Pith. sign in

REVIEW 3 major objections 5 minor 20 references

GUARD: Grounding Uncertainty and Ablation-Based Risk Detection for Diffusion-Based VLAs

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read GUARD detects failures in diffusion-based robot policies by erasing the visual and language evidence an action most depends on and measuring how much the action changes.

desk verdict Genuinely new angle on VLA failure detection, but the headline sensitivity signal comes from a synthetic one-step probe and is never validated against the actual action-level effect. read the letter →

arxiv 2608.04510 v1 pith:WVT5SK6O submitted 2026-08-05 cs.RO cs.AI

classification cs.ROcs.AI
keywords vision-language-actionmodelsfailuredetectionKV-cacheablationcounterfactualinterventiondiffusionpoliciesgroundingrobotmanipulationautonomousdriving
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

GUARD asks a different question from most failure detectors: not whether an observation looks novel or whether actions wobble, but whether the generated action actually depends on the visual and language evidence that defines the task. It answers this at test time without retraining the policy by finding the most influential token-indexed entries in the VLM's final KV cache, replacing the top fraction in each modality with that modality's mean, and measuring how much the denoised action changes. A large change means the action is grounded in task evidence; a small change means the policy is producing a plausible-looking action while ignoring what it should be doing. Across five held-out-task settings with three policies and four benchmarks, this signal detects unseen-task failures with an average ROC-AUC of 88.84%, beating the strongest runtime monitor and staying within 0.19 points of the best seen-task average.

What carries the argument

The central object is the token-indexed final key–value (KV) cache of the vision-language model, the conditioning memory passed to the diffusion action head. GUARD's machinery is a saliency backward pass computing $g_i = \|\partial \Phi(A_t)/\partial z_i\|_2$ with $\Phi(A_t)=\|\mathrm{vec}(A_t)\|_2$, modality-balanced top-$K$ selection of the most salient visual and language positions, mean-ablated counterfactual caches $\bar{z}_m = \frac{1}{|M_m|}\sum_{i\in M_m} z_i$, and a batched single-step denoising probe that evaluates original and ablated caches from the same noisy action input. From these probes it derives sensitivity $S_t$, modality bias $B_t$, attention entropy $E_t$, grounding efficiency $G_t=S_t/(E_t+\varepsilon)$, an entropy-adjusted adaptive threshold $\theta_t$, and the low-sensitivity indicator $b_t$; a lightweight temporal classifier and functional conformal prediction convert the seven-dimensional stream into online alarms. This machinery isolates the action head's causal dependence on task evidence while leaving the pretrained policy frozen.

What would settle it

Collect failed and successful rollouts on held-out tasks with human-annotated task-relevant objects or language tokens, then check whether GUARD's low-sensitivity events coincide with the action ignoring the annotated evidence: a successful rollout that keeps low sensitivity all the way through, or a failure with uniformly high sensitivity, would break the claimed link between counterfactual sensitivity and grounding.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that counterfactual dependence is a transferable failure signal: a reliable action chunk must remain functionally coupled to the salient multimodal evidence specifying the task, so when the diffusion action head becomes insensitive to the visual and language tokens that most influence it, the rollout is heading toward failure even if the action looks smooth. GUARD measures that coupling by computing gradient saliency of the action-norm objective with respect to the final VLM KV-cache entries, selecting the top 10% of salient entries within each modality, constructing counterfactual caches in which those entries are replaced by modality means, and comparing the denoising responses from the same noisy action state. Low sensitivity to the joint ablation is treated as a low-sensitivity event and, after online calibration and temporal classification, produces an alarm. The paper reports that this grounding probe transfers across policies, tasks, embodiments, and domains, with best unseen-task ROC-AUC on four of five settings and second on the remaining one.

Load-bearing premise

The signal stands on the premise that the gradient saliency of the action norm with respect to final KV-cache entries singles out evidence the action is actually grounded in, and that replacing the top 10% of those entries per modality with the modality mean is a faithful counterfactual; the paper does not validate saliency against a ground-truth notion of task-relevant evidence.

Editorial extensions

If this is right

  • A VLA's failure risk can be scored during deployment from the policy's own frozen computation, with only seen-task labels for classifier training and no unseen-task failure labels.
  • The grounding signal transfers across architectures (Pi0, SmolVLA, Alpamayo-1.5), embodiments, and domains, so one detector recipe replaces task-specific monitors.
  • Focused counterfactual intervention matters: ablating the top 10% salient entries per modality beats random or least-salient selection by 8.10 and 9.84 average unseen-task ROC-AUC points, and ablating more than 10% degrades performance.
  • Online alarms come early enough for intervention: on the MetaWorld setting GUARD fires at roughly 20% of rollout progress while keeping the best balanced accuracy.

Reading between the lines

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

  • If low sensitivity really tracks weak grounding, the same probe could serve as an interpretability tool: it not only raises alarms but names which modality (vision or language) the action stopped listening to, via the bias ratio $B_t$.
  • The saliency-faithfulness assumption is the place to stress-test next; a direct comparison of GUARD's salient tokens against human-annotated task-relevant regions or against causal interventions on known-critical objects would show whether the signal measures grounding or merely sensitivity to any perturbation.
  • Because the mechanism is architectural (KV-cache conditioning) rather than task-specific, a natural extension is to autoregressive VLAs by applying the same ablation to the tokens the action decoder attends to, although the paper explicitly leaves that outside its current scope.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes GUARD, a test-time failure detection method for diffusion/flow-based vision-language-action (VLA) policies. GUARD computes gradient saliency of the action-norm objective with respect to token-indexed entries in the final VLM KV cache, constructs counterfactual caches by replacing the top saliency fraction per modality with modality-mean entries, and evaluates a single denoising response under the original and ablated caches from a common synthetic noisy input. From these responses it derives diagnostics (sensitivity, modality bias, attention entropy, grounding efficiency, an adaptive threshold, calibration status, and a low-sensitivity-event indicator), which are fed to a lightweight temporal classifier trained with a from-first-LSE target and thresholded via functional conformal prediction. Experiments on LIBERO, SimplerEnv, MetaWorld, and PhysicalAI-AV with Pi0, SmolVLA, and Alpamayo report an average unseen-task ROC-AUC of 88.84%, best in four of five settings, together with ablations on selection mode, ablation fraction, feature importance, and KV-cache versus embedding targets.

Significance. If the central proxy is valid, GUARD contributes a genuinely different failure signal: instead of embedding distance, action variability, or representation likelihood, it probes the functional dependence of action generation on multimodal evidence through counterfactual KV-cache intervention. The empirical scope is substantial—three policies, four benchmarks, five held-out settings, and several carefully designed ablations (top/random/least selection, ablation percentage, leave-one-feature-out, KV-cache vs embedding target)—which strengthens the transferability claim. The appendix is appropriately cautious in stating that the mathematical properties are intuition rather than a proof of detection. However, the main quantitative results rest on a sensitivity signal whose relation to the final generated action is not validated, and some aspects of the LSE-based training target and calibration mechanism need clarification before the headline claims are fully supported.

major comments (3)
  1. [§3.3, Eq. (5)–(6); Appendix A.4] Eq. (6) defines S_t from a single denoising response at the synthetic input \tilde n_{K-1} = (1-\lambda_p) A_t + \lambda_p \xi, not from the actual penultimate state of the K-step generation, and Property 4 in Appendix A.4 expands only the one-step response. The paper therefore does not establish that S_t measures the functional coupling of the final generated action to the ablated KV evidence; Table 2 validates the selection rule (top vs random vs least) but not the action-level claim. Because S_t feeds θ_t, b_t, G_t, and the LSE training target (Eqs. (10)–(14)), this gap propagates to every reported ROC-AUC. I request an experiment comparing S_t with the full multi-step action difference under the same ablation, or applying the ablation during the actual rollout, and, if possible, a sanity check of the saliency ranking against a causal effect estimate on the final action.
  2. [§3.6, Eq. (14); Appendix C.2] The timestep targets y_t are constructed from the first LSE, which is a threshold on S_t itself, so the classifier is trained to predict a quantity that is a deterministic function of its own input features. Appendix C.2 shows that removing the calibration-status feature c_t reduces unseen ROC-AUC by 25.3 points on MetaWorld (98.8 to 73.5), the largest drop of any feature; c_t is a calibration-window indicator that is shared across episodes and marks a fixed rollout phase. This raises the concern that part of the reported transfer gain comes from the classifier exploiting the calibration window rather than from the grounding diagnostics. Please report ROC-AUC with c_t and b_t removed jointly, and analyze whether the learned alarms are actually triggered by S_t dynamics rather than by the phase-dependent masking.
  3. [§3.6, Eq. (14) and §B (Eqs. (44)–(45))] The first-LSE target is defined over all timesteps, but the threshold θ_t in Eq. (11) depends on the episode-level constant C computed from the calibration window in Eq. (10), which runs from C_start to C_end (Table 14). For timesteps before C_end, θ_t is not computable at inference time without future information. The paper suppresses alarms during calibration, but the training target does not explicitly restrict t_LSE to the post-calibration period, so the classifier may be trained to fire on timesteps at which the required threshold is not yet available online. Please either restrict t_LSE to t ≥ C_end, or report the distribution of t_LSE relative to C_end and show that the current definition does not introduce training/inference inconsistency.
minor comments (5)
  1. [Table 1] The FIPER row contains only eight entries and no Alpamayo values, while other rows have ten; please clarify the missing entries and whether FIPER was evaluated on all five settings.
  2. [Table 14] Table 14 refers to λ_p as defined in Eq. (8) and γ as defined in Eqs. (13)–(14), but λ_p is introduced in Eq. (5) and γ in Eqs. (10)–(11); the cross-references should be corrected.
  3. [Eq. (15)] Eq. (15) is typeset incompletely: the summation index and limits are missing, making the definition of q_mean ambiguous; please fix the equation.
  4. [Figure captions] The parenthetical instructions "(Please zoom in 500%)" in Figures 1 and 3 are inappropriate for a journal submission and should be removed.
  5. [Eq. (8)] The definition of \bar P_{q,i} as a head-averaged attention probability is stated in prose, but no explicit averaging formula is provided; please specify the head count and aggregation explicitly.

Circularity Check

1 steps flagged · score 4.0 of 10

The temporal classifier's training target is constructed from its own low-sensitivity-event input feature; the main ROC-AUC result is still externally validated against true labels.

  1. self definitional [Section 3.6, Eq. (14); Section 3.4, Eq. (12) and definition of b_t]
    "For a failed episode, we identify the first LSE as tLSE = min{t: St < θt}. The timestep targets are then yt = 0, t<tLSE, 1, t≥tLSE. (14) ... xt = [St, Bt, Et, Gt, θt, ct, bt] ∈ R^7 ... bt = 1[St < θt]."

    The supervised target y_t for failed episodes is defined as the step function that switches permanently at the first timestep where b_t=1, and b_t = 1[S_t < θ_t] is one of the seven input features in x_t consumed by the same classifier. Thus the training label is a temporal prefix-closure of an input feature: predicting y_t reduces, by construction, to detecting or smoothing the detector's own low-sensitivity alarm rather than predicting an externally supplied timestep-level failure label. The paper calls this a 'binary failure-classification task', but the labels are not independent of the features.

full rationale

GUARD's primary diagnostic S_t is computed from the frozen policy via gradient saliency and counterfactual KV-cache ablation, not fitted to success/failure labels, and the headline ROC-AUC numbers are evaluated against true rollout labels on held-out tasks. That central comparison is not circular. The one concrete definitional reduction is the from_first_lse training target in Eq. (14), which is constructed from b_t = 1[S_t < θ_t], a feature already present in the classifier input x_t; the classifier is therefore trained to predict an internally derived event rather than an independent label. This does not fully invalidate the method, because the final separability result depends on whether the LSE event correlates with true failure, which is an external empirical question. No load-bearing self-citation chain, imported uniqueness theorem, or ansatz-smuggling-via-citation was found.

Assumptions & free parameters 6 free parameters · 5 assumptions · 1 invented entities

The method introduces no new physical or architectural entities, but it relies on a set of hand-chosen diagnostic parameters (rho, lambda_p, gamma, mu), per-benchmark calibration windows, and an internally defined low-sensitivity event. The gradient-saliency grounding assumption is the key unvalidated domain assumption. The conformal calibration also relies on an exchangeability assumption that is acknowledged as violated for held-out tasks.

free parameters (6)
  • rho_vis and rho_lang (saliency fractions, 0.1) = 0.1
    The fraction of cache entries ablated within each modality is fixed at 10% after comparing 10%, 30%, and 50% in Table 2. It is a hand-chosen hyperparameter that materially affects the results.
  • lambda_p (probe-noise mixing scale) = 0.05
    Mixing scale for the probe noise in Eq. (8) is fixed across all experiments; no sensitivity analysis is reported.
  • gamma (entropy-correction exponent) = 2.0
    Exponent controlling the entropy correction in the adaptive threshold and calibration constant is fixed at 2.0 across all settings.
  • mu (safety margin) = 0.8
    Multiplier for the adaptive threshold theta_t in Eq. (11); chosen by hand and fixed across experiments.
  • C_start and C_end (calibration window) = steps 25-35 for LIBERO, 4-10 for SimplerEnv, 0-3 for Alpamayo
    The episode-specific calibration window is set differently per benchmark, which means the episode baseline is fitted using task-specific knowledge.
  • Temporal classifier hyperparameters = best per benchmark (Table 13)
    Model type, hidden dimension, dropout, learning rate, weight decay, batch size, and window configuration are selected per benchmark via hyperparameter search.
assumptions (5)
  • domain assumption The mapping from the final conditioning KV cache to the action objective is differentiable in a neighborhood of Z_t
    Invoked in Appendix A.1 to justify the first-order influence bound for the saliency score.
  • ad hoc to paper The gradient of the action norm with respect to KV-cache entries identifies the tokens the action is grounded in
    The central hypothesis of the paper; stated in Sections 3.2-3.3 and Figure 1, not validated against ground-truth task relevance.
  • ad hoc to paper Mean ablation of the top saliency entries is a conservative counterfactual that removes position-specific information without corrupting the conditioning
    Assumed in Section 3.3 and Appendix A.3; the least-squares property does not imply that the ablated cache is a faithful counterfactual for the action head.
  • domain assumption Exchangeability of the successful-rollout conformal calibration set with future successful rollouts
    Assumed in Appendix B to target the finite-sample false-crossing rate, with the paper acknowledging that the assumption is violated for held-out tasks.
  • ad hoc to paper The first low-sensitivity event (St < theta_t) is a meaningful failure-onset time for training the temporal classifier
    The from_first_lse training target in Eq. (14) depends on this internally defined event, which is calibrated using the fitted theta_t.
invented entities (1)
  • Low-sensitivity event (LSE)
    purpose: Defines the onset of weakly grounded action generation and provides the temporal training target for the failure classifier
    The LSE is defined internally via the adaptive threshold theta_t, which depends on the fitted calibration constant and the chosen safety margin, so it has no falsifiable handle outside the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GUARD: Grounding Uncertainty and Ablation-Based Risk Detection for Diffusion-Based VLAs." pith.science (2026). https://pith.science/paper/WVT5SK6O

@misc{pith2026260804510,
  author       = {Pith},
  title        = {Pith review of: GUARD: Grounding Uncertainty and Ablation-Based Risk Detection for Diffusion-Based VLAs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WVT5SK6O}},
  note         = {Machine review of arXiv:2608.04510}
}
read the original abstract

Diffusion-based vision-language-action (VLA) policies can generate plausible actions even when their predictions are weakly grounded in the visual and language evidence defining the task. We introduce GUARD, a test-time failure detection method that measures this grounding without modifying the pretrained policy. GUARD estimates the influence of token-indexed entries in the final vision-language model key-value (KV) cache, constructs counterfactual caches by ablating salient KV entries, and compares their denoising responses with the original conditioning. Based on the comparison, we derive GUARD diagnostic stream including sensitivity, attention entropy, modality bias, and grounding efficiency, which are calibrated online and processed by a lightweight temporal classifier. We evaluate GUARD under task-held-out splits across five policy-benchmark settings, using Pi0, SmolVLA, and Alpamayo-1.5 on LIBERO, SimplerEnv, MetaWorld, and PhysicalAI-AV. GUARD achieves the best ROC-AUC on four of five unseen-task settings and ranks second on the remaining setting, improving the average unseen-task ROC-AUC by 5.73 percentage points over the strongest competing runtime monitor while remaining within 0.19 points of the best seen-task average. These results show that directly probing action-head dependence on multimodal evidence provides a transferable failure signal across policies, tasks, embodiments, and domains.

Figures

Figures reproduced from arXiv: 2608.04510 by the authors.

Figure 1
Figure 1. Qualitative comparison of saliency across failed [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. GUARD diagnostic creation with VLA rollout, [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Thresholded online failure detection across five policy–benchmark settings for unseen tasks. Blue and green bars [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Single-action cross-attention profiles over conditioning positions for action-query positions 0, 25, and 49 across ten [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Entropy–sensitivity phase plot for surgical saliency diagnostics. Each point corresponds to one action-chunk generation [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: Diagnostic trajectories for failed Pi0–LIBERO rollouts aligned at their first low-sensitivity event (LSE). Relative index [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Online failure-classifier module. At each timestep, the most recent causal window of normalized GUARD diagnostics [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Leave-one-feature-out ROC-AUC analysis of the seven GUARD diagnostics. Each variant retrains the classifier after [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]
Figure 9
Figure 9. Figure 9: Saliency-target ablation comparing interventions on the input token embeddings and the final token-indexed KV cache. [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]
Figure 10
Figure 10. Figure 10: Effect of the functional conformal miscoverage level [PITH_FULL_IMAGE:figures/full_fig_p019_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 4 canonical work pages

  1. [4]

    Fawcett,T.2006

    Failure Prediction with Statistical Guarantees for Vision- Based Robot Control.arXiv preprint arXiv:2202.05894. Fawcett,T.2006. AnIntroductiontoROCAnalysis.Pattern Recognition Letters, 27(8): 861–874. Fong, R. C.; and Vedaldi, A

  2. [8]

    Li, X.; Hsu, K.; Gu, J.; Mees, O.; Pertsch, K.; Walke, H

    CogACT: A Foundational Vision-Language-Action Model for Syner- gizingCognitionandActioninRoboticManipulation.arXiv preprint arXiv:2411.19650. Li, X.; Hsu, K.; Gu, J.; Mees, O.; Pertsch, K.; Walke, H. R.; Fu, C.; Lunawat, I.; Sieh, I.; Kirmani, S.; Levine, S.; Wu, J.; Finn, C.; Su, H.; Vuong, Q.; and Xiao, T

  3. [10]

    Open X-Embodiment Collaboration; O’Neill, A.; Rehman, A.; Gupta, A.; Maddukuri, A.; et al

    Octo:AnOpen-SourceGeneralistRobotPolicy.arXiv preprint arXiv:2405.12213. Open X-Embodiment Collaboration; O’Neill, A.; Rehman, A.; Gupta, A.; Maddukuri, A.; et al

  4. [11]

    arXiv preprint arXiv:2310.08864

    Open X- Embodiment: Robotic Learning Datasets and RT-XModels. arXiv preprint arXiv:2310.08864. Papernot,N.;andMcDaniel,P.2018. Deepk-NearestNeigh- bors: Towards Confident, Interpretable and Robust Deep Learning.arXiv preprint arXiv:1803.04765. Qu, D.; Song, H.; Chen, Q.; Yao, Y.; Ye, X.; Ding, Y.; Wang, Z.; Gu, J.; Zhao, B.; Wang, D.; and Li, X

  5. [12]

    Römer,R.;Kobras,A.;Worbis,L.;andSchoellig,A.P.2025

    SpatialVLA: Exploring Spatial Representations for Vision- Language-ActionModels.arXivpreprintarXiv:2501.15830. Römer,R.;Kobras,A.;Worbis,L.;andSchoellig,A.P.2025. FailurePredictionatRuntimeforGenerativeRobotPolicies. arXiv preprint arXiv:2510.09459. Römer, R.; Seeliger, M.; Liu, S.; Sturgis, B.; Bagatella, M.; Marta, D.; Krause, A.; and Schoellig, A. P

  6. [14]

    Simonyan, K.; Vedaldi, A.; and Zisserman, A

    SmolVLA: A Vision-Language- Action Model for Affordable and Efficient Robotics.arXiv preprint arXiv:2506.01844. Simonyan, K.; Vedaldi, A.; and Zisserman, A

  7. [17]

    Xu, C.; Nguyen, T

    Alpamayo- R1: Bridging Reasoning and Action Prediction for General- izableAutonomousDrivingintheLongTail.arXivpreprint arXiv:2511.00088. Xu, C.; Nguyen, T. K.; Dixon, E.; Rodriguez, C.; Miller, P.; Lee, R.; Shah, P.; Ambrus, R.; Nishimura, H.; and Itkina, M

  8. [18]

    Yu, T.; Quillen, D.; He, Z.; Julian, R.; Hausman, K.; Finn, C.; and Levine, S

    Can We Detect Failures Without Failure Data? Uncertainty-Aware Runtime Failure Detection for Imitation Learning Policies.arXiv preprint arXiv:2503.08558. Yu, T.; Quillen, D.; He, Z.; Julian, R.; Hausman, K.; Finn, C.; and Levine, S

Show all 20 references
  1. [19]

    InProceedings of the Conference on Robot Learning, volume 100 ofProceedings of Machine Learning Research, 1094–1100

    Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learn- ing. InProceedings of the Conference on Robot Learning, volume 100 ofProceedings of Machine Learning Research, 1094–1100. Zhao,T.Z.;Kumar,V.;Levine,S.;andFinn,C.2023. Learn- ing Fine-Grained Bi...

  2. [20]

    InProceedings of The 7th Conference on Robot Learning, volume 229 ofProceedings of Machine Learning Research, 2165–2183

    RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control. InProceedings of The 7th Conference on Robot Learning, volume 229 ofProceedings of Machine Learning Research, 2165–2183. Appendix A Mathematical Intuition and Properties of GUARD Diagnostics This se...

  3. [2013]

    Sundararajan, M.; Taly, A.; and Yan, Q

    Deep Inside Convolutional Networks: Visualising Image Clas- sification Models and Saliency Maps.arXiv preprint arXiv:1312.6034. Sundararajan, M.; Taly, A.; and Yan, Q

  4. [2017]

    InProceedings of the 34th International Conference on Machine Learning, volume 70 ofProceedings of Machine Learning Research, 3319–3328

    Axiomatic Attribution for Deep Networks. InProceedings of the 34th International Conference on Machine Learning, volume 70 ofProceedings of Machine Learning Research, 3319–3328. Vovk,V.;Gammerman,A.;andShafer,G.2005.Algorithmic Learning in a Random World. Springer. Walke, H. R...

  5. [2019]

    InProceedings of the 2019 Conference of the North AmericanChapteroftheAssociationforComputationalLin- guistics: Human Language Technologies, 3543–3556

    Attention is not Expla- nation. InProceedings of the 2019 Conference of the North AmericanChapteroftheAssociationforComputationalLin- guistics: Human Language Technologies, 3543–3556. Khazatsky,A.;Pertsch,K.;Nair,S.;Balakrishna,A.;Dasari, S.; Karamcheti, S.; Nasiriany, S.; Sri...

  6. [2020]

    InProceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 4190–4197

    Quantifying Attention Flow in Transformers. InProceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 4190–4197. Agia, C.; Sinha, R.; Yang, J.; Cao, Z.-a.; Antonova, R.; Pavone,M.;andBohg,J.2025. UnpackingFailureModesof GenerativePolicies:Runti...

  7. [2021]

    Black,K.;Brown,N.;Driess,D.;Esmail,A.;Equi,M.;Finn, C.; Fusai, N.; Groom, L.; Hausman, K.; Ichter, B.; et al

    A Gentle Intro- duction to Conformal Prediction and Distribution-Free Un- certainty Quantification.arXiv preprint arXiv:2107.07511. Black,K.;Brown,N.;Driess,D.;Esmail,A.;Equi,M.;Finn, C.; Fusai, N.; Groom, L.; Hausman, K.; Ichter, B.; et al. 2024.π 0: A Vision-Language-Action ...

  8. [2022]

    Chi, C.; Xu, Z.; Feng, S.; Cousineau, E.; Du, Y.; Burchfiel, B.;Tedrake,R.;andSong,S.2023

    RT-1: Robotics Transformer for Real-World Control at Scale.arXiv preprint arXiv:2212.06817. Chi, C.; Xu, Z.; Feng, S.; Cousineau, E.; Du, Y.; Burchfiel, B.;Tedrake,R.;andSong,S.2023. DiffusionPolicy:Visuo- motor Policy Learning via Action Diffusion.arXiv preprint arXiv:2303.04...

  9. [2023]

    InAdvances in Neural In- formation Processing Systems, volume 36, 44776–44791

    LIBERO: Benchmarking Knowledge Trans- fer in Lifelong Robot Learning. InAdvances in Neural In- formation Processing Systems, volume 36, 44776–44791. Liu, H.; Zhang, Y.; Betala, V.; Zhang, E.; Liu, J.; Ding, C.; andZhu,Y.2024a.Multi-TaskInteractiveRobotFleetLearn- ingwithVisual...

  10. [2024]

    Kim,M.J.;Finn,C.;andLiang,P.2025.Fine-TuningVision- Language-ActionModels:OptimizingSpeedandSuccess.In Robotics: Science and Systems

    DROID: A Large-Scale In-the-Wild Robot Manipulation Dataset.arXiv preprint arXiv:2403.12945. Kim,M.J.;Finn,C.;andLiang,P.2025.Fine-TuningVision- Language-ActionModels:OptimizingSpeedandSuccess.In Robotics: Science and Systems. Kim, M. J.; Pertsch, K.; Karamcheti, S.; Xiao, T.;...

  11. [2025]

    Hendrycks, D.; and Gimpel, K

    SAFE: Multitask Failure De- tection for Vision-Language-Action Models.arXiv preprint arXiv:2506.09937. Hendrycks, D.; and Gimpel, K

  12. [2026]

    Serrano, S.; and Smith, N

    Un- certainty Quantification for Flow-Based Vision-Language- Action Models.arXiv preprint arXiv:2606.18043. Serrano, S.; and Smith, N. A

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.