Pith. sign in

REVIEW 3 major objections 6 minor 2 cited by

Emma-X: An Embodied Multimodal Action Model with Grounded Chain of Thought and Look-ahead Spatial Reasoning

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Adding grounded chain-of-thought and look-ahead spatial targets to a vision-language-action model nearly doubles its real-world manipulation success rate.

desk verdict Grounded CoT and look-ahead spatial reasoning produce real, consistent gains on a WidowX benchmark, but the load-bearing trajectory segmentation is not semantically validated and the evaluation needs error bars before I'd trust the numbers. read the letter →

arxiv 2412.11974 v2 pith:3NWI74FI submitted 2024-12-16 cs.RO cs.AIcs.CLcs.CV

Qi Sun , Pengfei Hong , Tej Deep Pala , Vernon Toh , U-Xuan Tan , Deepanway Ghosal , Soujanya Poria This is my paper · ORCID
classification cs.ROcs.AIcs.CLcs.CV
keywords vision-language-actionmodelsrobotmanipulationchain-of-thoughtreasoningspatialtrajectorysegmentationBridgeV2groundedpolicygeneralization
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

The paper argues that a vision-language-action (VLA) robot policy can generalize much better if, before predicting each action, it is trained to generate a grounded chain of thought: a subtask label and scene reasoning, plus a look-ahead spatial target consisting of the 2D gripper position at the next segment boundary and a 3D movement plan to reach it. To make that training signal reliable, the authors build a 60,000-trajectory dataset from BridgeV2 in which demonstrations are split into segments by gripper state changes and motion-trajectory clustering, and each segment is labeled by a multimodal model from images rather than from text alone. On 12 real-world WidowX-250 tasks, the resulting 7B model, EMMA-X, reaches average success rates of 71.66% (half-success) and 57.50% (full success), versus 45.41%/33.33% for OpenVLA and 28.75%/15.00% for the text-reasoning baseline ECoT. The largest gains appear on spatial-relation and out-of-distribution instruction tasks, supporting the claim that look-ahead spatial reasoning and visually grounded reasoning, not extra fine-tuning alone, drive the improvement.

What carries the argument

The load-bearing mechanism is a two-stage data-annotation pipeline. First, trajectory segmentation: HDBSCAN clusters end-effector poses under a custom distance $d(i,j) = \|p_i-p_j\|_2 + \lambda\|r_i-r_j\|_2 + \beta|t_i-t_j|$ (with $\lambda=1$, $\beta=0.03$), and a breakpoint is also inserted wherever the gripper open/close state changes ($gs_i \neq gs_{i+1}$); this yields segments meant to correspond to semantic subtasks. Second, per-segment supervision: Gemini receives the segmented images plus the task instruction and outputs a subtask label and reasoning, while OWLv2 and SAM detect the 2D gripper position, and the difference between the current and the next-segment-initial gripper poses is converted into a templated 3D movement plan following the motion-language idea of RT-H. These outputs are flattened into the text-token stream of OpenVLA, so the model is trained to predict reasoning, future gripper position, movement plan, and action tokens autoregressively.

What would settle it

Take the same 12 real-robot tasks and compare EMMA-X against a version trained with segment boundaries deliberately shifted by a few frames, either randomly or toward the middle of a semantic phase; if success rates do not drop as the shifts grow, the segmentation is not carrying the claimed load. Alternatively, manually annotate true subtask boundaries on a held-out subset of BridgeV2 and check whether EMMA-X's predicted subtasks and future gripper positions align with those boundaries better than chance.

Watch

Extended reading notes

Core claim

EMMA-X is a 7B-parameter model obtained by fine-tuning OpenVLA on a hierarchically annotated version of BridgeV2. For every state in each of 60,000 trajectories, the training data contains: a grounded chain-of-thought (subtask label plus reasoning generated by Gemini from segmented images), the 2D gripper position of the first state of the next segment, and a natural-language 3D movement plan from the current state to that future state. At inference the model emits all of this before outputting the 7-dimensional action, so the action is conditioned on a prediction of where the gripper should be next and how to get there. The authors' central claim is that this combination—visually grounded reasoning plus explicit look-ahead spatial targets—is what lets the policy handle unseen objects, instructions, and spatial relations. Ablations show that removing any one component (future gripper position, movement plan, grounded reasoning, or the HDBSCAN segmentation) drops success rates by 25 to 60 percentage points.

Load-bearing premise

Everything depends on the assumption that the automated segmentation boundaries—where HDBSCAN finds a new motion cluster or the gripper changes state—fall at the same places a human would draw subtask boundaries; if those boundaries are misplaced, the subtask labels, future gripper positions, and movement plans used as training targets are noisy, and the measured gains could come from that noise rather than from genuine reasoning.

Editorial extensions

If this is right

  • Training a VLA to predict a future gripper position and a movement plan before each action yields large gains on spatial-relation tasks, such as 'put the upper half of the carrot in the pot', which rises from 30% to 80% half-success.
  • Grounded reasoning that labels segments from images rather than from text alone reduces the label hallucination seen in ECoT-style text-only chain-of-thought data.
  • Fine-tuning OpenVLA on BridgeV2 by itself does not help and even hurts performance; the hierarchical annotations, not extra exposure to the same data, are what improve generalization.
  • Ablations attribute drops of 25 to 60 percentage points to removing the grounded reasoning, the future gripper position, the movement plan, or the HDBSCAN segmentation, indicating each component is necessary for the full effect.
  • The entire 60k-trajectory dataset is auto-annotated, so the pipeline can be applied to other demonstration datasets without manual labeling.
  • The paper's own limitations section flags high inference latency (about ten times the tokens of OpenVLA), restricted task scope, and SAM's failure to detect the gripper when occluded or out of frame; these set the practical bounds of the claimed gains.

Reading between the lines

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

  • The look-ahead spatial target is effectively a learned subgoal: predicting the gripper pose at the next segment boundary injects long-horizon structure without dense reward shaping or a separate planner, and the idea could transfer to other embodiments by redefining the 'future state' in that embodiment's state space.
  • Because the reasoning and spatial tokens multiply sequence length roughly tenfold, latency is the main practical bottleneck; a natural extension is to distill the reasoning into a short subgoal code or to predict actions for an entire segment and only re-plan on deviation, as the authors themselves suggest.
  • The method's upper bound is set by the 2D gripper detector: when SAM fails due to occlusion or an out-of-frame gripper, the future-position target is wrong, so replacing the detector with a more robust hand tracker should directly improve success; this is a testable swap.
  • The segmentation assumption could be tested on the data side: measure label noise by comparing Gemini's subtask labels against human annotations on a sample of BridgeV2 trajectories, a comparison the paper does not report.
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 / 6 minor

Summary. The paper introduces EMMA-X, a 7B-parameter Vision-Language-Action model obtained by fine-tuning OpenVLA on a newly constructed hierarchical embodiment dataset derived from BridgeV2. The dataset contains about 60,000 trajectories augmented with grounded chain-of-thought reasoning, future 2D gripper positions, and 3D movement plans. To generate these annotations, the authors propose a trajectory segmentation method that combines HDBSCAN clustering with a custom distance metric over end-effector position, orientation, and timestamp, together with gripper-state transitions. The segments are then used to prompt Gemini for subtask labels and reasoning, and to compute look-ahead spatial targets from segment boundaries. The paper reports real-world experiments on 12 WidowX-250 manipulation tasks, showing that EMMA-X outperforms OpenVLA and ECoT, with particularly large gains on spatial-relation and out-of-distribution instruction tasks. Ablation studies in Table 2 examine the contributions of the segmentation, the look-ahead 2D position, the 3D movement plan, and the grounded reasoning. The paper also includes a limitations section noting higher inference latency, limited generalization scale, and potential inaccuracies in SAM-based gripper detection.

Significance. If the empirical results hold, EMMA-X represents a meaningful practical improvement in VLA policies for real-world manipulation, and the released dataset and code would be useful assets for the community. The paper provides a clear ablation structure that separates the contributions of the main design components, and it makes falsifiable claims about the value of grounded reasoning and look-ahead spatial targets. However, the strength of these claims is currently limited by two gaps: the trajectory segmentation that underpins all generated labels is not validated semantically, and the evaluation relies on small samples without statistical uncertainty estimates. These gaps affect the attribution of the measured gains to the proposed mechanisms rather than to dataset or evaluation artifacts, so the significance is promising but not yet fully established.

major comments (3)
  1. [Section 3.1 and Appendix D] The trajectory segmentation, which is the foundation of the entire data-generation pipeline, is never validated for semantic correctness. The paper reports only average segment length (5.5 frames) and average segment count (6.9) in Appendix D, and the w/o HDBSCAN ablation in Table 2 merely shows that gripper-state-only segmentation performs worse; it does not establish that HDBSCAN boundaries align with human-perceived subtasks. Given that a one-frame boundary error mislabels roughly 18% of the frames in a segment, the authors should provide a human-agreement study on segment boundaries and/or a quantitative analysis of label noise before attributing the policy gains to grounded reasoning and look-ahead spatial reasoning.
  2. [Section 4.2 and Table 1] The main evaluation uses 10 trials per task and reports no confidence intervals, error bars, or significance tests. With n=10, differences such as 80% versus 60% on 'Put the left half of the lemon in the pan' are not statistically reliable, and the average success rates across 12 tasks may be driven by a few outliers. The paper should report per-task variance or bootstrap confidence intervals over trials, and justify the sample size for the central generalization claims.
  3. [Section 3.2 and Appendix B] The Gemini-generated subtask labels and reasoning justifications are used as training targets without any quantitative quality assessment. The paper's claim that the proposed segmentation reduces hallucination is supported only by the qualitative example in Figure 1, not by a human evaluation or a systematic comparison with alternative labeling schemes. Because the grounded chain-of-thought component is one of the two main contributions, the authors should measure the accuracy of the generated reasoning labels or demonstrate that the training signals are not dominated by label noise.
minor comments (6)
  1. [Section 3.1] The hyperparameters λ=1 and β=0.03 in Eq. (1) are chosen by manual selection for 'best segmentation', but no sensitivity analysis is reported; please report how the segmentation quality varies with these parameters.
  2. [Section 3.2] The phrase 'using the state policy of the robot' is unclear; the 3D positions of the end effector are presumably taken from recorded robot state, not a policy, so please clarify the wording.
  3. [Section 4.5] The paper states that the six prompts used in Table 2 are indicated in magenta in Appendix C, but the prompt list in the manuscript is not colored; please make the selection explicit so readers can identify the subset.
  4. [Appendix D] Reporting only the averages for frames per segment and segments per trajectory hides the variability; include standard deviations or histograms for these statistics.
  5. [Global notation] The model name is written inconsistently as 'EMMA-X' and 'EMMA -X' across the paper; please unify the notation.
  6. [Appendix F] In Algorithm 1, the line 'mt ← Template(gt - gend)' appears to compute the movement plan from the current gripper position to a future position, but Section 3.2 says the plan is from the current state to the end state of the current segment; ensure the notation gt and gend is consistent with the text.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reasoning targets are supervised auxiliary labels, and the performance claims are evaluated against external real-robot baselines and ablations.

full rationale

The paper's hierarchical dataset derives grounded chain-of-thought and look-ahead spatial targets (2D future gripper position, 3D movement plan) deterministically from the same BridgeV2 expert trajectories used for action supervision. This is standard auxiliary-task training, not circularity: the targets are supervised labels rather than fitted parameters, and the central empirical claim is that predicting these intermediate tokens improves downstream action prediction. The claim is supported by ablations (w/o mt, w/o gt, w/o GRt, w/o HDBSCAN) and compared against OpenVLA, OpenVLA fine-tuned on BridgeV2, and ECoT on 12 real-robot tasks. No load-bearing self-citation or imported uniqueness theorem is present; citations to ECoT, OpenVLA, Gemini, OWLv2/SAM, RT-H, and HDBSCAN are external prior work, and the distance metric of Eq. (1) is stated with its hyperparameters rather than fitted to the evaluation. The unvalidated trajectory segmentation is a legitimate empirical/correctness risk (label noise, boundary misplacement) but not a circularity: an unvalidated intermediate representation is not equivalent by construction to the output being predicted. Consequently, no step in the derivation chain reduces to its own inputs or to a self-citation chain.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests mainly on the quality of the auto-generated reasoning labels. The segmentation weights are hand-tuned but not fitted to task success; they affect data quality. The paper postulates that Gemini, SAM, and OpenVLA are reliable enough to create useful training signal, but does not validate label accuracy quantitatively.

free parameters (2)
  • lambda (λ) in distance metric = 1
    Weights the orientation difference in the HDBSCAN distance metric (Eq. 1). Chosen 'for best segmentation' (Section 3.1), directly affects segment boundaries and thus all downstream labels.
  • beta (β) in distance metric = 0.03
    Weights temporal distance in Eq. 1; chosen 'for best segmentation'. Affects segmentation granularity and look-ahead targets.
assumptions (4)
  • domain assumption OpenVLA provides a sufficiently strong VLA backbone whose pretrained representations are generalizable across objects and instructions.
    The method fine-tunes OpenVLA; if the base model were weak, the improvements could be an artifact of poor baselines. The paper relies on OpenVLA's pretraining on OXE.
  • domain assumption Gemini-1.5-pro is capable of producing accurate, grounded subtask annotations from segmented image sequences.
    All grounded reasoning labels (GRt) are generated by Gemini without human verification (Section 3.2). The paper states Gemini can hallucinate, and claims segmentation mitigates it, but does not measure residual error.
  • domain assumption OWLv2 + SAM reliably detect the 2D gripper position in current and future frames.
    Used to generate 2D gripper position labels (gt, gend) and at inference to detect current gripper position. The paper itself notes SAM can be inaccurate when the gripper is occluded or out of frame (Limitations).
  • ad hoc to paper The custom distance metric d(i,j) in Eq. 1 with Euclidean norms and fixed weights is an appropriate similarity measure for segmenting end-effector trajectories into semantic subtasks.
    No principled derivation is given for the combination of position, orientation, and time; the weights are hand-picked. The segmentation quality is not directly evaluated against human annotation, yet all downstream reasoning labels depend on it.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Emma-X: An Embodied Multimodal Action Model with Grounded Chain of Thought and Look-ahead Spatial Reasoning." pith.science (2026). https://pith.science/paper/3NWI74FI

@misc{pith2026241211974,
  author       = {Pith},
  title        = {Pith review of: Emma-X: An Embodied Multimodal Action Model with Grounded Chain of Thought and Look-ahead Spatial Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3NWI74FI}},
  note         = {Machine review of arXiv:2412.11974}
}
read the original abstract

Traditional reinforcement learning-based robotic control methods are often task-specific and fail to generalize across diverse environments or unseen objects and instructions. Visual Language Models (VLMs) demonstrate strong scene understanding and planning capabilities but lack the ability to generate actionable policies tailored to specific robotic embodiments. To address this, Visual-Language-Action (VLA) models have emerged, yet they face challenges in long-horizon spatial reasoning and grounded task planning. In this work, we propose the Embodied Multimodal Action Model with Grounded Chain of Thought and Look-ahead Spatial Reasoning, Emma-X. Emma-X leverages our constructed hierarchical embodiment dataset based on BridgeV2, containing 60,000 robot manipulation trajectories auto-annotated with grounded task reasoning and spatial guidance. Additionally, we introduce a trajectory segmentation strategy based on gripper states and motion trajectories, which can help mitigate hallucination in grounding subtask reasoning generation. Experimental results demonstrate that Emma-X achieves superior performance over competitive baselines, particularly in real-world robotic tasks requiring spatial reasoning.

Figures

Figures reproduced from arXiv: 2412.11974 by the authors.

Figure 1
Figure 1. Comparison of our EMMA-X with ECoT in task reasoning. While both approaches utilize Gemini, our method also incorporates image sequence input, whereas ECoT relies solely on text input. We also illustrate an example of spatial reasoning. beyond task-specific training data and limiting their applicability (Brohan et al., 2023b; Chi et al.). Recent advancements in foundation models for vision and language have highligh… view at source ↗
Figure 2
Figure 2. Construction of our hierarchical embodied dataset. We first segment the trajectory. Then, we generate the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The overview of EMMA-X fine-tuned from OpenVLA using our hierarchical embodiment dataset. Why Look-ahead Spatial Reasoning? Consider the robot as a delivery driver tasked with delivering a package to a specific destination (the goal). The driver has access to a detailed high-level map of the city, which provides potential landmarks or check￾points (st+k) along the way to the destination. To reach the goal efficientl… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Experimental results on different categories [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Qualitative examples of successful and failed cases with [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. VANE: Reliable Test-Time Training for Vision-Language-Action Models via Future Visual Representation Prediction

    cs.RO 2026-08 conditional novelty 6.0 of 10

    VANE combines context-routed latent prompts, future visual representation prediction, and validation-gated shadow updates, yielding modest but consistent success-rate gains on a simulated robot arm while cutting optim...

  2. RoboInter1.5: A Holistic Intermediate Representation Suite for Embodied World Modeling and Robotic Manipulation

    cs.RO 2026-07 conditional novelty 6.0 of 10

    Dense per-frame intermediate representations (traces, masks, grasp poses, subtasks) improve embodied VQA, VLA action generation, and world-model video prediction in the new 230k-episode RoboInter-Data suite.

Reference graph

Works this paper leans on

24 extracted references · 7 canonical work pages · cited by 2 Pith papers

  1. [1]

    Suneel Belkhale, Tianli Ding, Ted Xiao, Pierre Sermanet, Quon Vuong, Jonathan Tompson, Yevgen Chebotar, Debidatta Dwibedi, and Dorsa Sadigh. 2024. https://arxiv.org/abs/2403.01823 Rt-h: Action hierarchies using language . Preprint, arXiv:2403.01823

    arXiv 2024

  2. [2]

    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...

    arXiv 2023

  3. [3]

    Anthony Brohan, Noah Brown, Justice Carbajal, Yevgen Chebotar, Joseph Dabis, Chelsea Finn, Keerthana Gopalakrishnan, Karol Hausman, Alex Herzog, Jasmine Hsu, Julian Ibarz, Brian Ichter, Alex Irpan, Tomas Jackson, Sally Jesmonth, Nikhil J Joshi, Ryan Julian, Dmitry Kalashnikov, Yuheng Kuang, Isabel Leal, Kuang-Huei Lee, Sergey Levine, Yao Lu, Utsav Malla, ...

    arXiv 2023

  4. [4]

    Diffusion policy: Visuomotor policy learning via action diffusion

    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. The International Journal of Robotics Research, page 02783649241273668

    work page

  5. [5]

    Open X-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, Alexander Khazatsky, Anant Rai, Anchit Gupta, Andrew Wang, Andrey Kolobov, Anikait Singh, Animesh Garg, An...

    arXiv 2023

  6. [6]

    Danny Driess, Fei Xia, Mehdi S. M. Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, Wenlong Huang, Yevgen Chebotar, Pierre Sermanet, Daniel Duckworth, Sergey Levine, Vincent Vanhoucke, Karol Hausman, Marc Toussaint, Klaus Greff, Andy Zeng, Igor Mordatch, and Pete Florence. 2023. https://arxiv....

    arXiv 2023

  7. [7]

    Frederik Ebert, Yanlai Yang, Karl Schmeckpeper, Bernadette Bucher, Georgios Georgakis, Kostas Daniilidis, Chelsea Finn, and Sergey Levine. 2021. https://arxiv.org/abs/2109.13396 Bridge data: Boosting generalization of robotic skills with cross-domain datasets . Preprint, arXiv:2109.13396

    arXiv 2021

  8. [8]

    Huy Ha, Pete Florence, and Shuran Song. 2023. https://arxiv.org/abs/2307.14535 Scaling up and distilling down: Language-guided robot skill acquisition . Preprint, arXiv:2307.14535

    arXiv 2023

Show all 24 references
  1. [9]

    Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakrishna, Suraj Nair, Rafael Rafailov, Ethan P Foster, Pannag R Sanketi, Quan Vuong, Thomas Kollar, Benjamin Burchfiel, Russ Tedrake, Dorsa Sadigh, Sergey Levine, Percy Liang, and Chelsea Finn. 2024. https://...

    2024

  2. [10]

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. 2023. Segment anything. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4015--4026

    2023

  3. [11]

    Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwasawa. 2022. https://openreview.net/forum?id=e2TBb5y0yFf Large language models are zero-shot reasoners . In Advances in Neural Information Processing Systems

    2022

  4. [12]

    Jacky Liang, Wenlong Huang, Fei Xia, Peng Xu, Karol Hausman, Brian Ichter, Pete Florence, and Andy Zeng. 2023. https://arxiv.org/abs/2209.07753 Code as policies: Language model programs for embodied control . Preprint, arXiv:2209.07753

    2023 arXiv

  5. [13]

    Yueen Ma, Zixing Song, Yuzheng Zhuang, Jianye Hao, and Irwin King. 2024. A survey on vision-language-action models for embodied ai. arXiv preprint arXiv:2405.14093

    2024 arXiv

  6. [14]

    Leland McInnes, John Healy, and S. Astels. 2017. https://api.semanticscholar.org/CorpusID:53231359 hdbscan: Hierarchical density based clustering . J. Open Source Softw., 2:205

    2017

  7. [15]

    Matthias Minderer, Alexey Gritsenko, and Neil Houlsby. 2024. Scaling open-vocabulary object detection. Advances in Neural Information Processing Systems, 36

    2024

  8. [16]

    Octo Model Team , Dibya Ghosh, Homer Walke, Karl Pertsch, Kevin Black, Oier Mees, Sudeep Dasari, Joey Hejna, Charles Xu, Jianlan Luo, Tobias Kreiman, You Liang Tan, Pannag Sanketi, Quan Vuong, Ted Xiao, Dorsa Sadigh, Chelsea Finn, and Sergey Levine. 2024. Octo: An open-source ...

    2024

  9. [17]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learni...

    2021

  10. [18]

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, Katie Millican, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805

    2023 arXiv

  11. [19]

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288

    2023 arXiv

  12. [20]

    Zhao, Quan Vuong, Chongyi Zheng, Philippe Hansen-Estruch, Andre Wang He, Vivek Myers, Moo Jin Kim, Max Du, Abraham Lee, Kuan Fang, Chelsea Finn, and Sergey Levine

    Homer Rich Walke, Kevin Black, Tony Z. Zhao, Quan Vuong, Chongyi Zheng, Philippe Hansen-Estruch, Andre Wang He, Vivek Myers, Moo Jin Kim, Max Du, Abraham Lee, Kuan Fang, Chelsea Finn, and Sergey Levine. 2023. https://openreview.net/forum?id=f55MlAT1Lu Bridgedata v2: A dataset ...

    2023

  13. [21]

    Micha Zawalski, William Chen, Karl Pertsch, Oier Mees, Chelsea Finn, and Sergey Levine. 2024. Robotic control via embodied chain-of-thought reasoning. arXiv preprint arXiv:2407.08693

    2024 arXiv

  14. [22]

    Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, and Lucas Beyer. 2023. Sigmoid loss for language image pre-training. In 2023 IEEE/CVF International Conference on Computer Vision (ICCV), pages 11941--11952. IEEE

    2023

  15. [23]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  16. [24]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

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