale , author=. Robotics: Science and Systems, (RSS) , year=

    work page

  2. [2]

    Conference on Robot Learning, (CoRL) , pages=

    Rt-2: Vision-language-action models transfer web knowledge to robotic control , author=. Conference on Robot Learning, (CoRL) , pages=. 2023 , organization=

    work page 2023

  3. [3]

    2024 IEEE International Conference on Robotics and Automation, (ICRA) , pages=

    Open x-embodiment: Robotic learning datasets and rt-x models: Open x-embodiment collaboration 0 , author=. 2024 IEEE International Conference on Robotics and Automation, (ICRA) , pages=. 2024 , organization=

    work page 2024

  4. [4]

    Robotics: Science and Systems, (RSS) , year=

    DROID: A large-scale in-the-wild robot manipulation dataset , author=. Robotics: Science and Systems, (RSS) , year=

    work page

  5. [5]

    Robotics: Science and Systems, (RSS) , year=

    Octo: An Open-Source Generalist Robot Policy , author=. Robotics: Science and Systems, (RSS) , year=

    work page

  6. [6]

    arXiv preprint arXiv:2502.19645 , year=

    Fine-tuning vision-language-action models: Optimizing speed and success , author=. arXiv preprint arXiv:2502.19645 , year=

    arXiv

  7. [7]

    Proceedings of The 8th Conference on Robot Learning, (CoRL) , series=

    OpenVLA: An Open-Source Vision-Language-Action Model , author=. Proceedings of The 8th Conference on Robot Learning, (CoRL) , series=

    work page

  8. [8]

    arXiv preprint arXiv:2410.24164 , year=

    pi\_0 : A Vision-Language-Action Flow Model for General Robot Control , author=. arXiv preprint arXiv:2410.24164 , year=

    arXiv

Show all 65 references
  1. [9]

    arXiv preprint arXiv:2501.09747 , year=

    Fast: Efficient action tokenization for vision-language-action models , author=. arXiv preprint arXiv:2501.09747 , year=

    arXiv

  2. [10]

    arXiv preprint arXiv:2501.15830 , year=

    Spatialvla: Exploring spatial representations for visual-language-action model , author=. arXiv preprint arXiv:2501.15830 , year=

    arXiv

  3. [11]

    arXiv preprint arXiv:2503.14734 , year=

    Gr00t n1: An open foundation model for generalist humanoid robots , author=. arXiv preprint arXiv:2503.14734 , year=

    arXiv

  4. [12]

    arXiv preprint arXiv:2504.19854 , year=

    Nora: A small open-sourced generalist vision language action model for embodied tasks , author=. arXiv preprint arXiv:2504.19854 , year=

    arXiv

  5. [13]

    arXiv preprint arXiv:2505.03912 , year=

    Openhelix: A short survey, empirical analysis, and open-source dual-system vla model for robotic manipulation , author=. arXiv preprint arXiv:2505.03912 , year=

    arXiv

  6. [14]

    GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Data , author=

  7. [15]

    arXiv preprint arXiv:2505.06111 , year=

    Univla: Learning to act anywhere with task-centric latent actions , author=. arXiv preprint arXiv:2505.06111 , year=

    arXiv

  8. [16]

    arXiv preprint arXiv:2506.19850 , year=

    Unified vision-language-action model , author=. arXiv preprint arXiv:2506.19850 , year=

    arXiv

  9. [17]

    arXiv preprint arXiv:2506.01844 , year=

    Smolvla: A vision-language-action model for affordable and efficient robotics , author=. arXiv preprint arXiv:2506.01844 , year=

    arXiv

  10. [18]

    Advances in Neural Information Processing Systems, (NeurIPS) , volume=

    Vla-os: Structuring and dissecting planning representations and paradigms in vision-language-action models , author=. Advances in Neural Information Processing Systems, (NeurIPS) , volume=

  11. [19]

    arXiv preprint arXiv:2506.21539 , year=

    Worldvla: Towards autoregressive action world model , author=. arXiv preprint arXiv:2506.21539 , year=

    arXiv

  12. [20]

    Advances in Neural Information Processing Systems, (NeurIPS) , volume=

    4d-vla: Spatiotemporal vision-language-action pretraining with cross-scene calibration , author=. Advances in Neural Information Processing Systems, (NeurIPS) , volume=

  13. [21]

    2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=

    PD-VLA: Accelerating Vision-Language-Action Model Integrated with Action Chunking via Parallel Decoding , author=. 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages=. 2025 , organization=

    2025

  14. [22]

    Advances in Neural Information Processing Systems, (NeurIPS) , volume=

    Thinkact: Vision-language-action reasoning via reinforced visual latent planning , author=. Advances in Neural Information Processing Systems, (NeurIPS) , volume=

  15. [23]

    arXiv preprint arXiv:2508.07917 , year=

    Molmoact: Action reasoning models that can reason in space , author=. arXiv preprint arXiv:2508.07917 , year=

    arXiv

  16. [24]

    arXiv preprint arXiv:2508.18269 , year=

    Flowvla: Visual chain of thought-based motion reasoning for vision-language-action models , author=. arXiv preprint arXiv:2508.18269 , year=

  17. [25]

    Proceedings of the Computer Vision and Pattern Recognition Conference, (CVPR) , pages=

    Cot-vla: Visual chain-of-thought reasoning for vision-language-action models , author=. Proceedings of the Computer Vision and Pattern Recognition Conference, (CVPR) , pages=

  18. [26]

    International Conference on Learning Representations, (ICLR) , volume=

    Tracevla: Visual trace prompting enhances spatial-temporal awareness for generalist robotic policies , author=. International Conference on Learning Representations, (ICLR) , volume=

  19. [27]

    Wang, Yihao and Ding, Pengxiang and Li, Lingxiao and Cui, Can and Ge, Zirui and Tong, Xinyang and Song, Wenxuan and Zhao, Han and Zhao, Wei and Hou, Pengxu and others , booktitle=

  20. [28]

    IEEE Robotics and Automation Letters , volume=

    Calvin: A benchmark for language-conditioned policy learning for long-horizon robot manipulation tasks , author=. IEEE Robotics and Automation Letters , volume=. 2022 , publisher=

    2022

  21. [29]

    Advances in Neural Information Processing Systems, (NeurIPS) , volume=

    Libero: Benchmarking knowledge transfer for lifelong robot learning , author=. Advances in Neural Information Processing Systems, (NeurIPS) , volume=

  22. [30]

    arXiv preprint arXiv:2412.15115 , year=

    Qwen2.5 Technical Report , author=. arXiv preprint arXiv:2412.15115 , year=

    arXiv

  23. [31]

    European Conference on Computer Vision , pages=

    End-to-End Object Detection with Transformers , author=. European Conference on Computer Vision , pages=

  24. [32]

    arXiv preprint arXiv:2107.14795 , year=

    Perceiver IO: A General Architecture for Structured Inputs and Outputs , author=. arXiv preprint arXiv:2107.14795 , year=

    arXiv

  25. [33]

    International Conference on Learning Representations, (ICLR) , year=

    Predictive Inverse Dynamics Models are Scalable Learners for Robotic Manipulation , author=. International Conference on Learning Representations, (ICLR) , year=

  26. [34]

    The International Journal of Robotics Research , volume=

    Diffusion policy: Visuomotor policy learning via action diffusion , author=. The International Journal of Robotics Research , volume=

  27. [35]

    Advances in Neural Information Processing Systems, (NeurIPS) , year=

    Attention Is All You Need , author=. Advances in Neural Information Processing Systems, (NeurIPS) , year=

  28. [36]

    Advances in Neural Information Processing Systems, (NeurIPS) , year=

    Decision Transformer: Reinforcement Learning via Sequence Modeling , author=. Advances in Neural Information Processing Systems, (NeurIPS) , year=

  29. [37]

    Advances in Neural Information Processing Systems, (NeurIPS) , year=

    Flamingo: a Visual Language Model for Few-Shot Learning , author=. Advances in Neural Information Processing Systems, (NeurIPS) , year=

  30. [38]

    Advances in Neural Information Processing Systems, (NeurIPS) , year=

    Object-Centric Learning with Slot Attention , author=. Advances in Neural Information Processing Systems, (NeurIPS) , year=

  31. [39]

    Transactions on Machine Learning Research, (TMLR) , year=

    A Generalist Agent , author=. Transactions on Machine Learning Research, (TMLR) , year=

  32. [40]

    Proceedings of the 40th International Conference on Machine Learning, (ICML) , year=

    PaLM-E: An Embodied Multimodal Language Model , author=. Proceedings of the 40th International Conference on Machine Learning, (ICML) , year=

  33. [41]

    Proceedings of the 40th International Conference on Machine Learning, (ICML) , year=

    BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models , author=. Proceedings of the 40th International Conference on Machine Learning, (ICML) , year=

  34. [42]

    International Conference on Learning Representations, (ICLR) , year=

    LoRA: Low-Rank Adaptation of Large Language Models , author=. International Conference on Learning Representations, (ICLR) , year=

  35. [43]

    arXiv preprint arXiv:2307.09288 , year=

    Llama 2: Open foundation and fine-tuned chat models , author=. arXiv preprint arXiv:2307.09288 , year=

    arXiv

  36. [44]

    Qiao, Guanren and Ouyang, Ruixiang and Xu, Sheng and Jin, Ruixing and Deng, Yueci and Tai, Yunxin and Jia, Kui and Liu, Guiliang , booktitle =

  37. [45]

    arXiv preprint arXiv:2408.15538 , year=

    Trafficgamer: Reliable and flexible traffic simulation for safety-critical scenarios with game-theoretic oracles , author=. arXiv preprint arXiv:2408.15538 , year=

  38. [46]

    Proceedings of the Forty-third International Conference on Machine Learning, (ICML) , year =

    DyGRO-VLA: Cross-Task Scaling of Vision Language Action Models via Dynamic Grouped Residual Optimization , author =. Proceedings of the Forty-third International Conference on Machine Learning, (ICML) , year =

  39. [47]

    arXiv preprint arXiv:2605.13403 , year=

    RotVLA: Rotational Latent Action for Vision-Language-Action Model , author=. arXiv preprint arXiv:2605.13403 , year=

    arXiv

  40. [48]

    arXiv preprint arXiv:2605.13382 , year=

    BlockVLA: Accelerating Autoregressive VLA via Block Diffusion Finetuning , author=. arXiv preprint arXiv:2605.13382 , year=

    arXiv

  41. [49]

    2026 , eprint=

    ActQuant: Sub-4-bit Action-Guided Quantization for Vision-Language-Action Models , author=. 2026 , eprint=

    2026

  42. [50]

    arXiv preprint arXiv:2605.17517 , year=

    AffordVLA: Injecting Affordance Representations into Vision-Language-Action Models via Implicit Feature Alignment , author=. arXiv preprint arXiv:2605.17517 , year=

    arXiv

  43. [51]

    arXiv preprint arXiv:2505.18719 , year=

    Vla-rl: Towards masterful and general robotic manipulation with scalable reinforcement learning , author=. arXiv preprint arXiv:2505.18719 , year=

    arXiv

  44. [52]

    arXiv preprint arXiv:2509.09674 , year=

    Simplevla-rl: Scaling vla training via reinforcement learning , author=. arXiv preprint arXiv:2509.09674 , year=

    arXiv

  45. [53]

    arXiv preprint arXiv:2505.17016 , year=

    Interactive post-training for vision-language-action models , author=. arXiv preprint arXiv:2505.17016 , year=

    arXiv

  46. [54]

    arXiv preprint arXiv:2605.02881 , year=

    MolmoAct2: Action Reasoning Models for Real-world Deployment , author=. arXiv preprint arXiv:2605.02881 , year=

    arXiv

  47. [55]

    and others , journal=

    Black, Kevin and Brown, Noah and Darpinian, James and Dhabalia, Karan and Driess, Danny and Esmail, Adnan and Equi, Michael and Finn, Chelsea and Fusai, Niccolo and Galliker, Manuel Y. and others , journal=. _

  48. [56]

    arXiv preprint arXiv:2509.04996 , year=

    FLOWER: Democratizing Generalist Robot Policies with Efficient Vision-Language-Action Flow Policies , author=. arXiv preprint arXiv:2509.04996 , year=

    arXiv

  49. [57]

    2026 , eprint=

    Beyond Accuracy: Measuring Bias Acknowledgment in Chain-of-Thought Reasoning for Responsible AI Evaluation , author=. 2026 , eprint=

    2026

  50. [58]

    arXiv preprint arXiv:2606.26432 , year=

    Embedding Foundation Model Predictions in Discrete-Choice Models with Structural Guarantees , author=. arXiv preprint arXiv:2606.26432 , year=

    arXiv

  51. [59]

    ArXiv , year=

    Is Your Driving World Model an All-Around Player? , author=. ArXiv , year=

  52. [60]

    European Conference on Computer Vision, (ECCV) , pages =

    Optical Flow Training under Limited Label Budget via Active Learning , author =. European Conference on Computer Vision, (ECCV) , pages =. 2022 , publisher =

    2022

  53. [61]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    LIBERO-Plus: A Progressive Robustness Benchmark for Vision-Language-Action Models , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  54. [62]

    ArXiv , year=

    SignBot: Learning Human-to-Humanoid Sign Language Interaction , author=. ArXiv , year=

  55. [63]

    Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages =

    Deep Layer Aggregation , author =. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages =

  56. [64]

    Proceedings of the European Conference on Computer Vision , pages =

    Sparsely Aggregated Convolutional Networks , author =. Proceedings of the European Conference on Computer Vision , pages =

  57. [65]

    2026 , eprint =

    AffordanceVLA: A Vision-Language-Action Model Empowering Action Generation through Affordance-Aware Understanding , author =. 2026 , eprint =

    2026

Pith tools

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