Pith. sign in

REVIEW 3 major objections 6 minor 294 references

V-Simba: Unleashing the Architectural Potential of RL in Visual Continuous Control

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a norm-stabilized network architecture (V-Simba) alone can match or beat leading visual continuous-control methods on 29 tasks, while training faster than DrQ-v2.

desk verdict Useful architecture paper, but the 'architecture alone' claim is not cleanly isolated across the main benchmarks because the algorithm changes (DDPG to SAC, distributional critic) at the same time. read the letter →

arxiv 2608.07870 v1 pith:KW72DWTG submitted 2026-08-08 cs.LG cs.RO

Donghu Kim , Youngdo Lee , Hojoon Lee , Johan Obando-Ceron , Byungkun Lee , Aaron Courville , Pablo Samuel Castro , Jaegul Choo
show 1 more author
Clare Lyle
This is my paper · ORCID
classification cs.LGcs.RO
keywords deepreinforcementlearningvisualcontinuouscontrolneuralarchitecturedesignlayernormalizationdistributionalcriticsampleefficiencyweightregularizationplasticityloss
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 asks whether the architectural design principles that recently improved state-based RL can be transferred to pixel-based continuous control, where progress has come mostly from algorithms rather than network design. It claims the answer is yes: a redesigned convolutional network called V-Simba, built on SAC with data augmentation, matches or outperforms leading visual RL methods on 29 tasks across DMC, Adroit, and Meta-World using a single hyperparameter set, while training in less wall-clock time than DrQ-v2 and with fewer parameters. The design idea is that visual RL's standard shallow encoder becomes unstable — sharp loss landscapes, dormant units, feature collapse, and growing feature, parameter, and gradient norms — and that constraining those norms with layer normalization, $\ell^2$ weight decay, and a distributional critic restores stable, efficient learning. If true, the result matters because it offers a cheap, algorithm-agnostic performance boost: swap the network, keep the rest.

What carries the argument

The load-bearing machinery is a set of components that jointly constrain feature, parameter, and gradient norms. Layer normalization is placed at the input and after each stage, exploiting the scale-invariance property $\mathrm{Norm}(cWx) = \mathrm{Norm}(Wx)$, which makes gradients scale as $1/c$ with weight magnitude and therefore demands explicit control of weight growth via $\ell^2$ decay. A categorical distributional critic with KL divergence and reward normalization keeps Q-value targets bounded and the loss landscape smooth. Wrapped around these are compute-saving choices — an early stride-2 convolution, 2×2 max-pooling, and pointwise 1×1 convolutions in inverted-bottleneck residual blocks — that keep the normalized network fast enough for 84×84 visual observations. These components work together to hold effective norms within a narrow band, which the paper links to smoother loss landscapes and higher final returns.

What would settle it

Run a factorial comparison on a subset of DMC tasks: V-Simba's architecture with DrQ-v2's algorithm (DDPG, MSE critic) versus DrQ-v2's architecture with V-Simba's algorithm (SAC, categorical critic, reward normalization). If the architecture-only swap reproduces most of the performance gap at 1 million steps, the central claim holds; if the algorithm-only swap does, it fails. A second check: for every task where V-Simba wins, compute the top Hessian eigenvalue of both trained critics; the proposed mechanism predicts V-Simba's landscape is consistently flatter.

Watch

Extended reading notes

Core claim

The paper identifies that the standard visual-RL architecture popularized by DrQ-v2 — a shallow convolutional encoder with a single normalization layer — develops severe training instabilities: the loss landscape becomes sharp, a growing fraction of neurons go dormant, features collapse toward low diversity, and feature, parameter, and gradient norms grow without bound. It proposes V-Simba, which counters these instabilities by placing layer normalization at the input and after every block, adding $\ell^2$ weight regularization, and replacing the mean-squared-error critic with a categorical distributional critic trained by KL divergence under reward normalization. To make this extra machinery affordable on high-resolution pixel inputs, V-Simba applies an early stride-2 convolution, uses 2×2 max-pooling for downsampling, and builds its convolutional blocks from lightweight pointwise (1×1) convolutions in an inverted-bottleneck residual layout. The paper reports that this architecture, without any new exploration strategy, world model, or auxiliary task, matches or outperforms DrQ-v2, DrM, TACO, TD-MPC2, MR.Q, and A-LIX across 18 DMC tasks, 3 sparse Adroit tasks, and 8 Meta-World tasks at 1 million environment steps, and that it reduces training time from 5.4 to 4.8 hours per 1M DMC steps.

Load-bearing premise

The performance gains are attributed to the architecture itself, but the headline benchmark comparisons change the base algorithm together with the network, so the claim that architecture alone is responsible rests on the assumption that the concurrent switch to SAC with a distributional critic and reward normalization is not driving most of the improvement.

Editorial extensions

If this is right

  • Adopting V-Simba as a drop-in replacement for DrQ-v2's encoder and heads should improve sample efficiency for any visual continuous-control algorithm that currently uses the standard shallow CNN.
  • Because V-Simba uses a single hyperparameter set across 29 tasks, its design reduces the per-task tuning burden that many algorithmic baselines require.
  • The diagnostic metrics — sharpness, dormant ratio, feature diversity, and norm growth — offer a concrete checklist for evaluating future visual RL architectures, even though no single metric is predictive alone.
  • V-Simba's lower parameter count and training time mean that the same norm-control approach can be scaled to wider or deeper networks without incurring the computational cost that previously made such scaling prohibitive for pixel inputs.
  • The success of transferring Simba-style principles from state-based to visual RL suggests that further state-based architectural advances may also carry over to pixel domains.

Reading between the lines

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

  • The headline comparisons mix an architecture change with a base-algorithm change (SAC + distributional critic + reward normalization versus DDPG + MSE critic), so the cleanest reading is that architecture drives a large share of the gain, while the full margin may also depend on the algorithmic switches.
  • The same norm-control recipe could be tested on other high-dimensional observation regimes, such as multi-camera or partially observable robotic inputs, where the standard CNN encoder likely exhibits similar instability.
  • Since the paper notes sharpness aligns most consistently with performance ordering, a direct testable extension is whether a sharpness-reduction method on the standard DrQ-v2 architecture would close a substantial part of the gap without architectural change.
  • A natural stress test for the generality claim is to attach V-Simba's encoder to model-based methods such as TD-MPC2 or to offline datasets, where representation stability matters differently.
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 V-Simba, an architecture for visual continuous control built on top of SAC with data augmentation. The architecture adds layer normalization, residual connections, pointwise convolutions, early downsampling, weight decay, a distributional (categorical) critic, and reward normalization. The authors claim that V-Simba matches or outperforms state-of-the-art methods across DMC, Adroit, and Meta-World (29 tasks) with a single hyperparameter set, while being more computationally efficient than DrQ-v2, and they provide diagnostic metrics (loss sharpness, dormant ratio, feature diversity, norm growth) and an ablation study to support the design. The paper includes controlled architecture comparisons on Dog Stand under both DDPG and SAC.

Significance. If the results hold, the paper would demonstrate that a carefully designed convolutional architecture can yield large sample-efficiency gains in visual continuous control without complex algorithmic add-ons, offering a simple, reusable component for the community. The paper ships public code and provides an unusually detailed stability analysis of common visual RL encoders, which is a useful contribution regardless of the headline claim. The ablation study gives explicit evidence that each component contributes to the final performance. However, the central claim that 'architectural design alone' drives the benchmark gains is not cleanly supported by the main experiments because the algorithm changes simultaneously with the architecture.

major comments (3)
  1. [Section 5.2 (DMC Medium paragraph) and Figures 5–8] The headline benchmark comparisons vary the RL algorithm and the architecture at the same time. V-Simba is built on SAC (Section 4), while DrQ-v2 uses DDPG, and V-Simba further adds a categorical distributional critic and reward normalization (Section 4.1). The sentence 'simply replacing DrQ-v2’s neural network with our proposed architecture' is therefore misleading, because the learning algorithm also changes. The controlled algorithm-fixed comparisons in Figure 2 and Appendix 7.3 provide supporting evidence on Dog Stand, but only on a single environment. As a result, the abstract and Contribution 3 claim that 'architectural design alone' achieves superior performance across 29 tasks is not established by the main benchmark results; the claim needs either a fixed-algorithm comparison across a wider task set or a more careful wording that attributes the gains to the combined architectural and algorithmic package.
  2. [Contribution 2/3 and Section 4.1] The paper describes the distributional critic and reward normalization as part of the V-Simba architecture, but these are algorithmic components rather than architectural changes in the usual sense. Contribution 3 states that 'robust architectural design alone can achieve superior sample and compute efficiency without introducing complex algorithmic add-ons,' which overstates the purity of the architectural contribution. The categorical critic and reward scaling are value-learning modifications, not neural-network layout choices, and the abstract’s emphasis on architecture should be reconciled with this.
  3. [Abstract and Section 5.2 (DMC Hard)] The abstract claims V-Simba 'matches or outperforms' state-of-the-art methods, but the DMC Hard paragraph states only that V-Simba 'performs competitively with MR.Q,' which is a weaker claim. The paper should either present quantitative aggregates showing that V-Simba matches or exceeds all baselines across the full benchmark, or soften the abstract to 'is competitive with' where appropriate.
minor comments (6)
  1. [Section 5.2] The phrase 'our base algorithm, DrQ-v2' is incorrect: the base algorithm is SAC, while DrQ-v2 is a DDPG-based method. Please rephrase to avoid confusion.
  2. [Section 4.1] 'We first opt LayerNorm' should read 'We first adopt LayerNorm'.
  3. [Table 1 caption] 'highlighted base on their relative percentile difference' contains a typo: 'base' should be 'based'.
  4. [References] The Burda et al. exploration paper appears twice (2018a and 2018b) with identical content; please consolidate.
  5. [Appendix 7.2] The definition of the dormant ratio is typeset in a garbled way; please reformat the equation for readability.
  6. [Figure 2 caption] The caption says both architectures are evaluated on SAC, but the surrounding text also discusses DrQ-v2's default DDPG. Please clarify in the caption that this is a controlled comparison holding the algorithm fixed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: empirical architecture study evaluated against external baselines, with a non-circular attribution confound in the headline comparisons.

full rationale

This is an empirical architecture paper, not a derivation, so the circularity burden is low. The central comparisons are against externally developed baselines (DrQ-v2, MR.Q, TD-MPC2, DrM, TACO, A-LIX) on standard benchmarks with official or reported numbers, and the ablations vary one component at a time. The paper's diagnostic metrics are explicitly labeled non-predictive ('We stress that these metrics are diagnostic rather than predictive...'), which removes the risk that the metrics are fitted to the final returns. The design borrows from the authors' prior Simba and SimbaV2 work, but those citations motivate the architecture rather than prove the benchmark result; the benchmark result is obtained by running the method. The one substantive concern is not circularity: Section 5.2's statement that V-Simba is 'simply replacing DrQ-v2's neural network' is imprecise because V-Simba also changes the base learner from DDPG to SAC and adds a distributional critic and reward normalization, so the headline comparisons vary algorithm and architecture together. That is an experimental attribution/validity issue, not a case where a prediction is equivalent to its input by construction. No fitted parameter is renamed as a prediction and no uniqueness theorem or definitional identity forces the reported gains.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical or mathematical entities are introduced. The contribution is an architecture whose design rests on established normalization properties, benchmark representativeness, and diagnostic validity. The free parameters are hyperparameters of the architecture, not fitted to a target result.

free parameters (5)
  • weight_decay = 1e-2
    L2 regularization strength chosen to constrain parameter growth (Table 2). It is a hand-selected hyperparameter central to the stabilization principle.
  • return_support = [-5, 5]
    Bounds for the distributional critic's support, listed in Table 2. This ad hoc choice defines the range of normalized returns and affects the Q-value distribution.
  • encoder_hidden_channels = 32
    Number of channels in encoder blocks (Table 2). Hand-chosen width that balances capacity and compute.
  • num_atoms = 101
    Number of atoms in the categorical critic (Table 2). Chosen by hand, larger than the original C51 style and impacts gradient stability.
  • learning_rate = 1e-4
    Adam learning rate in Table 2. A standard but nonetheless hand-selected value that the results depend on.
assumptions (4)
  • standard math LayerNorm scale invariance and inverse gradient scaling (Eqs. 3 and 4) accurately describe the optimization dynamics of the proposed network.
    Invoked in Section 4.1 to motivate weight regularization. These properties are established from prior literature (Lyle et al., 2024a; Lee et al., 2025) and are not re-derived here.
  • domain assumption DMC, Adroit, and Meta-World at 1M environment steps are representative of the visual continuous control problem class.
    The paper evaluates only these benchmarks and generalizes its conclusions to visual RL broadly. Standard practice, but an assumption nonetheless.
  • domain assumption Sharpness of the loss landscape, dormant ratio, and feature diversity are valid diagnostic indicators of optimization instability and capacity loss.
    Section 7.2 defines these metrics and the paper itself stresses they are 'diagnostic rather than predictive'. The architectural design is motivated by these metrics, so the causal link is assumed.
  • ad hoc to paper A single fixed hyperparameter set generalizes across all 29 tasks without per-task tuning.
    Section 5.1 states the method uses the same hyperparameters everywhere. The paper does not describe how this set was selected or whether it was tuned on a subset of tasks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of V-Simba: Unleashing the Architectural Potential of RL in Visual Continuous Control." pith.science (2026). https://pith.science/paper/KW72DWTG

@misc{pith2026260807870,
  author       = {Pith},
  title        = {Pith review of: V-Simba: Unleashing the Architectural Potential of RL in Visual Continuous Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KW72DWTG}},
  note         = {Machine review of arXiv:2608.07870}
}
read the original abstract

Improving sample efficiency remains a core challenge in reinforcement learning (RL), especially in real-world settings like robotics, where data collection is costly. This challenge is pronounced in visual RL, where high-dimensional inputs often obscure learning signals. While prior work in visual RL has focused on algorithmic solutions, such as better dynamics models or exploration strategies, recent advances in state-based RL show that architectural design alone can lead to significant gains in sample efficiency. This raises an important question: Can these architectural principles transfer to visual RL? In response, we introduce V-Simba, a simple yet effective visual RL architecture inspired by the Simba architecture from state-based RL. Built on top of Soft Actor-Critic (SAC) with data augmentation, V-Simba modifies the architecture by adding normalization layers to stabilize training and using pointwise convolutions to reduce computation. Despite its simplicity, V-Simba matches or outperforms the state-of-the-art methods across the DMC, Adroit, and Meta-World benchmarks, while being more computationally efficient than DrQ-v2. We make our code publicly available at https://github.com/DAVIAN-Robotics/V-Simba.

Figures

Figures reproduced from arXiv: 2608.07870 by the authors.

Figure 1
Figure 1. Benchmark Summary. We evaluate the effectiveness of V-Simba across 29 visual contin￾uous control tasks spanning multiple domains, with a single set of hyperparameters. By incorporat￾ing V-Simba into Soft Actor-Critic with data augmentation, it matches or outperforms state-of-the￾art visual RL methods, demonstrating better sample and compute efficiency. 1 Introduction Deep reinforcement learning (RL) has long been a … view at source ↗
Figure 2
Figure 2. DrQ-v2 vs. V-Simba. Comparison of the DrQ-v2 architecture and our V-Simba in the Dog Stand environment. Both are evaluated on Soft Actor-Critic (SAC) with results averaged over 5 seeds. (a) V-Simba is substantially more efficient than DrQ-v2 in terms of both sample and compute. (b) Unlike DrQ-v2, V-Simba has stable learning dynamics, indicated by a smooth loss landscape, a low dormant ratio, high feature diversity, … view at source ↗
Figure 3
Figure 3. V-Simba architecture. We aim to develop an architecture that constrains its feature norm, weight norm, and gradient norm for better stability and generalization. Precisely, we make extensive use of layer normalization and residual connections for stable features and gradients, and incorporate point-wise convolutions in tandem with spatial convolutions for computational efficiency. 4.2 Design Philosophy II: Maintaini… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Environment Visualization. We evaluate our V-Simba on 3 visual continuous control benchmarks: DeepMind Control Suite (Tassa et al., 2018), Adroit (Rajeswaran et al., 2017), and Meta-World (Yu et al., 2020) [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: DeepMind Control Suite - Hard. Average episode returns on 7 hard-level tasks from DeepMind Control Suite (Tassa et al., 2018). Each curve represents the mean performance across 5 random seeds per algorithm; shaded areas indicate 95% bootstrap confidence intervals. leve…
Figure 6
Figure 6. Figure 6: DeepMind Control Suite - Medium. Average episode returns on 11 medium-level tasks from DeepMind Control Suite (Tassa et al., 2018). Each curve represents the mean performance across 5 random seeds per algorithm; shaded areas indicate 95% bootstrap confidence intervals.…
Figure 7
Figure 7. Figure 7: Adroit - Sparse. Average success rates on 3 sparse-reward tasks from Adroit (Rajeswaran et al., 2017). Each curve represents the mean performance across 5 random seeds per algorithm; shaded areas indicate 95% bootstrap confidence intervals. 0 0.2 0.4 0.6 0.8 1.0 0 25 5…
Figure 8
Figure 8. Figure 8: Meta-World. Average success rates on 8 tasks from the Meta-World (Yu et al., 2020). Each curve represents the mean performance across 5 random seeds per algorithm; shaded areas indicate 95% bootstrap confidence intervals [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Comparison of Neural Architectures under DDPG. We evaluate and compare the neural network architectures proposed by DrQ-v2, IMPALA, MR.Q and our V-Simba, using DDPG with data augmentation in the Dog Stand environment. V-Simba maintains greater stability throughout trai…
Figure 10
Figure 10. Figure 10: Comparison of Neural Architectures under SAC. We evaluate and compare the neural network architectures proposed by DrQ-v2, IMPALA, MR.Q and our V-Simba, using SAC with data augmentation in the Dog Stand environment. V-Simba maintains greater stability throughout train…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

294 extracted references · 6 canonical work pages

  1. [1]

    Reinforcement Learning Conference , year=

    On the consistency of hyper-parameter selection in value-based deep reinforcement learning , author=. Reinforcement Learning Conference , year=

  2. [2]

    Forty-first International Conference on Machine Learning , year=

    In value-based deep reinforcement learning, a pruned network is a good network , author=. Forty-first International Conference on Machine Learning , year=

  3. [3]

    1995 , publisher=

    Python reference manual , author=. 1995 , publisher=

    1995

  4. [4]

    Nature , volume=

    Array programming with NumPy , author=. Nature , volume=. 2020 , publisher=

    2020

  5. [5]

    Computing in science & engineering , volume=

    Matplotlib: A 2D graphics environment , author=. Computing in science & engineering , volume=. 2007 , publisher=

    2007

  6. [6]

    JAX: composable transformations of Python+ NumPy programs , author=

  7. [7]

    , journal=

    Oliphant, Travis E. , journal=. Python for Scientific Computing , year=

  8. [8]

    IOS Press , year = 2016, pages =

    Jupyter Notebooks a publishing format for reproducible computational workflows. IOS Press , year = 2016, pages =. doi:10.3233/978-1-61499-649-1-87 , adsurl =

    doi:10.3233/978-1-61499-649-1-87 2016

Show all 294 references
  1. [9]

    Python for Data Analysis: Data Wrangling with Pandas,

    McKinney, Wes , biburl =. Python for Data Analysis: Data Wrangling with Pandas,

  2. [10]

    Forty-second International Conference on Machine Learning , year=

    The Impact of On-Policy Parallelized Data Collection on Deep Reinforcement Learning Networks , author=. Forty-second International Conference on Machine Learning , year=

  3. [11]

    The Thirteenth International Conference on Learning Representations , year=

    Neuroplastic Expansion in Deep Reinforcement Learning , author=. The Thirteenth International Conference on Learning Representations , year=

  4. [12]

    Mixture of Experts in a Mixture of

    Timon Willi and Johan Samir Obando Ceron and Jakob Nicolaus Foerster and Gintare Karolina Dziugaite and Pablo Samuel Castro , booktitle=. Mixture of Experts in a Mixture of. 2024 , url=

    2024

  5. [13]

    Forty-third International Conference on Machine Learning , year=

    Stable Deep Reinforcement Learning via Isotropic Gaussian Representations , author=. Forty-third International Conference on Machine Learning , year=

  6. [14]

    The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

    Stable Gradients for Stable Learning at Scale in Deep Reinforcement Learning , author=. The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

  7. [15]

    Advances in Neural Information Processing Systems , volume=

    Measure gradients, not activations! enhancing neuronal activity in deep reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  8. [16]

    arXiv preprint arXiv:2506.17518 , year=

    A survey of state representation learning for deep reinforcement learning , author=. arXiv preprint arXiv:2506.17518 , year=

    arXiv

  9. [17]

    International Conference on Learning Representations , volume=

    Simplicial embeddings improve sample efficiency in actor--critic agents , author=. International Conference on Learning Representations , volume=

  10. [18]

    arXiv preprint arXiv:2210.04561 , year=

    A comprehensive survey of data augmentation in visual reinforcement learning , author=. arXiv preprint arXiv:2210.04561 , year=

    arXiv

  11. [19]

    MENTOR: Mixture-of-Experts Network with Task-Oriented Perturbation for Visual Reinforcement Learning , author=

  12. [20]

    Advances in Neural Information Processing Systems , volume=

    TACO: Temporal Latent Action-Driven Contrastive Loss for Visual Reinforcement Learning , author=. Advances in Neural Information Processing Systems , volume=

  13. [21]

    International conference on machine learning , pages=

    Decoupling representation learning from reinforcement learning , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  14. [22]

    arXiv preprint arXiv:2501.16142 , year=

    Towards General-Purpose Model-Free Reinforcement Learning , author=. arXiv preprint arXiv:2501.16142 , year=

    arXiv

  15. [23]

    arXiv preprint arXiv:2007.05929 , year=

    Data-efficient reinforcement learning with self-predictive representations , author=. arXiv preprint arXiv:2007.05929 , year=

    2007 arXiv

  16. [24]

    Advances in Neural Information Processing Systems , volume=

    Pretraining representations for data-efficient reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  17. [25]

    2016 IEEE/RSJ international conference on intelligent robots and systems (IROS) , pages=

    Stable reinforcement learning with autoencoders for tactile and visual data , author=. 2016 IEEE/RSJ international conference on intelligent robots and systems (IROS) , pages=. 2016 , organization=

    2016

  18. [26]

    International conference on machine learning , pages=

    Deepmdp: Learning continuous latent space models for representation learning , author=. International conference on machine learning , pages=. 2019 , organization=

    2019

  19. [27]

    Advances in Neural Information Processing Systems , volume=

    Stochastic latent actor-critic: Deep reinforcement learning with a latent variable model , author=. Advances in Neural Information Processing Systems , volume=

  20. [28]

    Advances in Neural Information Processing Systems , volume=

    Playvirtual: Augmenting cycle-consistent virtual trajectories for reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  21. [29]

    Advances in Neural Information Processing Systems , volume=

    Mask-based latent reconstruction for reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  22. [30]

    International Conference on Machine Learning , pages=

    A deep reinforcement learning approach to marginalized importance sampling with the successor representation , author=. International Conference on Machine Learning , pages=. 2021 , organization=

    2021

  23. [31]

    arXiv preprint arXiv:2110.04935 , year=

    Learning temporally-consistent representations for data-efficient reinforcement learning , author=. arXiv preprint arXiv:2110.04935 , year=

    arXiv

  24. [32]

    International Conference on Machine Learning , pages=

    Reinforcement learning with action-free pre-training from videos , author=. International Conference on Machine Learning , pages=. 2022 , organization=

    2022

  25. [33]

    , author=

    Self-Predictive Dynamics for Generalization of Vision-based Reinforcement Learning. , author=. IJCAI , pages=

  26. [34]

    arXiv preprint arXiv:2401.08898 , year=

    Bridging state and history representations: Understanding self-predictive rl , author=. arXiv preprint arXiv:2401.08898 , year=

    arXiv

  27. [35]

    arXiv preprint arXiv:2502.03550 , year=

    TD-M (PC) ^2 : Improving Temporal Difference MPC Through Policy Constraint , author=. arXiv preprint arXiv:2502.03550 , year=

    arXiv

  28. [36]

    Conference on robot learning , pages=

    Daydreamer: World models for physical robot learning , author=. Conference on robot learning , pages=. 2023 , organization=

    2023

  29. [37]

    arXiv preprint arXiv:1803.10122 , year=

    World models , author=. arXiv preprint arXiv:1803.10122 , year=

    arXiv

  30. [38]

    2016 IEEE International Conference on Robotics and Automation (ICRA) , pages=

    Deep spatial autoencoders for visuomotor learning , author=. 2016 IEEE International Conference on Robotics and Automation (ICRA) , pages=. 2016 , organization=

    2016

  31. [39]

    Advances in neural information processing systems , volume=

    Embed to control: A locally linear latent dynamics model for control from raw images , author=. Advances in neural information processing systems , volume=

  32. [40]

    Advances in neural information processing systems , volume=

    Reinforcement learning with augmented data , author=. Advances in neural information processing systems , volume=

  33. [41]

    Advances in neural information processing systems , volume=

    Byol-explore: Exploration by bootstrapped prediction , author=. Advances in neural information processing systems , volume=

  34. [42]

    arXiv preprint arXiv:2310.07220 , year=

    Coplanner: Plan to roll out conservatively but to explore optimistically for model-based rl , author=. arXiv preprint arXiv:2310.07220 , year=

    arXiv

  35. [43]

    International conference on machine learning , pages=

    Curiosity-driven exploration by self-supervised prediction , author=. International conference on machine learning , pages=. 2017 , organization=

    2017

  36. [44]

    International conference on machine learning , pages=

    Planning to explore via self-supervised world models , author=. International conference on machine learning , pages=. 2020 , organization=

    2020

  37. [45]

    arXiv preprint arXiv:2002.06038 , year=

    Never give up: Learning directed exploration strategies , author=. arXiv preprint arXiv:2002.06038 , year=

    2002 arXiv

  38. [46]

    Advances in Neural Information Processing Systems , volume=

    Accelerating reinforcement learning with value-conditional state entropy exploration , author=. Advances in Neural Information Processing Systems , volume=

  39. [47]

    arXiv preprint arXiv:2412.12098 , year=

    MaxInfoRL: Boosting exploration in reinforcement learning through information gain maximization , author=. arXiv preprint arXiv:2412.12098 , year=

    arXiv

  40. [48]

    nature , volume=

    Human-level control through deep reinforcement learning , author=. nature , volume=. 2015 , publisher=

    2015

  41. [49]

    Proceedings of the aaai conference on artificial intelligence , volume=

    Improving sample efficiency in model-free reinforcement learning from images , author=. Proceedings of the aaai conference on artificial intelligence , volume=

  42. [50]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Convnext v2: Co-designing and scaling convnets with masked autoencoders , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  43. [51]

    arXiv preprint arXiv:2410.09754 , year=

    SimBa: Simplicity Bias for Scaling Up Parameters in Deep Reinforcement Learning , author=. arXiv preprint arXiv:2410.09754 , year=

    arXiv

  44. [52]

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

    Neural Redshift: Random Networks are not Random Functions , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  45. [53]

    Proceedings of Thirty Third Conference on Learning Theory , pages =

    Kernel and Rich Regimes in Overparametrized Models , author =. Proceedings of Thirty Third Conference on Learning Theory , pages =. 2020 , editor =

    2020

  46. [54]

    Proceedings of Thirty Third Conference on Learning Theory , pages =

    Implicit regularization for deep neural networks driven by an Ornstein-Uhlenbeck like process , author =. Proceedings of Thirty Third Conference on Learning Theory , pages =. 2020 , editor =

    2020

  47. [55]

    International conference on machine learning , pages=

    On the optimization of deep networks: Implicit acceleration by overparameterization , author=. International conference on machine learning , pages=. 2018 , organization=

    2018

  48. [56]

    Advances in Neural Information Processing Systems , volume=

    Implicit regularization in deep matrix factorization , author=. Advances in Neural Information Processing Systems , volume=

  49. [57]

    2022 , url=

    Aviral Kumar and Rishabh Agarwal and Tengyu Ma and Aaron Courville and George Tucker and Sergey Levine , booktitle=. 2022 , url=

    2022

  50. [58]

    International Conference on Learning Representations , year=

    Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning , author=. International Conference on Learning Representations , year=

  51. [59]

    Proceedings of the 36th International Conference on Machine Learning , pages =

    On the Spectral Bias of Neural Networks , author =. Proceedings of the 36th International Conference on Machine Learning , pages =. 2019 , editor =

    2019

  52. [60]

    arXiv preprint arXiv:2502.15280 , year=

    Hyperspherical Normalization for Scalable Deep Reinforcement Learning , author=. arXiv preprint arXiv:2502.15280 , year=

    arXiv

  53. [61]

    Advances in neural information processing systems , volume=

    Sgd on neural networks learns functions of increasing complexity , author=. Advances in neural information processing systems , volume=

  54. [62]

    arXiv preprint arXiv:2006.05826 , year=

    Transient non-stationarity and generalisation in deep reinforcement learning , author=. arXiv preprint arXiv:2006.05826 , year=

    2006 arXiv

  55. [63]

    arXiv preprint arXiv:2206.01078 , year=

    Deep transformer q-networks for partially observable reinforcement learning , author=. arXiv preprint arXiv:2206.01078 , year=

    arXiv

  56. [64]

    arXiv preprint arXiv:2302.01327 , year=

    Dual patchnorm , author=. arXiv preprint arXiv:2302.01327 , year=

    arXiv

  57. [65]

    Neural Networks: Tricks of the trade , pages=

    Early stopping-but when? , author=. Neural Networks: Tricks of the trade , pages=. 2002 , publisher=

    2002

  58. [66]

    Advances in neural information processing systems , volume=

    On warm-starting neural network training , author=. Advances in neural information processing systems , volume=

  59. [67]

    Neural computation , volume=

    Toward training recurrent neural networks for lifelong learning , author=. Neural computation , volume=. 2020 , publisher=

    2020

  60. [68]

    Proceedings of the IEEE international conference on computer vision , pages=

    Delving deep into rectifiers: Surpassing human-level performance on imagenet classification , author=. Proceedings of the IEEE international conference on computer vision , pages=

  61. [69]

    Advances in Neural Information Processing Systems , volume=

    Gradient starvation: A learning proclivity in neural networks , author=. Advances in Neural Information Processing Systems , volume=

  62. [70]

    The Primacy Bias in Deep Reinforcement Learning , author=

  63. [71]

    Understanding and preventing capacity loss in reinforcement learning , author=

  64. [72]

    arXiv preprint arXiv:2303.07507 , year=

    Loss of Plasticity in Continual Deep Reinforcement Learning , author=. arXiv preprint arXiv:2303.07507 , year=

    arXiv

  65. [73]

    international conference on machine learning , pages=

    Understanding and improving convolutional neural networks via concatenated rectified linear units , author=. international conference on machine learning , pages=. 2016 , organization=

    2016

  66. [74]

    arXiv preprint arXiv:2006.05990 , year=

    What matters in on-policy reinforcement learning? a large-scale empirical study , author=. arXiv preprint arXiv:2006.05990 , year=

    2006 arXiv

  67. [75]

    Understanding plasticity in neural networks , author=

  68. [76]

    arXiv preprint arXiv:2304.10466 , year=

    Efficient Deep Reinforcement Learning Requires Regulating Overfitting , author=. arXiv preprint arXiv:2304.10466 , year=

    arXiv

  69. [77]

    The Twelfth International Conference on Learning Representations , year=

    CrossQ: Batch Normalization in Deep Reinforcement Learning for Greater Sample Efficiency and Simplicity , author=. The Twelfth International Conference on Learning Representations , year=

  70. [78]

    The Eleventh International Conference on Learning Representations , year=

    Sample-efficient reinforcement learning by breaking the replay ratio barrier , author=. The Eleventh International Conference on Learning Representations , year=

  71. [79]

    International Conference on Machine Learning , pages=

    Bigger, better, faster: Human-level atari with human-level efficiency , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  72. [80]

    arXiv preprint arXiv:2308.11958 , year=

    Maintaining plasticity via regenerative regularization , author=. arXiv preprint arXiv:2308.11958 , year=

    arXiv

  73. [81]

    arXiv preprint arXiv:2306.13812 , year=

    Maintaining Plasticity in Deep Continual Learning , author=. arXiv preprint arXiv:2306.13812 , year=

    arXiv

  74. [82]

    arXiv preprint arXiv:2312.00246 , year=

    Curvature Explains Loss of Plasticity , author=. arXiv preprint arXiv:2312.00246 , year=

    arXiv

  75. [83]

    arXiv preprint arXiv:2307.04345 , year=

    Continual learning as computationally constrained reinforcement learning , author=. arXiv preprint arXiv:2307.04345 , year=

    arXiv

  76. [84]

    Proceedings of the thirteenth international conference on artificial intelligence and statistics , pages=

    Understanding the difficulty of training deep feedforward neural networks , author=. Proceedings of the thirteenth international conference on artificial intelligence and statistics , pages=. 2010 , organization=

    2010

  77. [85]

    arXiv preprint arXiv:1912.02178 , year=

    Fantastic generalization measures and where to find them , author=. arXiv preprint arXiv:1912.02178 , year=

    1912 arXiv

  78. [86]

    International Conference on Machine Learning , pages=

    Disentangling trainability and generalization in deep neural networks , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  79. [87]

    arXiv preprint arXiv:1409.1556 , year=

    Very deep convolutional networks for large-scale image recognition , author=. arXiv preprint arXiv:1409.1556 , year=

    arXiv

  80. [88]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Deep residual learning for image recognition , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  81. [89]

    arXiv preprint arXiv:2010.11929 , year=

    An image is worth 16x16 words: Transformers for image recognition at scale , author=. arXiv preprint arXiv:2010.11929 , year=

    2010 arXiv

  82. [90]

    arXiv preprint arXiv:2303.08774 , year=

    GPT-4 Technical Report , author=. arXiv preprint arXiv:2303.08774 , year=

    arXiv

  83. [91]

    arXiv preprint arXiv:2312.11805 , year=

    Gemini: a family of highly capable multimodal models , author=. arXiv preprint arXiv:2312.11805 , year=

    arXiv

  84. [92]

    2009 , institution=

    Learning multiple layers of features from tiny images , author=. 2009 , institution=

    2009

  85. [93]

    http://yann.lecun.com/exdb/mnist/ , year=

    The MNIST database of handwritten digits , author=. http://yann.lecun.com/exdb/mnist/ , year=

  86. [94]

    arXiv preprint arXiv:1711.05101 , year=

    Decoupled weight decay regularization , author=. arXiv preprint arXiv:1711.05101 , year=

    arXiv

  87. [95]

    arXiv preprint arXiv:1706.02677 , year=

    Accurate, large minibatch sgd: Training imagenet in 1 hour , author=. arXiv preprint arXiv:1706.02677 , year=

    arXiv

  88. [96]

    International conference on machine learning , pages=

    Batch normalization: Accelerating deep network training by reducing internal covariate shift , author=. International conference on machine learning , pages=. 2015 , organization=

    2015

  89. [97]

    arXiv preprint arXiv:1607.06450 , year=

    Layer normalization , author=. arXiv preprint arXiv:1607.06450 , year=

    arXiv

  90. [98]

    arXiv preprint arXiv:2306.17833 , year=

    Resetting the optimizer in deep RL: An empirical study , author=. arXiv preprint arXiv:2306.17833 , year=

    arXiv

  91. [99]

    International conference on machine learning , pages=

    Training data-efficient image transformers & distillation through attention , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  92. [100]

    International journal of computer vision , volume=

    Imagenet large scale visual recognition challenge , author=. International journal of computer vision , volume=. 2015 , publisher=

    2015

  93. [101]

    Journal of mathematics and mechanics , pages=

    A Markovian decision process , author=. Journal of mathematics and mechanics , pages=. 1957 , publisher=

    1957

  94. [102]

    arXiv preprint arXiv:1509.02971 , year=

    Continuous control with deep reinforcement learning , author=. arXiv preprint arXiv:1509.02971 , year=

    arXiv

  95. [103]

    arXiv preprint arXiv:2405.16158 , year=

    Bigger, Regularized, Optimistic: scaling for compute and sample-efficient continuous control , author=. arXiv preprint arXiv:2405.16158 , year=

    arXiv

  96. [104]

    International conference on machine learning , pages=

    Impala: Scalable distributed deep-rl with importance weighted actor-learner architectures , author=. International conference on machine learning , pages=. 2018 , organization=

    2018

  97. [105]

    IEEE Transactions on Circuits and Systems for Video Technology , volume=

    Data augmentation using random image cropping and patching for deep CNNs , author=. IEEE Transactions on Circuits and Systems for Video Technology , volume=. 2019 , publisher=

    2019

  98. [106]

    Artificial intelligence and statistics , pages=

    Communication-efficient learning of deep networks from decentralized data , author=. Artificial intelligence and statistics , pages=. 2017 , organization=

    2017

  99. [107]

    Advances in neural information processing systems , volume=

    Continual learning with deep generative replay , author=. Advances in neural information processing systems , volume=

  100. [108]

    Trends in cognitive sciences , volume=

    What learning systems do intelligent agents need? Complementary learning systems theory updated , author=. Trends in cognitive sciences , volume=. 2016 , publisher=

    2016

  101. [109]

    International conference on machine learning , pages=

    Neural episodic control , author=. International conference on machine learning , pages=. 2017 , organization=

    2017

  102. [110]

    arXiv preprint arXiv:2201.12604 , year=

    Learning fast, learning slow: A general continual learning method based on complementary learning system , author=. arXiv preprint arXiv:2201.12604 , year=

    arXiv

  103. [111]

    arXiv preprint arXiv:2202.11672 , year=

    Learning fast and slow for online time series forecasting , author=. arXiv preprint arXiv:2202.11672 , year=

    arXiv

  104. [112]

    , author=

    Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory. , author=. Psychological review , volume=. 1995 , publisher=

    1995

  105. [113]

    Advances in neural information processing systems , volume=

    A simple weight decay can improve generalization , author=. Advances in neural information processing systems , volume=

  106. [114]

    arXiv preprint arXiv:2001.08361 , year=

    Scaling laws for neural language models , author=. arXiv preprint arXiv:2001.08361 , year=

    2001 arXiv

  107. [115]

    International Conference on Learning Representations , year=

    Two-Timescale Networks for Nonlinear Value Function Approximation , author=. International Conference on Learning Representations , year=

  108. [116]

    arXiv preprint arXiv:2106.00042 , year=

    A study on the plasticity of neural networks , author=. arXiv preprint arXiv:2106.00042 , year=

    arXiv

  109. [117]

    arXiv preprint arXiv:2107.09645 , year=

    Mastering visual continuous control: Improved data-augmented reinforcement learning , author=. arXiv preprint arXiv:2107.09645 , year=

    arXiv

  110. [118]

    arXiv preprint arXiv:2207.00986 , year=

    Stabilizing off-policy deep reinforcement learning from pixels , author=. arXiv preprint arXiv:2207.00986 , year=

    arXiv

  111. [119]

    International Conference on Machine Learning , pages=

    The dormant neuron phenomenon in deep reinforcement learning , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  112. [120]

    arXiv preprint arXiv:2310.19668 , year=

    DrM: Mastering Visual Reinforcement Learning through Dormant Ratio Minimization , author=. arXiv preprint arXiv:2310.19668 , year=

    arXiv

  113. [121]

    arXiv preprint arXiv:2302.03281 , year=

    Utility-based Perturbed Gradient Descent: An Optimizer for Continual Learning , author=. arXiv preprint arXiv:2302.03281 , year=

    arXiv

  114. [122]

    Communications of the ACM , volume=

    Temporal difference learning and TD-Gammon , author=. Communications of the ACM , volume=

  115. [123]

    2018 , publisher=

    Reinforcement learning: An introduction , author=. 2018 , publisher=

    2018

  116. [124]

    Advances in neural information processing systems , volume=

    Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results , author=. Advances in neural information processing systems , volume=

  117. [125]

    arXiv preprint arXiv:2310.16828 , year=

    Td-mpc2: Scalable, robust world models for continuous control , author=. arXiv preprint arXiv:2310.16828 , year=

    arXiv

  118. [126]

    International Conference on Machine Learning , pages=

    Spectral normalisation for deep reinforcement learning: an optimisation perspective , author=. International Conference on Machine Learning , pages=. 2021 , organization=

    2021

  119. [127]

    Advances in Neural Information Processing Systems , year=

    Deep Reinforcement Learning at the Edge of the Statistical Precipice , author=. Advances in Neural Information Processing Systems , year=

  120. [128]

    Proceedings on , pages=

    When Does Re-initialization Work? , author=. Proceedings on , pages=. 2023 , organization=

    2023

  121. [129]

    arXiv preprint arXiv:2004.13649 , year=

    Image augmentation is all you need: Regularizing deep reinforcement learning from pixels , author=. arXiv preprint arXiv:2004.13649 , year=

    2004 arXiv

  122. [130]

    Submitted to Transactions on Machine Learning Research , year=

    Exponential Moving Average of Weights in Deep Learning: Dynamics and Benefits , author=. Submitted to Transactions on Machine Learning Research , year=

  123. [131]

    arXiv preprint arXiv:2306.02451 , year=

    For SALE: State-Action Representation Learning for Deep Reinforcement Learning , author=. arXiv preprint arXiv:2306.02451 , year=

    arXiv

  124. [132]

    arXiv preprint arXiv:1512.08562 , year=

    Taming the noise in reinforcement learning via soft updates , author=. arXiv preprint arXiv:1512.08562 , year=

    arXiv

  125. [133]

    arXiv preprint arXiv:2312.11669 , year=

    Prediction and Control in Continual Reinforcement Learning , author=. arXiv preprint arXiv:2312.11669 , year=

    arXiv

  126. [134]

    Trends in neurosciences , volume=

    Hippocampal neurogenesis and forgetting , author=. Trends in neurosciences , volume=. 2013 , publisher=

    2013

  127. [135]

    Nature , volume=

    The forgotten part of memory , author=. Nature , volume=. 2019 , publisher=

    2019

  128. [136]

    Nature Reviews Neuroscience , volume=

    Forgetting as a form of adaptive engram cell plasticity , author=. Nature Reviews Neuroscience , volume=. 2022 , publisher=

    2022

  129. [137]

    Advances in Neural Information Processing Systems , volume=

    Dualnet: Continual learning, fast and slow , author=. Advances in Neural Information Processing Systems , volume=

  130. [138]

    Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

    Plasticity-Optimized Complementary Networks for Unsupervised Continual Learning , author=. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

  131. [139]

    arXiv preprint arXiv:1611.02779 , year=

    Rl ^2 : Fast reinforcement learning via slow reinforcement learning , author=. arXiv preprint arXiv:1611.02779 , year=

    arXiv

  132. [140]

    arXiv preprint arXiv:2202.06709 , year=

    How do vision transformers work? , author=. arXiv preprint arXiv:2202.06709 , year=

    arXiv

  133. [141]

    arXiv preprint arXiv:2106.01548 , year=

    When vision transformers outperform resnets without pre-training or strong data augmentations , author=. arXiv preprint arXiv:2106.01548 , year=

    arXiv

  134. [142]

    arXiv preprint arXiv:2012.09816 , year=

    Towards understanding ensemble, knowledge distillation and self-distillation in deep learning , author=. arXiv preprint arXiv:2012.09816 , year=

    2012 arXiv

  135. [143]

    Advances in neural information processing systems , volume=

    Towards deeper deep reinforcement learning with spectral normalization , author=. Advances in neural information processing systems , volume=

  136. [144]

    arXiv preprint arXiv:2207.02099 , year=

    An empirical study of implicit regularization in deep offline rl , author=. arXiv preprint arXiv:2207.02099 , year=

    arXiv

  137. [145]

    arXiv preprint arXiv:2108.06325 , year=

    Continual backprop: Stochastic gradient descent with persistent randomness , author=. arXiv preprint arXiv:2108.06325 , year=

    arXiv

  138. [146]

    Advances in Neural Information Processing Systems , volume=

    On plasticity, invariance, and mutually frozen weights in sequential task learning , author=. Advances in Neural Information Processing Systems , volume=

  139. [147]

    arXiv preprint arXiv:2312.07551 , year=

    Language Model Alignment with Elastic Reset , author=. arXiv preprint arXiv:2312.07551 , year=

    arXiv

  140. [148]

    arXiv preprint arXiv:2202.00155 , year=

    Fortuitous forgetting in connectionist networks , author=. arXiv preprint arXiv:2202.00155 , year=

    arXiv

  141. [149]

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

    A closer look at rehearsal-free continual learning , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  142. [150]

    Proceedings of the 2016 ACM SIGSAC conference on computer and communications security , pages=

    Deep learning with differential privacy , author=. Proceedings of the 2016 ACM SIGSAC conference on computer and communications security , pages=

    2016

  143. [151]

    arXiv preprint arXiv:1903.00374 , year=

    Model-based reinforcement learning for atari , author=. arXiv preprint arXiv:1903.00374 , year=

    1903 arXiv

  144. [152]

    arXiv preprint arXiv:2310.07996 , year=

    Reset It and Forget It: Relearning Last-Layer Weights Improves Continual and Transfer Learning , author=. arXiv preprint arXiv:2310.07996 , year=

    arXiv

  145. [153]

    arXiv preprint arXiv:2404.00781 , year=

    Addressing Loss of Plasticity and Catastrophic Forgetting in Continual Learning , author=. arXiv preprint arXiv:2404.00781 , year=

    arXiv

  146. [154]

    arXiv preprint arXiv:2402.18762 , year=

    Disentangling the Causes of Plasticity Loss in Neural Networks , author=. arXiv preprint arXiv:2402.18762 , year=

    arXiv

  147. [155]

    arXiv preprint arXiv:1904.00962 , year=

    Large batch optimization for deep learning: Training bert in 76 minutes , author=. arXiv preprint arXiv:1904.00962 , year=

    1904 arXiv

  148. [156]

    arXiv preprint arXiv:1412.6980 , year=

    Adam: A method for stochastic optimization , author=. arXiv preprint arXiv:1412.6980 , year=

    arXiv

  149. [157]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Rainbow: Combining improvements in deep reinforcement learning , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  150. [158]

    arXiv preprint arXiv:2403.00514 , year=

    Overestimation, Overfitting, and Plasticity in Actor-Critic: the Bitter Lesson of Reinforcement Learning , author=. arXiv preprint arXiv:2403.00514 , year=

    arXiv

  151. [159]

    Advances in Neural Information Processing Systems , volume=

    Small batch deep reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  152. [161]

    Information processing letters , volume=

    Occam's razor , author=. Information processing letters , volume=. 1987 , publisher=

    1987

  153. [162]

    International Conference on Machine Learning , pages=

    Scaling vision transformers to 22 billion parameters , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  154. [163]

    Journal of Statistical Mechanics: Theory and Experiment , volume=

    Deep double descent: Where bigger models and more data hurt , author=. Journal of Statistical Mechanics: Theory and Experiment , volume=

  155. [164]

    Conference on learning theory , pages=

    Implicit bias of gradient descent for wide two-layer neural networks trained with the logistic loss , author=. Conference on learning theory , pages=. 2020 , organization=

    2020

  156. [165]

    Advances in neural information processing systems , volume=

    Implicit bias of gradient descent on linear convolutional networks , author=. Advances in neural information processing systems , volume=

  157. [166]

    Advances in Neural Information Processing Systems , volume=

    Implicit bias of sgd for diagonal linear networks: a provable benefit of stochasticity , author=. Advances in Neural Information Processing Systems , volume=

  158. [167]

    IEEE transactions on neural networks and learning systems , volume=

    Why resnet works? residuals generalize , author=. IEEE transactions on neural networks and learning systems , volume=. 2020 , publisher=

    2020

  159. [168]

    On the foundations of shortcut learning , author=

  160. [169]

    arXiv preprint arXiv:2301.04104 , year=

    Mastering diverse domains through world models , author=. arXiv preprint arXiv:2301.04104 , year=

    arXiv

  161. [170]

    Advances in Neural Information Processing Systems , volume=

    Understanding, predicting and better resolving Q-value divergence in offline-RL , author=. Advances in Neural Information Processing Systems , volume=

  162. [171]

    arXiv preprint arXiv:2407.04811 , year=

    Simplifying Deep Temporal Difference Learning , author=. arXiv preprint arXiv:2407.04811 , year=

    arXiv

  163. [172]

    Advances in Neural Information Processing Systems , volume=

    Plastic: Improving input and label plasticity for sample efficient reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  164. [173]

    Omega , volume=

    Effect of data standardization on neural network training , author=. Omega , volume=. 1996 , publisher=

    1996

  165. [174]

    ArXiv e-prints , pages=

    Layer normalization , author=. ArXiv e-prints , pages=

  166. [175]

    International Conference on Machine Learning , pages=

    On layer normalization in the transformer architecture , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  167. [176]

    Forty-first International Conference on Machine Learning , year=

    Craftax: A Lightning-Fast Benchmark for Open-Ended Reinforcement Learning , author=. Forty-first International Conference on Machine Learning , year=

  168. [177]

    International Conference on Learning Representations , year=

    Benchmarking the Spectrum of Agent Capabilities , author=. International Conference on Learning Representations , year=

  169. [178]

    Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , year =

    Heinrich K. Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , year =

  170. [179]

    International conference on machine learning , pages=

    Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor , author=. International conference on machine learning , pages=. 2018 , organization=

    2018

  171. [180]

    arXiv preprint arXiv:2301.13442 , year=

    Scaling laws for single-agent reinforcement learning , author=. arXiv preprint arXiv:2301.13442 , year=

    arXiv

  172. [181]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Distributional reinforcement learning with quantile regression , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  173. [182]

    arXiv preprint arXiv:1810.12894 , year=

    Exploration by random network distillation , author=. arXiv preprint arXiv:1810.12894 , year=

    arXiv

  174. [183]

    arXiv preprint arXiv:1707.06347 , year=

    Proximal policy optimization algorithms , author=. arXiv preprint arXiv:1707.06347 , year=

    arXiv

  175. [184]

    International Conference on Machine Learning , pages=

    Rethinking bias-variance trade-off for generalization of neural networks , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  176. [185]

    GitHub repository , howpublished =

    Dhariwal, Prafulla and Hesse, Christopher and Klimov, Oleg and Nichol, Alex and Plappert, Matthias and Radford, Alec and Schulman, John and Sidor, Szymon and Wu, Yuhuai and Zhokhov, Peter , title =. GitHub repository , howpublished =. 2017 , publisher =

    2017

  177. [186]

    2020 , journal=

    Acme: A Research Framework for Distributed Reinforcement Learning , author=. 2020 , journal=

    2020

  178. [187]

    Journal of Machine Learning Research , year =

    Antonin Raffin and Ashley Hill and Adam Gleave and Anssi Kanervisto and Maximilian Ernestus and Noah Dormann , title =. Journal of Machine Learning Research , year =

  179. [188]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Simplicity Bias in Overparameterized Machine Learning , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  180. [189]

    International conference on machine learning , pages=

    A closer look at memorization in deep networks , author=. International conference on machine learning , pages=. 2017 , organization=

    2017

  181. [190]

    arXiv preprint arXiv:1805.08522 , year=

    Deep learning generalizes because the parameter-function map is biased towards simple functions , author=. arXiv preprint arXiv:1805.08522 , year=

    arXiv

  182. [191]

    arXiv preprint arXiv:1909.11522 , year=

    Neural networks are a priori biased towards boolean functions with low entropy , author=. arXiv preprint arXiv:1909.11522 , year=

    1909 arXiv

  183. [192]

    arXiv preprint arXiv:2103.10427 , year=

    The low-rank simplicity bias in deep networks , author=. arXiv preprint arXiv:2103.10427 , year=

    arXiv

  184. [193]

    Advances in Neural Information Processing Systems , volume=

    The pitfalls of simplicity bias in neural networks , author=. Advances in Neural Information Processing Systems , volume=

  185. [194]

    Journal of Machine Learning Research , volume=

    The implicit bias of gradient descent on separable data , author=. Journal of Machine Learning Research , volume=

  186. [195]

    The 22nd International Conference on Artificial Intelligence and Statistics , pages=

    Convergence of gradient descent on separable data , author=. The 22nd International Conference on Artificial Intelligence and Statistics , pages=. 2019 , organization=

    2019

  187. [196]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Deep reinforcement learning with double q-learning , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  188. [197]

    International conference on machine learning , pages=

    Addressing function approximation error in actor-critic methods , author=. International conference on machine learning , pages=. 2018 , organization=

    2018

  189. [198]

    arXiv preprint arXiv:1706.01905 , year=

    Parameter space noise for exploration , author=. arXiv preprint arXiv:1706.01905 , year=

    arXiv

  190. [199]

    The Eleventh International Conference on Learning Representations , year=

    Pink noise is all you need: Colored noise exploration in deep reinforcement learning , author=. The Eleventh International Conference on Learning Representations , year=

  191. [200]

    arXiv preprint arXiv:2110.02034 , year=

    Dropout q-functions for doubly efficient reinforcement learning , author=. arXiv preprint arXiv:2110.02034 , year=

    arXiv

  192. [201]

    arXiv preprint arXiv:2101.05982 , year=

    Randomized ensembled double q-learning: Learning fast without a model , author=. arXiv preprint arXiv:2101.05982 , year=

    arXiv

  193. [202]

    Advances in Neural Information Processing Systems , year=

    Attention is all you need , author=. Advances in Neural Information Processing Systems , year=

  194. [203]

    arXiv preprint arXiv:2406.02596 , year=

    Slow and Steady Wins the Race: Maintaining Plasticity with Hare and Tortoise Networks , author=. arXiv preprint arXiv:2406.02596 , year=

    arXiv

  195. [204]

    Journal of Machine Learning Research , volume=

    Rademacher and Gaussian complexities: Risk bounds and structural results , author=. Journal of Machine Learning Research , volume=

  196. [205]

    Advances in Neural Information Processing Systems , volume=

    Random deep neural networks are biased towards simple functions , author=. Advances in Neural Information Processing Systems , volume=

  197. [206]

    Advances in Neural Information Processing Systems , volume=

    The alignment property of SGD noise and how it helps select flat minima: A stability analysis , author=. Advances in Neural Information Processing Systems , volume=

  198. [207]

    arXiv preprint arXiv:2010.01412 , year=

    Sharpness-aware minimization for efficiently improving generalization , author=. arXiv preprint arXiv:2010.01412 , year=

    2010 arXiv

  199. [208]

    Incomplete Ideas (blog) , volume=

    The bitter lesson , author=. Incomplete Ideas (blog) , volume=

  200. [209]

    Advances in Neural Information Processing Systems , volume=

    Learning to discover skills through guidance , author=. Advances in Neural Information Processing Systems , volume=

  201. [210]

    arXiv preprint arXiv:2406.00324 , year=

    Do's and Don'ts: Learning Desirable Skills with Instruction Videos , author=. arXiv preprint arXiv:2406.00324 , year=

    arXiv

  202. [211]

    Proceedings of the IRE , volume=

    Communication in the presence of noise , author=. Proceedings of the IRE , volume=. 1949 , publisher=

    1949

  203. [212]

    Root mean square layer normalization , author=

  204. [213]

    2012 IEEE/RSJ international conference on intelligent robots and systems , pages=

    Mujoco: A physics engine for model-based control , author=. 2012 IEEE/RSJ international conference on intelligent robots and systems , pages=. 2012 , organization=

    2012

  205. [214]

    arXiv preprint arXiv:1606.01540 , year=

    OpenAI Gym , author=. arXiv preprint arXiv:1606.01540 , year=

    arXiv

  206. [215]

    arXiv preprint arXiv:1801.00690 , year=

    Deepmind control suite , author=. arXiv preprint arXiv:1801.00690 , year=

    arXiv

  207. [216]

    arXiv preprint arXiv:2205.13600 , year=

    MyoSuite--A contact-rich simulation suite for musculoskeletal motor control , author=. arXiv preprint arXiv:2205.13600 , year=

    arXiv

  208. [217]

    arXiv preprint arXiv:2403.10506 , year=

    Humanoidbench: Simulated humanoid benchmark for whole-body locomotion and manipulation , author=. arXiv preprint arXiv:2403.10506 , year=

    arXiv

  209. [218]

    arXiv preprint arXiv:2407.17032 , year=

    Gymnasium: A standard interface for reinforcement learning environments , author=. arXiv preprint arXiv:2407.17032 , year=

    arXiv

  210. [219]

    arXiv preprint arXiv:2410.08896 , year=

    MAD-TD: Model-Augmented Data stabilizes High Update Ratio RL , author=. arXiv preprint arXiv:2410.08896 , year=

    arXiv

  211. [220]

    arXiv preprint arXiv:2406.02696 , year=

    iQRL--Implicitly Quantized Representations for Sample-efficient Reinforcement Learning , author=. arXiv preprint arXiv:2406.02696 , year=

    arXiv

  212. [221]

    International conference on machine learning , pages=

    Can increasing input dimensionality improve deep reinforcement learning? , author=. International conference on machine learning , pages=. 2020 , organization=

    2020

  213. [222]

    International Conference on Machine Learning , pages=

    Controlling overestimation bias with truncated mixture of continuous distributional quantile critics , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  214. [223]

    arXiv preprint arXiv:2411.12155 , year=

    Reinforcement Learning with Action Sequence for Data-Efficient Robot Learning , author=. arXiv preprint arXiv:2411.12155 , year=

  215. [224]

    COMPSTAT 1982 5th Symposium held at Toulouse 1982: Part I: Proceedings in Computational Statistics , pages=

    Updating formulae and a pairwise algorithm for computing sample variances , author=. COMPSTAT 1982 5th Symposium held at Toulouse 1982: Part I: Proceedings in Computational Statistics , pages=. 1982 , organization=

    1982

  216. [225]

    International conference on machine learning , pages=

    A distributional perspective on reinforcement learning , author=. International conference on machine learning , pages=. 2017 , organization=

    2017

  217. [226]

    Uncertainty in Artificial Intelligence , pages=

    Rezero is all you need: Fast convergence at large depth , author=. Uncertainty in Artificial Intelligence , pages=. 2021 , organization=

    2021

  218. [227]

    Deep Reinforcement Learning Workshop NeurIPS 2022 , year=

    Sample-efficient reinforcement learning by breaking the replay ratio barrier , author=. Deep Reinforcement Learning Workshop NeurIPS 2022 , year=

    2022

  219. [228]

    (No Title) , year=

    A comprehensive introduction to differential geometry , author=. (No Title) , year=

  220. [229]

    1992 , publisher=

    Riemannian geometry , author=. 1992 , publisher=

    1992

  221. [230]

    2008 , publisher=

    Optimization algorithms on matrix manifolds , author=. 2008 , publisher=

    2008

  222. [231]

    2006 , publisher=

    Riemannian manifolds: an introduction to curvature , author=. 2006 , publisher=

    2006

  223. [232]

    Stochastic Gradient Descent on Riemannian Manifolds , volume=

    Bonnabel, Silvere , year=. Stochastic Gradient Descent on Riemannian Manifolds , volume=. IEEE Transactions on Automatic Control , publisher=. doi:10.1109/tac.2013.2254619 , number=

    2013

  224. [233]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    High-resolution image synthesis with latent diffusion models , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  225. [234]

    International conference on machine learning , pages=

    Zero-shot text-to-image generation , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  226. [235]

    arXiv preprint arXiv:1912.02975 , year=

    Observational overfitting in reinforcement learning , author=. arXiv preprint arXiv:1912.02975 , year=

    1912 arXiv

  227. [236]

    arXiv preprint arXiv:1810.00123 , year=

    Generalization and regularization in dqn , author=. arXiv preprint arXiv:1810.00123 , year=

    arXiv

  228. [237]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Deep reinforcement learning that matters , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  229. [238]

    International Conference on Machine Learning , pages=

    Hyperparameters in reinforcement learning and how to tune them , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  230. [239]

    arXiv preprint arXiv:2407.01800 , year=

    Normalization and effective learning rates in reinforcement learning , author=. arXiv preprint arXiv:2407.01800 , year=

    arXiv

  231. [240]

    arXiv preprint arXiv:2410.01131 , year=

    ngpt: Normalized transformer with representation learning on the hypersphere , author=. arXiv preprint arXiv:2410.01131 , year=

    arXiv

  232. [241]

    Advances in neural information processing systems , volume=

    A minimalist approach to offline reinforcement learning , author=. Advances in neural information processing systems , volume=

  233. [242]

    arXiv preprint arXiv:2110.06169 , year=

    Offline reinforcement learning with implicit q-learning , author=. arXiv preprint arXiv:2110.06169 , year=

    arXiv

  234. [243]

    arXiv preprint arXiv:2208.06193 , year=

    Diffusion policies as an expressive policy class for offline reinforcement learning , author=. arXiv preprint arXiv:2208.06193 , year=

    arXiv

  235. [244]

    arXiv preprint arXiv:2301.02328 , year=

    Extreme q-learning: Maxent rl without entropy , author=. arXiv preprint arXiv:2301.02328 , year=

    arXiv

  236. [245]

    arXiv preprint arXiv:2304.10573 , year=

    Idql: Implicit q-learning as an actor-critic method with diffusion policies , author=. arXiv preprint arXiv:2304.10573 , year=

    arXiv

  237. [246]

    arXiv preprint arXiv:2306.05167 , year=

    Decision s4: Efficient sequence-based rl via state spaces layers , author=. arXiv preprint arXiv:2306.05167 , year=

    arXiv

  238. [247]

    arXiv preprint arXiv:2310.03022 , year=

    Decision convformer: Local filtering in metaformer is sufficient for decision making , author=. arXiv preprint arXiv:2310.03022 , year=

    arXiv

  239. [248]

    arXiv preprint arXiv:2004.07219 , year=

    D4rl: Datasets for deep data-driven reinforcement learning , author=. arXiv preprint arXiv:2004.07219 , year=

    2004 arXiv

  240. [249]

    arXiv preprint arXiv:2410.14606 , year=

    Streaming Deep Reinforcement Learning Finally Works , author=. arXiv preprint arXiv:2410.14606 , year=

    arXiv

  241. [250]

    arXiv preprint arXiv:2402.08609 , year=

    Mixtures of experts unlock parameter scaling for deep rl , author=. arXiv preprint arXiv:2402.08609 , year=

    arXiv

  242. [251]

    International conference on machine learning , pages=

    Understanding contrastive representation learning through alignment and uniformity on the hypersphere , author=. International conference on machine learning , pages=. 2020 , organization=

    2020

  243. [252]

    arXiv preprint arXiv:1808.10805 , year=

    Spherical latent spaces for stable variational autoencoders , author=. arXiv preprint arXiv:1808.10805 , year=

    arXiv

  244. [253]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Sphereface: Deep hypersphere embedding for face recognition , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  245. [254]

    Proceedings of the 25th ACM international conference on Multimedia , pages=

    Normface: L2 hypersphere embedding for face verification , author=. Proceedings of the 25th ACM international conference on Multimedia , pages=

  246. [255]

    Advances in neural information processing systems , volume=

    Deep hyperspherical learning , author=. Advances in neural information processing systems , volume=

  247. [256]

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

    Analyzing and improving the training dynamics of diffusion models , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  248. [257]

    International conference on machine learning , pages=

    A simple framework for contrastive learning of visual representations , author=. International conference on machine learning , pages=. 2020 , organization=

    2020

  249. [258]

    International conference on machine learning , pages=

    Improving regression performance with distributional losses , author=. International conference on machine learning , pages=. 2018 , organization=

    2018

  250. [259]

    arXiv preprint arXiv:2405.09999 , year=

    Reward Centering , author=. arXiv preprint arXiv:2405.09999 , year=

    arXiv

  251. [260]

    arXiv preprint arXiv:2403.03950 , year=

    Stop regressing: Training value functions via classification for scalable deep rl , author=. arXiv preprint arXiv:2403.03950 , year=

    arXiv

  252. [261]

    arXiv preprint arXiv:2211.15144 , year=

    Offline q-learning on diverse multi-task data both scales and generalizes , author=. arXiv preprint arXiv:2211.15144 , year=

    arXiv

  253. [262]

    Advances in Neural Information Processing Systems , volume=

    Training scale-invariant neural networks on the sphere can happen in three regimes , author=. Advances in Neural Information Processing Systems , volume=

  254. [263]

    arXiv preprint arXiv:2105.05347 , year=

    Return-based scaling: Yet another normalisation trick for deep rl , author=. arXiv preprint arXiv:2105.05347 , year=

    arXiv

  255. [264]

    Reinforcement Learning Conference , year=

    Dissecting deep rl with high update ratios: Combatting value divergence , author=. Reinforcement Learning Conference , year=

  256. [265]

    Thirty-seventh Conference on Neural Information Processing Systems , year=

    Sample-Efficient and Safe Deep Reinforcement Learning via Reset Deep Ensemble Agents , author=. Thirty-seventh Conference on Neural Information Processing Systems , year=

  257. [266]

    arXiv preprint arXiv:2411.04832 , year=

    Plasticity Loss in Deep Reinforcement Learning: A Survey , author=. arXiv preprint arXiv:2411.04832 , year=

    arXiv

  258. [267]

    Advances in neural information processing systems , volume=

    Weight normalization: A simple reparameterization to accelerate training of deep neural networks , author=. Advances in neural information processing systems , volume=

  259. [268]

    Science Robotics , volume=

    Learning agile and dynamic motor skills for legged robots , author=. Science Robotics , volume=. 2019 , publisher=

    2019

  260. [269]

    arXiv preprint arXiv:2501.12948 , year=

    DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning , author=. arXiv preprint arXiv:2501.12948 , year=

    arXiv

  261. [270]

    Advances in neural information processing systems , volume=

    Training language models to follow instructions with human feedback , author=. Advances in neural information processing systems , volume=

  262. [271]

    arXiv preprint arXiv:1709.10087 , year=

    Learning complex dexterous manipulation with deep reinforcement learning and demonstrations , author=. arXiv preprint arXiv:1709.10087 , year=

    arXiv

  263. [272]

    Conference on robot learning , pages=

    Meta-world: A benchmark and evaluation for multi-task and meta reinforcement learning , author=. Conference on robot learning , pages=. 2020 , organization=

    2020

  264. [273]

    Advances in neural information processing systems , volume=

    Mastering atari games with limited data , author=. Advances in neural information processing systems , volume=

  265. [274]

    International Conference on Machine Learning , pages=

    Efficient online reinforcement learning with offline data , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  266. [275]

    arXiv preprint arXiv:2502.07523 , year=

    Scaling Off-Policy Reinforcement Learning with Batch and Weight Normalization , author=. arXiv preprint arXiv:2502.07523 , year=

    arXiv

  267. [276]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    A convnet for the 2020s , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  268. [277]

    arXiv preprint arXiv:2310.07418 , year=

    Revisiting plasticity in visual reinforcement learning: Data, modules and training stages , author=. arXiv preprint arXiv:2310.07418 , year=

    arXiv

  269. [278]

    arXiv preprint arXiv:2403.05996 , year=

    Dissecting deep rl with high update ratios: Combatting value divergence , author=. arXiv preprint arXiv:2403.05996 , year=

    arXiv

  270. [279]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Pointwise convolutional neural networks , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  271. [280]

    Don't flatten, tokenize! Unlocking the key to SoftMoE's efficacy in deep

    Ghada Sokar and Johan Samir Obando Ceron and Aaron Courville and Hugo Larochelle and Pablo Samuel Castro , booktitle=. Don't flatten, tokenize! Unlocking the key to SoftMoE's efficacy in deep. 2025 , url=

    2025

  272. [281]

    arXiv preprint arXiv:2503.05546 , year=

    Impoola: The Power of Average Pooling for Image-Based Deep Reinforcement Learning , author=. arXiv preprint arXiv:2503.05546 , year=

    arXiv

  273. [282]

    arXiv preprint arXiv:2403.03949 , year=

    Reconciling reality through simulation: A real-to-sim-to-real approach for robust manipulation , author=. arXiv preprint arXiv:2403.03949 , year=

    arXiv

  274. [283]

    arXiv preprint arXiv:2501.03575 , year=

    Cosmos world foundation model platform for physical ai , author=. arXiv preprint arXiv:2501.03575 , year=

    arXiv

  275. [284]

    arXiv preprint arXiv:2502.08844 , year=

    MuJoCo Playground , author=. arXiv preprint arXiv:2502.08844 , year=

    arXiv

  276. [285]

    arXiv preprint arXiv:2108.10470 , year=

    Isaac gym: High performance gpu-based physics simulation for robot learning , author=. arXiv preprint arXiv:2108.10470 , year=

    arXiv

  277. [286]

    arXiv preprint arXiv:2107.03380 , year=

    Rrl: Resnet as representation for reinforcement learning , author=. arXiv preprint arXiv:2107.03380 , year=

    arXiv

  278. [287]

    Matrix Computations

    Lanczos Methods , author=. Matrix Computations. Baltimore: Johns Hopkins University Press , pages=

  279. [288]

    Advances in neural information processing systems , volume=

    Imagenet classification with deep convolutional neural networks , author=. Advances in neural information processing systems , volume=

  280. [289]

    arXiv preprint arXiv:2505.17749 , year=

    Mind the GAP! The Challenges of Scale in Pixel-based Deep Reinforcement Learning , author=. arXiv preprint arXiv:2505.17749 , year=

  281. [290]

    arXiv preprint arXiv:2505.15345 , year=

    Hadamax Encoding: Elevating Performance in Model-Free Atari , author=. arXiv preprint arXiv:2505.15345 , year=

    arXiv

  282. [291]

    arXiv preprint arXiv:2505.23150 , year=

    Bigger, Regularized, Categorical: High-Capacity Value Functions are Efficient Multi-Task Learners , author=. arXiv preprint arXiv:2505.23150 , year=

    arXiv

  283. [292]

    arXiv preprint arXiv:2602.11800 , year=

    Temporal Difference Learning with Constrained Initial Representations , author=. arXiv preprint arXiv:2602.11800 , year=

  284. [293]

    arXiv preprint arXiv:2507.02712 , year=

    A Forget-and-Grow Strategy for Deep Reinforcement Learning Scaling in Continuous Control , author=. arXiv preprint arXiv:2507.02712 , year=

    arXiv

  285. [294]

    arXiv preprint arXiv:2604.04539 , year=

    FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control , author=. arXiv preprint arXiv:2604.04539 , year=

    arXiv

  286. [295]

    arXiv preprint arXiv:2602.08040 , year=

    FIRE: Frobenius-Isometry Reinitialization for Balancing the Stability-Plasticity Tradeoff , author=. arXiv preprint arXiv:2602.08040 , year=

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.