Pith. sign in

REVIEW 3 major objections 6 minor 40 references

Meta-Controller: Few-Shot Imitation of Unseen Embodiments and Tasks in Continuous Control

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 Meta-Controller claims that a matching-based policy over joint-level tokens imitates unseen robot embodiments and tasks from five reward-free demonstrations, reporting an average normalized score of 71.1 against 37.9 for the strongest…

desk verdict A solid few-shot imitation system with a genuinely new integration, but the flagship 'unseen embodiment' result leans on a near-duplicate morphology and a warm start that is never ablated; the claims need rescoping, not the method. read the letter →

arxiv 2412.12147 v1 pith:BUQIZAJ3 submitted 2024-12-10 cs.LG cs.AIcs.RO

Seongwoong Cho , Donggyun Kim , Jinwoo Lee , Seunghoon Hong This is my paper · ORCID
classification cs.LGcs.AIcs.RO
keywords few-shotimitationlearningcontinuouscontrolembodimentgeneralizationmeta-learningbehaviorcloningmatching-basedpolicyjoint-levelrepresentationtransformers
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that a single behavior-cloning model can generalize to robot bodies and tasks never seen in training, using only five reward-free demonstrations. On the DeepMind Control suite, the proposed Meta-Controller reports an average normalized score of 71.1 across eight held-out embodiment-task pairs, compared with 37.9 for the best modular-policy baseline. The claim matters because, if it holds, a robot facing a new body shape or a new objective could be programmed by watching a short expert demonstration instead of being retrained or reward-engineered. The paper also isolates what carries the gain: the joint-level state and action representation plus a matching-based policy network, since removing the matching module drops the average to 60.3.

What carries the argument

The central machinery is the joint-level input-output representation plus a three-part network. Each robot is a set of hinge or slide joints, and per-joint state tokens (position, velocity, axis, motion type) with per-joint action tokens (torque or force command) standardize heterogeneous embodiments. The state encoder splits into a structure encoder, a bi-directional transformer over joints with embodiment-specific positional embeddings and parameter-efficient fine-tuning parameters, and a motion encoder, a causal transformer over time with embodiment- and task-specific parameters. The policy network uses the same state encoder on current and demonstration states, an action encoder on demonstration actions, a matching module that computes joint-wise cosine similarity and takes a weighted sum of action features, and a causal action decoder. That matching step is what makes the policy adaptive and non-parametric, and the paper argues it prevents overfitting by recombining local motor skills instead of memorizing whole trajectories.

What would settle it

Run the same five-shot behavior-cloning protocol on a held-out embodiment whose actuators are not hinge or slide joints, such as a wheeled rover or soft gripper, and compare the normalized score to the reported 71.1 average; if the model cannot tokenize the state and action spaces, the claimed generalization to arbitrary unseen embodiments is refuted.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that simultaneous few-shot generalization to unseen embodiments and tasks is achievable by decomposing control into per-joint tokens and learning to recombine demonstration snippets. States and actions of any multi-joint robot are tokenized joint-wise, so a single architecture can consume heterogeneous bodies. A structure encoder captures morphology with a small set of embodiment-specific parameters, and a motion encoder handles dynamics with embodiment- and task-specific parameters. The policy then predicts each joint's action as a similarity-weighted sum of action features from the demonstrations, which the authors interpret as composing local motor skills. In five-shot evaluations, this yields a 71.1 average normalized score on eight held-out tasks spanning the hopper, wolf, reacher-four, and walker embodiments, versus 37.9 for the best baseline, and the ablation attributes much of the gap to the matching mechanism.

Load-bearing premise

The framework assumes every relevant robot body can be represented as a collection of hinge or slide joints with per-joint state and action tokens, so any embodiment that does not decompose that way, such as a soft or wheeled actuator, lies outside the model's input space.

Editorial extensions

If this is right

  • Five reward-free demonstrations suffice to reach a 71.1 average normalized score on unseen embodiment-task pairs, outperforming both modular policy learning and few-shot imitation baselines whose best result is 37.9.
  • The matching-based policy is the load-bearing adaptive component: removing it drops the average from 71.1 to 60.3, so the architecture's gains do not come from fine-tuning alone.
  • Joint-level tokenization with parameter-efficient embodiment-specific parameters transfers to unseen bodies that are structurally similar to training bodies, as shown by the sharp reacher-four improvement when a morphologically similar reacher is present in meta-training.
  • Because actions are composed from demonstration snippets, the policy can imitate goals that differ from poses in the demonstrations, such as a reacher-four reaching a goal position rather than copying a demonstration pose.
  • Performance improves monotonically as demonstrations increase from 5 to 20 shots and remains leading at 3 shots, indicating the result is not tied to exactly five demonstrations.

Reading between the lines

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

  • If the matching operator truly composes local motor skills, the same model should transfer to tasks whose goals are expressible as reorderings of demonstrated motion primitives even when no demonstration of the exact task exists; this is a testable prediction beyond the paper's eight held-out tasks.
  • The paper's meta-training composition ablation shows that removing morphologically similar embodiments sharply hurts downstream performance, which implies the 'unseen embodiment' claim is bounded by the diversity of the meta-training distribution rather than by the architecture alone.
  • The paper's Appendix A acknowledges that the joint-level representation may not generalize to robots with significantly different morphologies or actuation mechanisms, so a direct stress test is a non-joint embodiment such as a wheeled or soft robot.
  • Because all experiments are simulated, real-world value depends on bridging the simulation-to-reality gap; the noise-robustness experiments perturb actions in simulation and do not test physical transfer, which the authors flag as a limitation.
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 proposes Meta-Controller, a few-shot behavior cloning framework for continuous control that aims to generalize to both unseen embodiments and unseen tasks from a few reward-free demonstrations. The method tokenizes states and actions into joint-level representations, encodes states with a structure-motion transformer that combines shared and embodiment-specific parameters (via PEFT), and conditions a matching-based policy network on the demonstrations. The model is trained with episodic meta-learning and then fine-tuned on a few demonstrations for a held-out embodiment-task pair. Experiments on the DeepMind Control suite report an average normalized score of 71.1 with five demonstrations on eight held-out tasks, compared with 37.9 for the best baseline (MetaMorph), and ablations are provided for the architectural components, the adaptive parameters, the number of demonstrations, and the meta-training composition.

Significance. The paper addresses an important and timely problem: simultaneous few-shot generalization across robot embodiments and tasks. The strongest contributions are the clean integration of joint-level tokenization with a matching-based policy, the explicit decomposition of shared versus embodiment-specific parameters, and a fairly comprehensive set of ablations, including 3-shot results, embodiment-variation experiments, and visualizations of learned embeddings. The code is released, which is a practical strength. If the findings hold, the method would be a useful step beyond modular policy learning (which is task-limited) and few-shot IL (which is embodiment-limited). The main weakness is that the headline 'unseen embodiment' claim is overstated: the flagship reacher-four result is obtained on an embodiment that is structurally nearly identical to a training embodiment, with embodiment-specific parameters warm-started from that training embodiment, and the paper's own composition ablation shows the result collapses without that near-duplicate in meta-training.

major comments (3)
  1. [Section 5.1, Implementation Details; Appendix D.3, Table 12] The reacher-four result is not clean evidence of generalization to an unseen embodiment. The paper discloses that for structurally similar downstream embodiments the encoder's embodiment-specific parameters are initialized from the trained reacher-three parameters during fine-tuning, and Table 12 shows that removing reacher-three from meta-training collapses reacher-four easy/hard from 56.1/50.8 to 5.6/4.6 and 4.6/3.8, while removing unrelated embodiments leaves 82.1/41.8. The warm-start initialization is never ablated (e.g., random initialization or initialization from a different embodiment), so the contribution of morphology similarity versus the method's cross-embodiment mechanism is confounded. Since reacher-four is the single largest margin over baselines in Table 1, the title-level claim of 'unseen embodiments' should be rescoped, or the experiment should be rerun with a control for the initialization.
  2. [Section 5.1, Evaluation Protocol] The evaluation protocol reports the best score over 10,000 fine-tuning iterations, evaluated every 1,000 iterations, and selects the low-rank parameter from {4,8,16} based on the same evaluation tasks. This is an oracle selection on the test tasks; it inflates absolute scores and can favor methods with higher variance. Even if all baselines are evaluated under the same protocol, the quantitative margins in Table 1 are not robust to this selection. Please report scores at a fixed iteration budget (e.g., the score at 10,000 iterations) or the mean/median over fine-tuning iterations, and include results over multiple fine-tuning seeds.
  3. [Abstract and Section 2; Appendix A] The paper claims generalization to 'unseen embodiments' and to arbitrary control problems, but the method is built on a joint-level tokenization over a pre-defined set of joint types (hinge and slide), and Section 3.1 states 'We consider a pre-defined set of joints whose compositions differ per embodiment.' Appendix A explicitly acknowledges that the unified joint-level representation may not generalize to robots with significantly different morphologies or actuation mechanisms. The claim should be qualified throughout to 'unseen compositions of known joint types' or 'embodiments within the supported joint vocabulary,' and the abstract should not overstate the generality beyond the evaluated setting.
minor comments (6)
  1. [Section 5.1] Section 5.1 lists hopper, reacher-four, and wolf as 'unseen embodiments'; given the warm-start disclosure for reacher-four, this embodiment should be described as a structurally related embodiment rather than an unseen one, or the warm start should be explicitly ablated.
  2. [Table 1 and Section 5.1] The standard errors in Table 1 are computed over 20 initial states, but the paper does not report variance over fine-tuning seeds. With best-checkpoint selection, seed variance is important; please include results over multiple fine-tuning seeds or clarify that the reported values are from a single seed.
  3. [Figure 5] In Figure 5, the axis labels appear garbled, with repeated 'Score' and 'Number of Demo.' labels; the figure should be regenerated with clean axes and a single legend.
  4. [Section 3.3] For few-shot fine-tuning, the paper says D is randomly split into two disjoint subsets, but the model at evaluation uses the whole D in the policy network. Please clarify how this split is used (e.g., support/validation split) and whether the baselines are given the same training/evaluation demonstration budget.
  5. [Table 3] Table 3 shows that the model without task-specific motion parameters (row 2) outperforms the full model on reacher-four easy/hard (70.8/48.1 vs 56.1/50.8); the main text should acknowledge this non-monotonicity rather than only reporting the overall average.
  6. [Appendix B.1] Appendix B.1 states that only embodiments operating on a 2D coordinate space are considered; this scope restriction should be stated in Section 5.1 or the abstract as a condition on the claimed generalization.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claim is an empirical architecture result supported by in-paper ablations and external benchmarks; self-citations are attribution, not load-bearing proof.

full rationale

Meta-Controller's central claim is an empirical architecture result: with joint-level I/O tokenization, a structure-motion encoder, and a matching-based policy, it achieves higher 5-shot BC scores than baselines on DMC. There is no derivation in which a predicted quantity is equal by construction to an input or fitted constant. The matching module is attributed to the authors' prior visual token matching work (Kim et al. [20,19]), but the present paper implements the module and supports it with in-paper ablations (Table 2: adding the matching module raises average normalized score from 60.3 to 71.1; Table 10), so the self-citation is attribution, not load-bearing. PEFT parameters are adapted on the few-shot demonstrations and then evaluated on new initial states, which is the standard few-shot BC protocol rather than evaluation on training demonstrations. The disclosed warm-start for structurally similar embodiments (Section 5.1: 'we initialize the encoder's embodiment-specific parameters using the trained parameters during fine-tuning') and the Appendix D.3 composition dependence (Table 12) are scope and validity concerns about how 'unseen' reacher-four is, not circular reductions; Appendix A also candidly limits the joint-level representation to similar morphologies. No self-definitional, fitted-input-as-prediction, imported-uniqueness, or renamed-known-result pattern is present. The derivation chain is self-contained against external benchmarks and in-paper ablations, so the circularity score is 0.

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

The central claim rests on the joint-level tokenization being a universal representation for any embodiment, the replay-buffer demonstrations being adequate expert behavior, the selected held-out tasks being representative of genuinely unseen embodiments and tasks, and the transformer architecture being able to capture morphological structure. Each of these is a domain assumption rather than a derived result; the paper itself acknowledges the joint-level representation limitation in Appendix A.

free parameters (4)
  • Rank of low-rank adaptation matrices in structure and motion encoders = 16
    Set by hand, with no sensitivity analysis for the proposed model; searched over {4,8,16} for baselines.
  • Temporal segment size for episodic meta-learning = 10
    Chosen to ensure query/demonstration consistency within replay buffers.
  • Number of demonstrations per meta-training episode = 4
    Architectural choice; inference uses 5-shot or 3-shot.
  • Maximum causal attention history size = 10
    Set to limit quadratic cost of transformers.
assumptions (4)
  • domain assumption Joint-level tokenization is a sufficient universal I/O representation for all relevant robot embodiments.
    Invoked in Sec 3.1 and acknowledged in Appendix A as a limitation for embodiments with non-joint actuators.
  • domain assumption Demonstrations drawn from DrQ-v2 replay buffers, filtered by cumulative reward > 10, are adequate expert demonstrations.
    Dataset construction in Appendix B.1.
  • domain assumption Performance on the 8 selected held-out tasks is representative of general few-shot generalization to unseen embodiments and tasks.
    Evaluation in Sec 5.1; held-out embodiments are structurally related to training embodiments.
  • domain assumption The bi-directional transformer over joint tokens can capture morphological structure necessary for cross-embodiment transfer.
    Design choice in Sec 3.2.1, supported only by ablations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Meta-Controller: Few-Shot Imitation of Unseen Embodiments and Tasks in Continuous Control." pith.science (2026). https://pith.science/paper/BUQIZAJ3

@misc{pith2026241212147,
  author       = {Pith},
  title        = {Pith review of: Meta-Controller: Few-Shot Imitation of Unseen Embodiments and Tasks in Continuous Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BUQIZAJ3}},
  note         = {Machine review of arXiv:2412.12147}
}
read the original abstract

Generalizing across robot embodiments and tasks is crucial for adaptive robotic systems. Modular policy learning approaches adapt to new embodiments but are limited to specific tasks, while few-shot imitation learning (IL) approaches often focus on a single embodiment. In this paper, we introduce a few-shot behavior cloning framework to simultaneously generalize to unseen embodiments and tasks using a few (\emph{e.g.,} five) reward-free demonstrations. Our framework leverages a joint-level input-output representation to unify the state and action spaces of heterogeneous embodiments and employs a novel structure-motion state encoder that is parameterized to capture both shared knowledge across all embodiments and embodiment-specific knowledge. A matching-based policy network then predicts actions from a few demonstrations, producing an adaptive policy that is robust to over-fitting. Evaluated in the DeepMind Control suite, our framework termed \modelname{} demonstrates superior few-shot generalization to unseen embodiments and tasks over modular policy learning and few-shot IL approaches. Codes are available at \href{https://github.com/SeongwoongCho/meta-controller}{https://github.com/SeongwoongCho/meta-controller}.

Figures

Figures reproduced from arXiv: 2412.12147 by the authors.

Figure 1
Figure 1. The overall framework of Meta-Controller. First, the states and actions of various robot [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The state encoder f consists of two component transformers. Joint-level state tokens are first encoded by the structure encoder fs along the joint axis, where the positional embedding and a part of backbone parameters adapt the model to each embodiment. The features are then passed to the motion encoder fm, which computes causal attentions of per-joint features along the temporal axis, where a part of backbone param… view at source ↗
Figure 3
Figure 3. An illustration of the matching-based policy network [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (15 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison on the hard task of the reacher-four embodiment, visualizing the final states of the demonstrations and the rollout trajectories of each model. In this task, the robot must move its limb tip to the goal position (visualized as a red ball). While …
Figure 5
Figure 5. Figure 5: Ablation study on the number of demonstrations. We plot the normalized scores for each [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Learning curves of ours and PDT+PEFT in 5-shot settings. TSNE visualization of state embeddings acrobot cartpole cartpole_two cheetah cup pendulum pointmass reacher reacher_three walker acrobot_elbow cartpole_slider cartpole_two_slider cheetah_bthigh cheetah_bshin chee…
Figure 7
Figure 7. Figure 7: t-SNE visualization of embeddings of state encoder [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]
Figure 8
Figure 8. Figure 8: Failure cases of our model. As indicated by red boxes, the agent begins to solve tasks [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]
Figure 9
Figure 9. Figure 9: Cumulative rewards of the failure cases in Figure 8. [PITH_FULL_IMAGE:figures/full_fig_p020_9.png]
Figure 10
Figure 10. Figure 10: Robustness analysis of our model in the presence of various noise levels in transition [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 11
Figure 11. Figure 11: Additional Qualitative Results on hopper-hop task. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_11.png]
Figure 12
Figure 12. Figure 12: Additional Qualitative Results on hopper-hop backwards task. 24 [PITH_FULL_IMAGE:figures/full_fig_p024_12.png]
Figure 13
Figure 13. Figure 13: Additional Qualitative Results on hopper-stand task. 25 [PITH_FULL_IMAGE:figures/full_fig_p025_13.png]
Figure 14
Figure 14. Figure 14: Additional Qualitative Results on wolf-walk task. 26 [PITH_FULL_IMAGE:figures/full_fig_p026_14.png]
Figure 15
Figure 15. Figure 15: Additional Qualitative Results on wolf-run task. 27 [PITH_FULL_IMAGE:figures/full_fig_p027_15.png]
Figure 16
Figure 16. Figure 16: Additional Qualitative Results on reacher four-easy task. 28 [PITH_FULL_IMAGE:figures/full_fig_p028_16.png]
Figure 17
Figure 17. Figure 17: Additional Qualitative Results on reacher four-hard task. 29 [PITH_FULL_IMAGE:figures/full_fig_p029_17.png]
Figure 18
Figure 18. Figure 18: Additional Qualitative Results on walker-walk backwards task. 30 [PITH_FULL_IMAGE:figures/full_fig_p030_18.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 29 canonical work pages

  1. [1]

    GitHub, 3, 2019

    Pytorch lightning. GitHub, 3, 2019

    work page 2019

  2. [2]

    Ben Zaken, Y

    E. Ben Zaken, Y . Goldberg, and S. Ravfogel. BitFit: Simple parameter-efficient fine-tuning for transformer-based masked language-models. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (V olume 2: Short Papers), 2022

    work page 2022

  3. [3]

    Brohan, N

    A. Brohan, N. Brown, J. Carbajal, Y . Chebotar, X. Chen, K. Choromanski, T. Ding, D. Driess, A. Dubey, C. Finn, et al. Rt-2: Vision-language-action models transfer web knowledge to robotic control. arXiv preprint arXiv:2307.15818, 2023

    arXiv 2023

  4. [4]

    L. Chen, K. Lu, A. Rajeswaran, K. Lee, A. Grover, M. Laskin, P. Abbeel, A. Srinivas, and I. Mordatch. Decision transformer: Reinforcement learning via sequence modeling. Advances in neural information processing systems , 34:15084–15097, 2021

    2021

  5. [5]

    K. M. Choromanski, V . Likhosherstov, D. Dohan, X. Song, A. Gane, T. Sarlos, P. Hawkins, J. Q. Davis, A. Mohiuddin, L. Kaiser, et al. Rethinking attention with performers. In International Conference on Learning Representations, 2020

    work page 2020

  6. [6]

    Devin, A

    C. Devin, A. Gupta, T. Darrell, P. Abbeel, and S. Levine. Learning modular neural network policies for multi-task and multi-robot transfer. In 2017 IEEE international conference on robotics and automation (ICRA), pages 2169–2176. IEEE, 2017

    work page 2017

  7. [7]

    Y . Duan, M. Andrychowicz, B. Stadie, O. Jonathan Ho, J. Schneider, I. Sutskever, P. Abbeel, and W. Zaremba. One-shot imitation learning. Advances in neural information processing systems, 30, 2017

    2017

  8. [8]

    C. Finn, P. Abbeel, and S. Levine. Model-agnostic meta-learning for fast adaptation of deep networks. In International conference on machine learning , pages 1126–1135. PMLR, 2017

    2017

Show all 40 references
  1. [9]

    J. Fu, A. Kumar, O. Nachum, G. Tucker, and S. Levine. D4rl: Datasets for deep data-driven reinforcement learning, 2020

    2020

  2. [10]

    Furuta, Y

    H. Furuta, Y . Iwasawa, Y . Matsuo, and S. S. Gu. A system for morphology-task generalization via unified representation and behavior distillation. In The Eleventh International Conference on Learning Representations, 2022

    2022

  3. [11]

    Ghadirzadeh, X

    A. Ghadirzadeh, X. Chen, P. Poklukar, C. Finn, M. Björkman, and D. Kragic. Bayesian meta-learning for few-shot policy adaptation across robotic platforms. In 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 1274–1280. IEEE, 2021

    2021

  4. [12]

    Gupta, L

    A. Gupta, L. Fan, S. Ganguli, and L. Fei-Fei. Metamorph: Learning universal controllers with transformers. In International Conference on Learning Representations , 2021

    2021

  5. [13]

    Hakhamaneshi, R

    K. Hakhamaneshi, R. Zhao, A. Zhan, P. Abbeel, and M. Laskin. Hierarchical few-shot imitation with skill transition models. In International Conference on Learning Representations , 2021

    2021

  6. [14]

    Hansen, H

    N. Hansen, H. Su, and X. Wang. Td-mpc2: Scalable, robust world models for continuous control. In The Twelfth International Conference on Learning Representations , 2023

    2023

  7. [15]

    Hendrycks and K

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

    2016 arXiv

  8. [16]

    S. Hong, D. Yoon, and K.-E. Kim. Structure-aware transformer policy for inhomogeneous multi-task reinforcement learning. In International Conference on Learning Representations, 2021

    2021

  9. [17]

    E. J. Hu, Y . Shen, P. Wallis, Z. Allen-Zhu, Y . Li, S. Wang, L. Wang, and W. Chen. LoRA: Low-rank adaptation of large language models. In International Conference on Learning Representations, 2022. 11

    2022

  10. [18]

    Huang, I

    W. Huang, I. Mordatch, and D. Pathak. One policy to control them all: Shared modular policies for agent-agnostic control. In International Conference on Machine Learning, pages 4455–4464. PMLR, 2020

    2020

  11. [19]

    D. Kim, S. Cho, S. Kim, C. Luo, and S. Hong. Chameleon: A data-efficient generalist for dense visual prediction in the wild. arXiv preprint arXiv:2404.18459, 2024

    2024 arXiv

  12. [20]

    D. Kim, J. Kim, S. Cho, C. Luo, and S. Hong. Universal few-shot learning of dense prediction tasks with visual token matching. In The Eleventh International Conference on Learning Representations, 2022

    2022

  13. [21]

    S. Kim, S. Shen, D. Thorsley, A. Gholami, W. Kwon, J. Hassoun, and K. Keutzer. Learned token pruning for transformers. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages 784–794, 2022

    2022

  14. [22]

    D. P. Kingma and J. Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  15. [23]

    Kurin, M

    V . Kurin, M. Igl, T. Rocktäschel, W. Boehmer, and S. Whiteson. My body is a cage: the role of morphology in graph-based incompatible control. In International Conference on Learning Representations, 2021

    2021

  16. [24]

    W. Liu, A. Rabinovich, and A. C. Berg. Parsenet: Looking wider to see better. arXiv preprint arXiv:1506.04579, 2015

    2015 arXiv

  17. [25]

    R. K. Mahabadi, S. Ruder, M. Dehghani, and J. Henderson. Parameter-efficient multi-task fine-tuning for transformers via shared hypernetworks. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference...

    2021

  18. [26]

    Majumdar, K

    A. Majumdar, K. Yadav, S. Arnaud, J. Ma, C. Chen, S. Silwal, A. Jain, V .-P. Berges, T. Wu, J. Vakil, et al. Where are we in the search for an artificial visual cortex for embodied intelligence? Advances in Neural Information Processing Systems , 36:655–677, 2023

    2023

  19. [27]

    V . Sanh, L. Debut, J. Chaumond, and T. Wolf. Distilbert, a distilled version of bert: smaller, faster, cheaper and lighter. In NeurIPS EMC2 Workshop, 2019

    2019

  20. [28]

    Schmied, M

    T. Schmied, M. Hofmarcher, F. Paischer, R. Pascanu, and S. Hochreiter. Learning to modulate pre-trained models in rl. Advances in Neural Information Processing Systems , 36, 2024

    2024

  21. [29]

    Schubert, J

    I. Schubert, J. Zhang, J. Bruce, S. Bechtle, E. Parisotto, M. Riedmiller, J. T. Springenberg, A. Byravan, L. Hasenclever, and N. Heess. A generalist dynamics model for control. arXiv preprint arXiv:2305.10912, 2023

    2023 arXiv

  22. [30]

    Todorov, T

    E. Todorov, T. Erez, and Y . Tassa. Mujoco: A physics engine for model-based control. In2012 IEEE/RSJ international conference on intelligent robots and systems , pages 5026–5033. IEEE, 2012

    2012

  23. [31]

    Tunyasuvunakool, A

    S. Tunyasuvunakool, A. Muldal, Y . Doron, S. Liu, S. Bohez, J. Merel, T. Erez, T. Lillicrap, N. Heess, and Y . Tassa. dm_control: Software and tasks for continuous control. Software Impacts, 2020

    2020

  24. [32]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin. Attention is all you need. Advances in neural information processing systems , 30, 2017

    2017

  25. [33]

    Vuong, S

    Q. Vuong, S. Levine, H. R. Walke, K. Pertsch, A. Singh, R. Doshi, C. Xu, J. Luo, L. Tan, D. Shah, et al. Open x-embodiment: Robotic learning datasets and rt-x models. In Towards Generalist Robots: Learning Paradigms for Scalable Skill Acquisition@ CoRL2023 , 2023

    2023

  26. [34]

    T. Wang, R. Liao, J. Ba, and S. Fidler. Nervenet: Learning structured policy with graph neural networks. In International conference on learning representations, 2018. 12

    2018

  27. [35]

    Xiong, Y

    R. Xiong, Y . Yang, D. He, K. Zheng, S. Zheng, C. Xing, H. Zhang, Y . Lan, L. Wang, and T. Liu. On layer normalization in the transformer architecture. In International Conference on Machine Learning, pages 10524–10533. PMLR, 2020

    2020

  28. [36]

    M. Xu, Y . Lu, Y . Shen, S. Zhang, D. Zhao, and C. Gan. Hyper-decision transformer for efficient online policy adaptation. In The Eleventh International Conference on Learning Representations, 2022

    2022

  29. [37]

    M. Xu, Y . Shen, S. Zhang, Y . Lu, D. Zhao, J. Tenenbaum, and C. Gan. Prompting decision transformer for few-shot policy generalization. In international conference on machine learning, pages 24631–24645. PMLR, 2022

    2022

  30. [38]

    Yarats, I

    D. Yarats, I. Kostrikov, and R. Fergus. Image augmentation is all you need: Regularizing deep reinforcement learning from pixels. In International Conference on Learning Representations , 2021

    2021

  31. [39]

    J. Yoon, T. Kim, O. Dia, S. Kim, Y . Bengio, and S. Ahn. Bayesian model-agnostic meta-learning. Advances in neural information processing systems , 31, 2018

    2018

  32. [40]

    Y . Zhou, S. Sonawani, M. Phielipp, S. Stepputtis, and H. Amor. Modularity through attention: Efficient training and transfer of language-conditioned policies for robot manipulation. In 6th Annual Conference on Robot Learning , 2022. 13 Appendix This document provides the cont...

    2022

Pith tools

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