EdgeVLA: Efficient Vision-Language-Action Models

Aaron Xie; Benjamin Bolte; Jingxiang Mo; Matthew Freed; Pawe{\l} Budzianowski; Tomasz M{\l}oduchowski; Viraj Tipnis; Wesley Maa; Winston Hsiao

REVIEW 4 major objections 3 minor 3 cited by

EdgeVLA: Efficient Vision-Language-Action Models

T0 review · 4 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read EdgeVLA claims a 1B-parameter vision-language-action model can match OpenVLA's training characteristics while predicting end-effector poses in one shot instead of autoregressively.

desk verdict Plausible architecture and honest early results, but the headline speedup is internally inconsistent and confounded, so the efficiency claim is not credible yet. read the letter →

arxiv 2507.14049 v1 pith:JIAELROS submitted 2025-07-18 cs.RO cs.CL

Pawe{l} Budzianowski , Wesley Maa , Matthew Freed , Jingxiang Mo , Winston Hsiao , Aaron Xie , Tomasz M{l}oduchowski , Viraj Tipnis

show 1 more author

Benjamin Bolte

This is my paper · ORCID

classification cs.ROcs.CL

keywords vision-language-actionmodelsedgedeploymentautoregressivedecodingremovaljointactionpredictionsmalllanguagerobotmanipulationOpenVLAcomparisonefficientinference

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

EdgeVLA (EVLA) is a roughly 1-billion-parameter vision-language-action model built from a 0.5B Qwen2 language model and two visual encoders, SigLIP and DINOv2. The paper's central claim is that predicting the end-effector position in one joint pass, instead of autoregressively token by token, does not hurt the model's ability to learn robot control, while removing the per-token decoding loop. On the BridgeData V2 and OpenX training suites, EVLA's loss and action-token accuracy curves track those of OpenVLA, a 7.5B-parameter model, with roughly 7x faster training iterations. The authors report inference dropping from 20 ms to 5 ms and memory from 16 GB to 4 GB on an A100 GPU, and argue this makes real-time VLA control feasible on edge devices. If the claim holds, learning-based manipulation policies could run on inexpensive, power-constrained robots instead of requiring large stationary GPU clusters.

What carries the argument

The load-bearing mechanism is the removal of the causal mask from the language model during action fine-tuning, turning end-effector prediction from a sequential decoding problem into a single joint prediction. Around that sits the small-model architecture: a 0.5B Qwen2 language backbone, a two-part visual encoder made of SigLIP and DINOv2, and a learned projection layer mapping visual features into the language model's token space. The design follows the OpenVLA-style fusion of two visual encoders and the Prismatic-style VLM pretraining recipe, then replaces autoregressive action-token generation with one-shot output of the full end-effector pose. This is what converts the 7.5B-parameter baseline's decoding loop into an edge-compatible forward pass.

What would settle it

Run the trained EVLA and OpenVLA policies on the same physical manipulation benchmark or a standardized simulation benchmark and compare task success rates: if EVLA finishes clearly below OpenVLA despite similar training curves, the central claim falls. A purely arithmetic check would also settle the speed claim: benchmark both models with identical attention kernels to determine whether the inference gap is 4x, 6x, or 7x.

Watch

Extended reading notes

Core claim

On the paper's own terms, EVLA establishes that a small language model can carry a vision-language-action policy once the autoregressive constraint on action generation is removed. The architecture fuses Qwen2-0.5B with SigLIP and DINOv2 visual encoders, trains first as a VLM on 1.2M image-text pairs following the Prismatic recipe, then fine-tunes on about 1M OpenX manipulation examples with the causal mask removed so the whole end-effector position is emitted in a single forward pass. The reported result is that these training curves are comparable to OpenVLA's on both BridgeData V2 and OpenX, while the smaller model and joint decoding yield a roughly sevenfold reduction in parameter count, 7x faster training iterations, 4x faster inference (20 ms to 5 ms), and 4x lower memory use (16 GB to 4 GB). The paper presents these as early results, with physical-robot evaluation across embodiments named as the required next step.

Load-bearing premise

The paper's case rests on treating early training loss and action-token accuracy as a stand-in for real robot manipulation skill, since no physical-robot or downstream success-rate evaluation is reported.

Editorial extensions

If this is right

Real-time closed-loop control with a VLA becomes possible on edge hardware: 5 ms inference and 4 GB memory fit a single GPU or high-end embedded device, where 20 ms and 16 GB did not.
VLA training becomes much cheaper to reproduce, since 7x faster training iterations and larger feasible batch sizes let the same OpenX data be processed at a fraction of the compute.
The autoregressive assumption can be dropped specifically for action output without sacrificing training-characteristic parity, separating control decoding from language generation in VLA design.
EVLA's reported efficiency numbers imply that mobile manipulators and humanoids could run learned visuomotor policies without a tether to a stationary compute server.

Reading between the lines

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

The paper leaves implicit that the speed advantage of joint prediction should grow with the number of action dimensions, since autoregressive decoding costs scale with each output token while joint decoding is nearly constant; this predicts even larger gains for whole-body or high-DoF control.
Because EVLA is compared against an OpenVLA baseline that uses FlashAttention-2 while EVLA runs in eager mode, the true speed gap under matched kernels is unresolved; measuring both with the same attention implementation would tighten the reported 4x, 6x, and 7x numbers.
The same architecture change could be applied to other VLA backbones, not just small language models; the paper's argument implies that any VLM trained with a causal action head could switch to joint prediction and inherit some of the speedup.
A natural testable extension is the authors' stated next step: evaluating EVLA on at least two humanoid platforms for few-shot manipulation, which would convert the training-curve claim into an embodiment-level claim.

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.

Desk Editor's Note

read the letter

Colleague, here's my take on EdgeVLA. The core idea is worth a look: predict the end-effector position jointly instead of autoregressively, using a small 0.5B Qwen2 backbone with SigLIP and DINOv2 as vision encoders. That combination is a natural extension of OpenVLA, and the authors ship code and checkpoints, which is good for the community. They also label the work as early results and admit they could not reproduce OpenVLA's full training. That honesty buys them some goodwill.

The soft spot is where the reader landed: the speedup claim is not internally consistent. The abstract and intro say 7x, Section III.B says six-times, and Table I's 20ms-to-5ms is 4x. That is not rounding error; it is a factor of two on the headline number. The Table I comparison is also confounded: OpenVLA runs with flash-attention2, EVLA in eager mode, and the backbones differ by 7.5B vs 1B. You cannot cleanly attribute the efficiency gain to the non-autoregressive change. The training comparison relies on early loss and action-token accuracy curves, with no final convergence data, no variance, and no physical robot evaluation. The authors themselves flag the physical-robot step as future work, but the abstract's claim of "comparable training performance" goes beyond what the figures support.

I would not call this a bad paper. The central idea is plausible and testable, and the limitations are honestly stated. The issues are fixable: reconcile the speedup numbers, run a controlled inference benchmark with matched kernel implementations and model sizes, and evaluate the trained policy on at least a few physical embodiments. I would send it to peer review rather than desk reject, because a careful reviewer can push the authors to tighten the claims and the community would benefit from a clearer answer either way. I would not cite it yet. Reading group: maybe, if you want a case study in how efficiency claims unravel under scrutiny.

Referee Report

4 major / 3 minor

Summary. The paper introduces Edge VLA (EVLA), a 1B-parameter vision-language-action model built from Qwen2-0.5B and two visual encoders (SigLIP, DINOv2), with a two-phase training procedure: VLM pretraining on image-text data, then manipulation fine-tuning on OpenX/BridgeData V2. The central claims are that removing the autoregressive constraint on end-effector position prediction yields a large inference speedup (stated as 7x in the abstract and Section I) and that EVLA achieves training performance comparable to OpenVLA at a fraction of the compute. The evidence consists of early training loss and action-token accuracy curves (Figures 2 and 3) and an efficiency comparison (Table I). The paper releases checkpoints and code.

Significance. If the claims were substantiated, the contribution would be practically valuable: a substantially smaller VLA that avoids autoregressive decoding would lower hardware barriers for real-time mobile manipulation. The paper deserves credit for using public datasets, releasing checkpoints and a training codebase, and posing a falsifiable architectural hypothesis (joint, non-autoregressive end-effector prediction) that is not circular. However, the current evidence does not establish the headline results: the speedup is reported inconsistently and in a confounded comparison, and the 'comparable training performance' claim rests on early training curves without final convergence, variance reporting, or closed-loop robot evaluation.

major comments (4)

[Abstract; Section III.B; Table I] The speedup claim is internally inconsistent and confounded. The abstract and Section I state a '7x speedup in inference', Section III.B says a 'six-times speedup', and Table I reports 20 ms for OpenVLA versus 5 ms for EVLA, which is a 4x speedup. Section IV.B refers to the training iteration being 'around 7 times faster', which is a different quantity. Because efficiency is the central value proposition, the authors must report one consistent, controlled inference benchmark—same hardware, attention kernels, batch size, and sequence length—and state explicitly what fraction of the speedup is attributable to the non-autoregressive architecture versus the smaller language model. The current Table I is not controlled: OpenVLA uses flash_attention2 kernels while EVLA is evaluated in eager mode, so the observed difference cannot be attributed to the proposed design.
[Section IV.A; Section IV.B; Figures 2 and 3] The 'comparable training performance' claim is not supported by the evidence. The paper shows only early training loss and action-token accuracy curves, without final converged values, standard deviations, or evaluation-set metrics. Section IV.B explicitly states that the full OpenVLA training was not reproduced, so the comparison lacks a matched baseline. The authors should report converged final metrics, train with multiple seeds, and either reproduce OpenVLA under the same protocol or clearly label the claim as preliminary and non-comparative.
[Section IV; Section V] No physical-robot or closed-loop evaluation is reported. Action-token accuracy on training data does not establish that the predicted end-effector positions are executable or that they transfer to real manipulation tasks. Section V itself identifies robot evaluation as the crucial next step, which is in tension with the abstract's wording that EVLA achieves 'comparable training performance' to OpenVLA. The authors should either add success-rate evaluations on at least one embodiment or substantially weaken the conclusions to match the available evidence.
[Section IV.A; Section IV.B] The term 'training efficiency' is used in contradictory ways: Section IV.A says EVLA's training efficiency is 'distinguishably slower' than OpenVLA, while Section IV.B says the training iteration is 'around 7 times faster'. Please clarify whether this refers to wall-clock time per training step, sample efficiency, or convergence speed, and report both per-step and end-to-end training costs.

minor comments (3)

[Figure 1 caption] The caption says 'the EVLA LLM is being retrained to generate end-effector position in an autoregressive fashion', which contradicts the method description in Section III.B; the caption or diagram should be corrected.
[Section I] The phrase 'Small Large Language Models (SMLs)' should be 'small language models (SLMs)' for consistency with the rest of the paper.
[Throughout] There are formatting typos such as 'a7-times' (Section I) and inconsistent spacing in '8A100-80GB' (Section IV.A); please proofread for spacing and consistent hardware naming.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: EVLA's claims are empirical comparisons against external models and datasets, with no fitted-input prediction loop.

full rationale

The paper's central comparison—EVLA versus OpenVLA on training loss and action-token accuracy—is grounded in external baselines and public datasets (BridgeData V2 and OpenX). The non-autoregressive end-effector prediction is an architectural choice that is measured empirically, not a quantity defined in terms of a fitted parameter. Phase 2 trains on OpenX and then reports training curves on that same dataset, but the paper explicitly frames these as 'training characteristics' rather than held-out predictions, so this is a limitation in generalization evidence, not a circular derivation. No load-bearing self-citation appears: the architecture follows external prior work (Prismatic, OpenVLA, Qwen2, SigLIP, DINOv2), and no result is justified solely by a same-author citation. The acknowledged absence of real-robot evaluation is an honest scope limitation. The internal inconsistency in the claimed speedup (abstract '7x', Section III.B 'six-times', Table I implying 4x from 20ms to 5ms) and the confounded efficiency comparison (different model sizes, flash-attention vs eager mode) are correctness and experimental-control problems, not circularity. Accordingly, the derivation chain is self-contained with respect to circularity, and the appropriate score is 0.

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

No free parameters are fitted to data in this empirical architecture paper; the only inputs are dataset choices and model-scale choices from prior work. The two domain assumptions listed are load-bearing because the paper's evidence is limited to training curves.

assumptions (2)

domain assumption Early training loss and action token accuracy are a valid proxy for final robot manipulation performance.
Section IV presents only training curves and never validates on physical robots; the paper's central claim of comparable performance depends on this proxy.
domain assumption The OpenX and BridgeData V2 datasets provide a fair and sufficient testbed for comparing VLA training behavior.
Section III.B trains on these public datasets; the comparison assumes they exercise both models in a representative way.

how reviews work

0 comments

Cite this review

Pith. "Pith review of EdgeVLA: Efficient Vision-Language-Action Models." pith.science (2026). https://pith.science/paper/JIAELROS

@misc{pith2026250714049,
  author       = {Pith},
  title        = {Pith review of: EdgeVLA: Efficient Vision-Language-Action Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JIAELROS}},
  note         = {Machine review of arXiv:2507.14049}
}

read the original abstract

Vision-Language Models (VLMs) have emerged as a promising approach to address the data scarcity challenge in robotics, enabling the development of generalizable visuomotor control policies. While models like OpenVLA showcase the potential of this paradigm, deploying large-scale VLMs on resource-constrained mobile manipulation systems remains a significant hurdle. This paper introduces Edge VLA (EVLA), a novel approach designed to significantly enhance the inference speed of Vision-Language-Action (VLA) models. EVLA maintains the representational power of these models while enabling real-time performance on edge devices. We achieve this through two key innovations: 1) Eliminating the autoregressive requirement for end-effector position prediction, leading to a 7x speedup in inference, and 2) Leveraging the efficiency of Small Language Models (SLMs), demonstrating comparable training performance to larger models with significantly reduced computational demands. Our early results demonstrate that EVLA achieves comparable training characteristics to OpenVLA while offering substantial gains in inference speed and memory efficiency. We release our model checkpoints and training \href{https://github.com/kscalelabs/evla }{codebase} to foster further research.

Figures

Figures reproduced from arXiv: 2507.14049 by the authors.

**Figure 2.** The loss (left) and action token accuracy (right) training curves for both OpenVLA and EVLA models during training [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** The loss (left) and the action token accuracy (right) training curves for both OpenVLA and EVLA models during [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

TIDAL: Temporally Interleaved Diffusion and Action Loop for High-Frequency VLA Control
cs.RO 2026-01 conditional novelty 6.0 of 10

TIDAL raises VLA control feedback from ~2.4 Hz to ~9 Hz by caching semantic intent and interleaving one-step flow generation with execution, doubling dynamic interception success in simulation.
FlashDrive: Flash Vision-Language-Action Inference for Autonomous Driving
cs.AI 2026-08 conditional novelty 5.0 of 10

A co-design of streaming KV-cache reuse, diffusion-based speculative decoding, adaptive flow-matching step caching, and CUDA Graph/kernel fusion cuts VLA autonomous-driving inference latency 4.7x with roughly unchange...
Large Model Empowered Embodied AI: A Survey on Decision-Making and Embodied Learning
cs.RO 2025-08 reject novelty 4.0 of 10

A review that categorizes large-model-empowered embodied AI into hierarchical and end-to-end decision-making, imitation and reinforcement learning, and world models.

Reference graph

Works this paper leans on

20 extracted references · 14 canonical work pages · cited by 3 Pith papers

[1]

Marah Abdin, Jyoti Aneja, Hany Awadalla, Ahmed Awadallah, Am- mar Ahmad Awan, Nguyen Bach, Amit Bahree, Arash Bakhtiari, Jianmin Bao, Harkirat Behl, Alon Benhaim, Misha Bilenko, Johan Bjorck, S ´ebastien Bubeck, Martin Cai, Qin Cai, Vishrav Chaudhary, Dong Chen, Dongdong Chen, Weizhu Chen, Yen-Chun Chen, Yi- Ling Chen, Hao Cheng, Parul Chopra, Xiyang Dai,...

work page 2024
[2]

Rt-2: Vision-language-action models transfer web knowledge to robotic control, 2023

Anthony Brohan, Noah Brown, Justice Carbajal, Yevgen Chebotar, Xi Chen, Krzysztof Choromanski, Tianli Ding, Danny Driess, Avinava Dubey, Chelsea Finn, Pete Florence, Chuyuan Fu, Montse Gonzalez Arenas, Keerthana Gopalakrishnan, Kehang Han, Karol Hausman, Alexander Herzog, Jasmine Hsu, Brian Ichter, Alex Irpan, Nikhil Joshi, Ryan Julian, Dmitry Kalashnikov...

work page 2023
[3]

Diffusion policy: Visuomotor policy learning via action diffusion, 2024

Cheng Chi, Zhenjia Xu, Siyuan Feng, Eric Cousineau, Yilun Du, Benjamin Burchfiel, Russ Tedrake, and Shuran Song. Diffusion policy: Visuomotor policy learning via action diffusion, 2024

2024
[4]

Embodiment Collaboration, Abby O’Neill, Abdul Rehman, Abhinav Gupta, Abhiram Maddukuri, Abhishek Gupta, Abhishek Padalkar, Abraham Lee, Acorn Pooley, Agrim Gupta, Ajay Mandlekar, Ajinkya Jain, Albert Tung, Alex Bewley, Alex Herzog, Alex Irpan, Alexan- der Khazatsky, Anant Rai, Anchit Gupta, Andrew Wang, Andrey Kolobov, Anikait Singh, Animesh Garg, Anirudd...

work page 2024
[5]

Flashattention-2: Faster attention with better parallelism and work partitioning, 2023

Tri Dao. Flashattention-2: Faster attention with better parallelism and work partitioning, 2023

2023
[6]

Zhao, and Chelsea Finn

Zipeng Fu, Tony Z. Zhao, and Chelsea Finn. Mobile aloha: Learning bimanual mobile manipulation with low-cost whole-body teleopera- tion. 2024

work page 2024
[7]

Flexattention: The flexibility of pytorch with the performance of flashattention, 2024

Yanbo Liang Horace He, Driss Guessous and Joy Dong. Flexattention: The flexibility of pytorch with the performance of flashattention, 2024

work page 2024
[8]

Prismatic vlms: Investigating the design space of visually-conditioned language models, 2024

Siddharth Karamcheti, Suraj Nair, Ashwin Balakrishna, Percy Liang, Thomas Kollar, and Dorsa Sadigh. Prismatic vlms: Investigating the design space of visually-conditioned language models, 2024

work page 2024

Show all 20 references

[9]

Openvla: An open-source vision-language-action model

Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakrishna, Suraj Nair, Rafael Rafailov, Ethan Foster, Grace Lam, Pannag Sanketi, Quan Vuong, Thomas Kollar, Benjamin Burchfiel, Russ Tedrake, Dorsa Sadigh, Sergey Levine, Percy Liang, and Chelsea Finn. Openvla...

2024
[10]

Jin Kim, Nur Muhammad Mahi Shafiullah, and Lerrel Pinto

Seungjae Lee, Yibin Wang, Haritheja Etukuru, H. Jin Kim, Nur Muhammad Mahi Shafiullah, and Lerrel Pinto. Behavior generation with latent actions, 2024

2024
[11]

Video-llava: Learning united visual representation by alignment before projection, 2023

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

2023
[12]

Dinov2: Learning robust visual features without supervision, 2024

Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mahmoud Assran, Nicolas Ballas, Woj- ciech Galuba, Russell Howes, Po-Yao Huang, Shang-Wen Li, Ishan Misra, Michae...

2024
[13]

Gemma Team, Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupatiraju, L ´eonard Hussenot, Thomas Mes- nard, Bobak Shahriari, Alexandre Ram ´e, Johan Ferret, Peter Liu, Pouya Tafti, Abe Friesen, Michelle Casbon, Sabela Ramos, Ravin Kumar, Charline ...

2024
[14]

Philippe Tillet, H. T. Kung, and David Cox. Triton: An intermediate language and compiler for tiled neural network computations. 2018

2018
[15]

Bridgedata v2: A dataset for robot learning at scale, 2024

Homer Walke, Kevin Black, Abraham Lee, Moo Jin Kim, Max Du, Chongyi Zheng, Tony Zhao, Philippe Hansen-Estruch, Quan Vuong, Andre He, Vivek Myers, Kuan Fang, Chelsea Finn, and Sergey Levine. Bridgedata v2: A dataset for robot learning at scale, 2024

2024
[16]

Bitnet: Scaling 1-bit transformers for large language models, 2023

Hongyu Wang, Shuming Ma, Li Dong, Shaohan Huang, Huaijie Wang, Lingxiao Ma, Fan Yang, Ruiping Wang, Yi Wu, and Furu Wei. Bitnet: Scaling 1-bit transformers for large language models, 2023

2023
[17]

Qwen2 technical report, 2024

An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Chengpeng Li, Chengyuan Li, Dayiheng Liu, Fei Huang, Guanting Dong, Haoran Wei, Huan Lin, Jialong Tang, Jialin Wang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Ma, Jianxin Yang, Jin Xu, Jingren Zhou, Jinze...

2024
[18]

Homerobot: Open-vocabulary mobile manipulation, 2024

Sriram Yenamandra, Arun Ramachandran, Karmesh Yadav, Austin Wang, Mukul Khanna, Theophile Gervet, Tsung-Yen Yang, Vidhi Jain, Alexander William Clegg, John Turner, Zsolt Kira, Manolis Savva, Angel Chang, Devendra Singh Chaplot, Dhruv Batra, Roozbeh Mot- taghi, Yonatan Bisk, an...

2024
[19]

Sigmoid loss for language image pre-training, 2023

Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, and Lucas Beyer. Sigmoid loss for language image pre-training, 2023

2023
[20]

Zhao, Vikash Kumar, Sergey Levine, and Chelsea Finn

Tony Z. Zhao, Vikash Kumar, Sergey Levine, and Chelsea Finn. Learning fine-grained bimanual manipulation with low-cost hardware. 2023

2023

Pith tools

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

[1] [1]

Marah Abdin, Jyoti Aneja, Hany Awadalla, Ahmed Awadallah, Am- mar Ahmad Awan, Nguyen Bach, Amit Bahree, Arash Bakhtiari, Jianmin Bao, Harkirat Behl, Alon Benhaim, Misha Bilenko, Johan Bjorck, S ´ebastien Bubeck, Martin Cai, Qin Cai, Vishrav Chaudhary, Dong Chen, Dongdong Chen, Weizhu Chen, Yen-Chun Chen, Yi- Ling Chen, Hao Cheng, Parul Chopra, Xiyang Dai,...

work page 2024

[2] [2]

Rt-2: Vision-language-action models transfer web knowledge to robotic control, 2023

Anthony Brohan, Noah Brown, Justice Carbajal, Yevgen Chebotar, Xi Chen, Krzysztof Choromanski, Tianli Ding, Danny Driess, Avinava Dubey, Chelsea Finn, Pete Florence, Chuyuan Fu, Montse Gonzalez Arenas, Keerthana Gopalakrishnan, Kehang Han, Karol Hausman, Alexander Herzog, Jasmine Hsu, Brian Ichter, Alex Irpan, Nikhil Joshi, Ryan Julian, Dmitry Kalashnikov...

work page 2023

[3] [3]

Diffusion policy: Visuomotor policy learning via action diffusion, 2024

Cheng Chi, Zhenjia Xu, Siyuan Feng, Eric Cousineau, Yilun Du, Benjamin Burchfiel, Russ Tedrake, and Shuran Song. Diffusion policy: Visuomotor policy learning via action diffusion, 2024

2024

[4] [4]

Embodiment Collaboration, Abby O’Neill, Abdul Rehman, Abhinav Gupta, Abhiram Maddukuri, Abhishek Gupta, Abhishek Padalkar, Abraham Lee, Acorn Pooley, Agrim Gupta, Ajay Mandlekar, Ajinkya Jain, Albert Tung, Alex Bewley, Alex Herzog, Alex Irpan, Alexan- der Khazatsky, Anant Rai, Anchit Gupta, Andrew Wang, Andrey Kolobov, Anikait Singh, Animesh Garg, Anirudd...

work page 2024

[5] [5]

Flashattention-2: Faster attention with better parallelism and work partitioning, 2023

Tri Dao. Flashattention-2: Faster attention with better parallelism and work partitioning, 2023

2023

[6] [6]

Zhao, and Chelsea Finn

Zipeng Fu, Tony Z. Zhao, and Chelsea Finn. Mobile aloha: Learning bimanual mobile manipulation with low-cost whole-body teleopera- tion. 2024

work page 2024

[7] [7]

Flexattention: The flexibility of pytorch with the performance of flashattention, 2024

Yanbo Liang Horace He, Driss Guessous and Joy Dong. Flexattention: The flexibility of pytorch with the performance of flashattention, 2024

work page 2024

[8] [8]

Prismatic vlms: Investigating the design space of visually-conditioned language models, 2024

Siddharth Karamcheti, Suraj Nair, Ashwin Balakrishna, Percy Liang, Thomas Kollar, and Dorsa Sadigh. Prismatic vlms: Investigating the design space of visually-conditioned language models, 2024

work page 2024