Pith. sign in

REVIEW 3 major objections 3 minor 3 cited by

Guiding Data Collection via Factored Scaling Curves

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

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

Pith's one-line read Factored scaling curves predict which demonstration data will most improve a robot policy.

desk verdict A practical data-collection method that mostly holds up in simulation; the real-world evidence is thinner than the headline margins suggest, but the idea and offline proxy are worth engaging. read the letter →

arxiv 2505.07728 v1 pith:BALNGYS6 submitted 2025-05-12 cs.RO cs.AIcs.LG

Lihan Zha , Apurva Badithela , Michael Zhang , Justin Lidard , Jeremy Bao , Emily Zhou , David Snyder , Allen Z. Ren
show 2 more authors
Dhruv Shah Anirudha Majumdar
This is my paper · ORCID
classification cs.ROcs.AIcs.LG
keywords factoredscalingcurvesdatacollectionimitationlearningrobotmanipulationlawspower-lawextrapolationpolicyembeddingsimilaritybudgetallocation
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

Factored scaling curves are the paper's answer to a practical problem: when a robot policy performs poorly in new environments, which additional demonstrations should be collected, and how many, under a fixed budget? The paper models each environmental factor, such as lighting, table texture, distractors, camera pose, or object pose, with its own curve mapping the amount of factor-specific demonstration data to policy success rate on the target distribution. A fitted power law is extrapolated to the budget horizon, and the steepest predicted slope identifies the factor or factor pair whose data will improve performance most. Across simulation and real-robot tasks, training from scratch or fine-tuning a pretrained policy, this allocation beats equal, greedy, and data-mixture-weighting baselines, with real-world success-rate advantages up to 26 percent. The method also works with an offline embedding-similarity proxy in place of expensive hardware evaluations.

What carries the argument

The load-bearing object is the factored scaling curve: a function $\Phi_i(n)$ that maps the number $n$ of demonstrations varying factor $f_i$ to the policy's expected success on the target environment distribution, holding other factor data fixed. The paper fits each curve as a power law $\hat\Phi_i(n)=1-a(n+|D\setminus D_i|)^b$, then uses the discrete slope $P_i^K$ over the budget horizon $K$ as the ranking signal. Grouping factors into disjoint pairs reduces the number of curves from $O(N^2)$ to $O(N)$ while retaining most of the predictive accuracy, and a version of the curve built on cosine similarity of policy embeddings gives an offline alternative to hardware evaluation. What makes the machinery work is the extrapolation: from as few as four fitted points, the slope at the budget determines which factor will yield the largest improvement.

What would settle it

Take a task and initial dataset, fit curves from four sizes such as $n=0,20,40,60$, predict the top factor at budgets $K=100$ and $K=500$, actually collect that much data for the top-ranked factor and for a lower-ranked factor, and evaluate on held-out factor combinations. The central claim fails if the observed gain from the top-ranked factor is no larger, within statistical error, than the gain from a lower-ranked factor or from an equal split, or if the measured success falls outside the fitted curve's prediction interval.

Watch

Extended reading notes

Core claim

The paper's central claim is that factored scaling curves can accurately predict how a policy's success rate changes when more demonstrations are collected for a particular environmental factor, and that this prediction is what should guide the next round of data collection. Starting from a dataset with demonstrations covering several factors, the method trains policies at a few intermediate dataset sizes for each factor or factor pair, evaluates them on the target environment distribution, and fits $\hat\Phi_i(n)=1-a(n+|D\setminus D_i|)^b$ with $a>0$ and $b<0$ to those points. The extrapolated slope $P_i^K=(\hat\Phi_i(|D_i|+K)-\hat\Phi_i(|D_i|))/K$ ranks factors by expected gain; allocating the whole budget to the top-ranked factor, or across factors when gains are similar, is the recommended strategy. Across simulation and real-robot experiments the paper reports that this allocation beats equal, greedy, and distributionally weighted baselines, with extrapolated curves matching actual performance in the tested cases and real-world success-rate advantages up to 26 percent.

Load-bearing premise

The whole method assumes that a power-law curve fitted through only a handful of noisy measurements keeps describing how success rate grows as more demonstrations are added, even well beyond the measured range.

Editorial extensions

If this is right

  • A practitioner with a fixed demonstration budget can compute a small set of curve points and allocate data to the factor pair with the steepest extrapolated slope, rather than spreading data uniformly.
  • The fitted curves give a quantitative expected success rate after collecting $K$ demonstrations, so a budget can be chosen by inverting the curve to hit a target success rate.
  • When the fitted curves show one factor clearly dominant, the Top strategy of allocating everything to it works best; when gains are similar, proportional All allocation is preferable.
  • FSC-Proxy can guide data collection using an offline embedding-similarity metric without large-scale real-world rollouts for every curve point.
  • The method transfers between training-from-scratch and fine-tuning settings and across visual and spatial factor types, so the same guide applies to different policy families.

Reading between the lines

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

  • Extension the paper leaves implicit: repeated re-fitting after each collection batch could turn FSC into an adaptive budget scheduler and likely reduce the large-budget degradation the paper reports at $K=500$.
  • Because the embedding-similarity proxy uses only initial observations, it could screen candidate data-generation setups, such as simulators or synthetic pipelines, before any real-robot runs, which the paper does not test.
  • The slope ranking could be read as a marginal-value signal for deciding whether to add a brand-new factor to the study, not just how much data to add to existing ones.
  • A natural stress test: transfer a fitted curve from one task to a related task and check whether the factor ranking remains stable across policy backbones.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The paper proposes Factored Scaling Curves (FSC) as a tool for deciding which environmental factors to prioritize when collecting additional demonstrations for imitation-learned robot policies. For each factor or factor pair, the method trains policies on increasing amounts of factor-specific data, evaluates them on a target distribution, fits a power law (Eq. 5), and computes an expected improvement slope (Eqs. 7 and 12). The budget is then allocated by Top, Top-Half, or All strategies. The paper evaluates FSC in ManiSkill3 simulation and on a Franka robot, using diffusion policies and fine-tuned pi0 models, and also presents FSC-Proxy, an offline variant where curves are built from policy embedding similarity instead of hardware success rates. The main claims are that factored scaling curves accurately predict the effect of additional factor data, and that the resulting data-collection strategy outperforms Equal, Greedy, and Re-Mix by roughly 10% on average in simulation and by up to 26% in real-world tasks.

Significance. If validated, FSC would be a practically useful and policy-aware method for allocating limited data-collection effort, a question of direct relevance to robot learning. The simulation evaluation is a genuine strength: each curve point is averaged over 3000-4800 rollouts (Appendix B.3), and Table 1 shows consistent gains over strong baselines across five tasks and two budget sizes. The paper is also admirably transparent about limitations, including the degradation at large budgets in Table 7. The FSC-Proxy idea, which replaces hardware evaluation with embedding similarity, is interesting and potentially high-impact. The main reservation is that the central predictive claim rests on real-world curves estimated from only 12-20 trials per policy, and the paper's own ablations show regimes in which FSC does not beat Equal; the real-world headline numbers therefore need either stronger uncertainty analysis or a more carefully scoped claim.

major comments (3)
  1. [Section 4.2 and Appendix C.4] The claim that factored scaling curves 'can accurately predict how policy performance changes with additional factor data' is not supported by uncertainty quantification in the real-world experiments. Appendix C.4 states that each policy is evaluated on 12-20 trials (Pick Place: 20; Fold Towel: 12; Mouse in Drawer: 18), so the standard error of a single success-rate estimate is roughly 0.11-0.14 at a success rate near 0.5. This is the same order of magnitude as the predicted improvements and as the FSC-vs-baseline gaps reported in Figure 3. Because the Top strategy commits the entire budget to the single highest-slope factor pair, trial-level noise can change the fitted exponent b in Eq. (5) and flip the recommended factor. Please provide bootstrap confidence intervals around the fitted curves and the resulting factor rankings, or explicitly restrict the predictive claim to settings in which ranking stability can be demonstrated.
  2. [Appendix A.4, Tables 6 and 7] The paper's own ablations show that FSC's advantage is regime-dependent, which contradicts the unqualified predictive claim in Section 4.2. In Peg Insertion - Spatial with an initial set of 480 demos, prediction-driven allocation falls below Equal at K=500 (All 59.4 vs Equal 63.0; Top-Half 48.4), and in Table 6, with a 300-demo initial set on Peg Insertion - Visual, Top achieves 58.4 at K=20 while Equal achieves 64.2. The Limitations section acknowledges this, but the abstract and Section 4.2 state the predictive accuracy without this scope. Please add a quantitative characterization of when extrapolation fails, for example in terms of budget horizon relative to the fitted range or performance saturation, and adjust the central claims accordingly.
  3. [Section 3.2, Eq. (5) and Eq. (12)] The selection statistic P^K_ij is derived from a power law fit to as few as four observed points, but the paper reports no goodness-of-fit, no model comparison, and no sensitivity analysis for the fitted exponents. Figure 4 itself notes an outlier at n=60 in Fold Towel - Visual that 'slightly distorts the fit,' which illustrates how sensitive the fitted exponent, and hence the factor ranking, can be to individual curve points. Since the entire allocation decision is a function of the ranking of slopes, please report fit residuals and prediction intervals, and show that the ranking is stable under reasonable perturbations of the curve points or under alternative parametric forms.
minor comments (3)
  1. [Section 4.4, Table 4] The statement that FSC-Proxy 'matches or surpasses' FSC is not supported by Table 4: FSC-Proxy is lower than FSC on Peg Insertion at K=100 (53.3 vs 57.9) and on Pull Cube Tool at K=100 (73.4 vs 83.4). Please soften this claim or add significance testing.
  2. [Section 4.3, Table 2] The claim that Group 'retains the full performance advantage' is overstated for Pick Place at K=100, where Group (64.4) is essentially tied with Equal (64.3) while Pairwise (78.8) is much higher. This does not invalidate the cost argument, but the wording should reflect the task-dependence of the comparison.
  3. [Appendix C.4 and Figure 3] Figure 3 reports real-world success rates without error bars or trial counts, even though Appendix C.4 gives only 12-20 trials per policy. Adding per-policy confidence intervals or raw trial-level data would greatly improve the reader's ability to judge the magnitude and reliability of the reported gains.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: FSC extrapolations are checked against newly trained policies, and the only self-citation is a non-load-bearing offline proxy metric.

full rationale

FSC curves are fit from policies trained on subsets of the available data and evaluated on a held-out target distribution (Eqs. 4-5); the predicted gain P^K_i (Eq. 7) is an extrapolation of that fit, and the paper's headline claim is tested by training new policies with the recommended additional data and measuring their success (Tables 1, 3, 4, Fig. 3-4). The test points are not the fitting points, so the prediction is not equal by construction to the input. The only self-reference is the offline embedding-similarity proxy imported from the authors' prior red-teaming paper [41]; that citation supplies a heuristic metric, not a theorem, and its ranking behavior is validated empirically in this paper (Figs. 5-6, Tables 4-5) rather than assumed. The paper's own limitations section acknowledges extrapolation error at large K and evaluation noise (Table 7, Appendix C.4), which are accuracy risks, not circularity. No step of the derivation reduces to its own inputs, so no significant circularity is found.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The method adds no new physical entities; the free parameters are the fitted power-law coefficients and tuning choices. The axioms above are the empirical premises that the extrapolation, the one-factor-at-a-time rule, the small real-world evaluation sizes, and the proxy metric rely on.

free parameters (4)
  • per-curve power-law coefficient a = e.g., 0.94 to 34.62 in Appendix A.5 fits
    Fit to observed success-rate points in Eq. (5); the extrapolated slope depends on a.
  • per-curve power-law exponent b = e.g., -0.89 to -0.01 in Appendix A.5 fits
    Fit from the same points; small changes in b strongly change predictions at |D|+K.
  • number of curve construction points m = m = 4 in the main experiments
    Section 3.2 states four values of n are often sufficient; this spacing choice is made by the authors.
  • k for FSC-Proxy k-NN = k = 1 (ablated to 5 and 10)
    Appendix A.3 uses k=1 for reported proxy results; it is a hand-set hyperparameter.
assumptions (6)
  • domain assumption Policy success for a factor is a smooth, saturating function of data volume, described by Phi_i(n) = 1 - a(n + |D_excl_i|)^b.
    Required for curve fitting and extrapolation in Eq. (5); no proof of monotonicity or functional form.
  • domain assumption Each demonstration varies exactly one factor at a time, and paired groups capture all interactions that matter.
    Section 3.1 fixes the collection rule and Section 3.2 defines Group-2, so cross-group interactions are ignored.
  • ad hoc to paper Evaluation on 12 to 20 real-world trials per policy estimates S accurately enough for curve fitting.
    Appendix C.4 reports these small trial counts without confidence intervals; the fitted slopes inherit this noise.
  • domain assumption New demonstrations for a factor follow the same distribution as the existing D_i, so the curve can be extrapolated beyond the fitted range.
    The method predicts performance at |D|+K; no distribution-shift correction is provided.
  • domain assumption Embedding similarity between training and evaluation observations is monotonically related to real success.
    FSC-Proxy replaces S with c_bar_pi in Eq. (6); the paper shows correlation plots but no guarantee of monotonicity.
  • domain assumption The initial dataset has equal per-factor sizes and the budget splits evenly within a selected factor pair.
    Appendix A.2 sets |D_i| = |D|/N and splits pairwise allocations in half; this prior is not derived from data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Guiding Data Collection via Factored Scaling Curves." pith.science (2026). https://pith.science/paper/BALNGYS6

@misc{pith2026250507728,
  author       = {Pith},
  title        = {Pith review of: Guiding Data Collection via Factored Scaling Curves},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BALNGYS6}},
  note         = {Machine review of arXiv:2505.07728}
}
abstract

Generalist imitation learning policies trained on large datasets show great promise for solving diverse manipulation tasks. However, to ensure generalization to different conditions, policies need to be trained with data collected across a large set of environmental factor variations (e.g., camera pose, table height, distractors) $-$ a prohibitively expensive undertaking, if done exhaustively. We introduce a principled method for deciding what data to collect and how much to collect for each factor by constructing factored scaling curves (FSC), which quantify how policy performance varies as data scales along individual or paired factors. These curves enable targeted data acquisition for the most influential factor combinations within a given budget. We evaluate the proposed method through extensive simulated and real-world experiments, across both training-from-scratch and fine-tuning settings, and show that it boosts success rates in real-world tasks in new environments by up to 26% over existing data-collection strategies. We further demonstrate how factored scaling curves can effectively guide data collection using an offline metric, without requiring real-world evaluation at scale.

Figures

Figures reproduced from arXiv: 2505.07728 by the authors.

Figure 1
Figure 1. To efficiently collect demonstrations so as to maximize policy performance under a fixed data bud￾get, we propose factored scaling curves: a principled tool to quantify how policy performance changes with the quantity of factor data. Based on factored scaling curves, we can allocate the data budget to collecting demonstrations that vary different factors based on their importance. 1 Introduction High-quality teleope… view at source ↗
Figure 2
Figure 2. Illustration of factored scaling curves used to inform data allocation. For the distractor fac￾tor, points are used to construct the scaling curve, and is the predicted policy success rate at K ad￾ditional demos of the factor over the initial dataset. At n = |Di |, the scaling curve represents policy performance when using the full available dataset D — comprising of demonstrations from all fac￾tors. Note that const… view at source ↗
Figure 3
Figure 3. Evaluating FSC in the real world. We visualize the task rollouts and report the average policy success rate trained with additional collected data. For pick-place task, we train the policies with diffusion policy. For all other experiments, we obtain policies by fine-tuning π0. FSC achieves the best performance in all tasks, achieving up to 26% more improvement over all baseline methods. Compared to the zero-shot se… view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Visualizing factored scaling curves for real world fine-tuning [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Expected improvement for π0 on three task settings using the Attention Weights from the last denoising step: Camera Pose – Distractor (CP-D), Table Texture – Lighting (TT-L), and Robot Pose – Object Pose (RP-OP). Cosine Similarity projections are normalized to have the…
Figure 6
Figure 6. Figure 6: Expected improvement for π0 on three task settings using the Latent Action from the first denoising step: Camera Pose – Distractor (CP-D), Table Texture – Lighting (TT-L), and Robot Pose – Object Pose (RP￾OP). Cosine Similarity projections are normalized to have the sa…
Figure 7
Figure 7. Figure 7: Factored scaling curves for the simulation Pick Place task. 0 50 100 150 Number of Factor Data 0.2 0.3 0.4 Success Rate Lighting Table Texture Reference Points i, j = 1 0.94 (x + 90) 0.02 Predicted Points Test Points 0 50 100 150 Number of Factor Data 0.2 0.3 0.4 Succe…
Figure 8
Figure 8. Figure 8: Factored scaling curves for the simulation Peg Insertion - Visual task. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]
Figure 9
Figure 9. Figure 9: Factored scaling curves for the simulation Pull Cube Tool - Visual task. 0 50 100 150 Number of Factor Data 0.3 0.4 0.5 0.6 Success Rate Background Table Height Reference Points i, j = 1 0.63 (x + 180) 0.01 Predicted Points Test Points 0 50 100 150 Number of Factor Dat…
Figure 10
Figure 10. Figure 10: Factored scaling curves for the simulation Peg Insertion - Spatial task. 19 [PITH_FULL_IMAGE:figures/full_fig_p019_10.png]
Figure 11
Figure 11. Figure 11: Factored scaling curves for the simulation Pull Cube Tool - Spatial task. 0 50 100 150 Number of Factor Data 0.4 0.5 0.6 0.7 Success Rate Lighting Table Texture Reference Points i, j = 1 34.62 (x + 90) 0.89 Predicted Points 0 50 100 150 Number of Factor Data 0.4 0.5 0…
Figure 12
Figure 12. Figure 12 [PITH_FULL_IMAGE:figures/full_fig_p020_12.png]
Figure 13
Figure 13. Figure 13: Illustrations of simulation tasks. camera. The object-pose range shown in [PITH_FULL_IMAGE:figures/full_fig_p021_13.png]
Figure 14
Figure 14. Figure 14: Visualization of simulation environment visual factor variations. 22 [PITH_FULL_IMAGE:figures/full_fig_p022_14.png]
Figure 15
Figure 15. Figure 15: Visualization of real environment factor variations. each factor. For Fold Towel - Spatial, we consider an initial dataset size of 180 demos, where we have 30 demos for each factor. Mouse in Drawer: the robot needs to open a drawer, pick up a mouse, place it in the op…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

  1. EgoVerse: An Egocentric Human Dataset for Robot Learning from Around the World

    cs.RO 2026-04 unverdicted novelty 6.5 of 10

    EgoVerse releases 1,362 hours of standardized egocentric human data across 1,965 tasks and shows via multi-lab experiments that robot policy performance scales with human data volume when the data aligns with robot ob...

  2. Active Real-World Factor-Based Evaluation for Generalist Robot Policies

    cs.LG 2026-07 conditional novelty 5.0 of 10

    An active evaluation framework selects the most informative task configurations for real-robot tests, matching random testing's accuracy in 20-40% fewer trials.

  3. RaC: Robot Learning for Long-Horizon Tasks by Scaling Recovery and Correction

    cs.RO 2025-09 conditional novelty 5.0 of 10

    Robot policies trained on human interventions that rewind to a familiar state and then correct the mistake achieve higher long-horizon success and better data efficiency than imitation on full demonstrations alone.

Reference graph

Works this paper leans on

52 extracted references · 12 canonical work pages · cited by 3 Pith papers

  1. [1]

    Brohan et al

    A. Brohan et al. Rt-2: Vision-language-action models transfer web knowledge to robotic control, 2023. URL https://arxiv.org/abs/2307.15818

    arXiv 2023

  2. [2]

    Black et al

    K. Black et al. π0: A vision-language-action flow model for general robot control, 2024. URL https://arxiv.org/abs/2410.24164

    arXiv 2024

  3. [3]

    M. J. Kim, K. Pertsch, S. Karamcheti, T. Xiao, A. Balakrishna, S. Nair, R. Rafailov, E. Foster, G. Lam, P. Sanketi, Q. Vuong, T. Kollar, B. Burchfiel, R. Tedrake, D. Sadigh, S. Levine, P. Liang, and C. Finn. Openvla: An open-source vision-language-action model, 2024. URL https://arxiv.org/abs/2406.09246

    arXiv 2024

  4. [4]

    G. R. Team et al. Gemini robotics: Bringing ai into the physical world, 2025. URL https: //arxiv.org/abs/2503.20020

    arXiv 2025

  5. [5]

    T. Z. Zhao, J. Tompson, D. Driess, P. Florence, K. Ghasemipour, C. Finn, and A. Wahid. Aloha unleashed: A simple recipe for robot dexterity. arXiv preprint arXiv:2410.13126, 2024

    arXiv 2024

  6. [6]

    Khazatsky et al

    A. Khazatsky et al. Droid: A large-scale in-the-wild robot manipulation dataset. 2024

    work page 2024

  7. [7]

    Walke, K

    H. Walke, K. Black, A. Lee, M. J. Kim, M. Du, C. Zheng, T. Zhao, P. Hansen-Estruch, Q. Vuong, A. He, V . Myers, K. Fang, C. Finn, and S. Levine. Bridgedata v2: A dataset for robot learning at scale, 2024. URL https://arxiv.org/abs/2308.12952

    arXiv 2024

  8. [8]

    O. X.-E. Collaboration et al. Open x-embodiment: Robotic learning datasets and rt-x models,

    work page

Show all 52 references
  1. [9]

    C. Chi, Z. Xu, C. Pan, E. Cousineau, B. Burchfiel, S. Feng, R. Tedrake, and S. Song. Universal manipulation interface: In-the-wild robot teaching without in-the-wild robots, 2024. URL https://arxiv.org/abs/2402.10329

    2024 arXiv

  2. [10]

    M. J. Kim, C. Finn, and P. Liang. Fine-tuning vision-language-action models: Optimizing speed and success, 2025. URL https://arxiv.org/abs/2502.19645

    2025 arXiv

  3. [11]

    J. Gao, A. Xie, T. Xiao, C. Finn, and D. Sadigh. Efficient data collection for robotic manipula- tion via compositional generalization, 2024. URL https://arxiv.org/abs/2403.05110

    2024 arXiv

  4. [12]

    Nguyen, W

    D. Nguyen, W. Yang, R. Anand, Y . Yang, and B. Mirzasoleiman. Mini-batch coresets for memory-efficient language model training on data mixtures, 2025. URL https://arxiv. org/abs/2407.19580

    2025 arXiv

  5. [13]

    Guruprasad, H

    P. Guruprasad, H. Sikka, J. Song, Y . Wang, and P. P. Liang. Benchmarking vision, language, & action models on robotic learning tasks, 2024. URLhttps://arxiv.org/abs/2411.05821

    2024 arXiv

  6. [14]

    Hejna, C

    J. Hejna, C. Bhateja, Y . Jian, K. Pertsch, and D. Sadigh. Re-mix: Optimizing data mixtures for large scale imitation learning, 2024. URL https://arxiv.org/abs/2408.14037. 11

    2024 arXiv

  7. [15]

    Hejna, S

    J. Hejna, S. Mirchandani, A. Balakrishna, A. Xie, A. Wahid, J. Tompson, P. Sanketi, D. Shah, C. Devin, and D. Sadigh. Robot data curation with mutual information estimators, 2025. URL https://arxiv.org/abs/2502.08623

    2025 arXiv

  8. [16]

    S. M. Xie, H. Pham, X. Dong, N. Du, H. Liu, Y . Lu, P. S. Liang, Q. V . Le, T. Ma, and A. W. Yu. Doremi: Optimizing data mixtures speeds up language model pretraining. In Advances in Neural Information Processing Systems, 2023

    2023

  9. [17]

    Q. Liu, X. Zheng, N. Muennighoff, G. Zeng, L. Dou, T. Pang, J. Jiang, and M. Lin. Regmix: Data mixture as regression for language model pre-training, 2025. URL https://arxiv. org/abs/2407.01492

    2025 arXiv

  10. [18]

    D. V . Lindley. On a measure of the information provided by an experiment. The Annals of Mathematical Statistics , 27(4):986–1005, 1956. ISSN 00034851, 21688990. URL http: //www.jstor.org/stable/2237191

    1956

  11. [19]

    Chaloner and I

    K. Chaloner and I. Verdinelli. Bayesian Experimental Design: A Review. Statistical Science, 10(3):273 – 304, 1995. doi:10.1214/ss/1177009939. URL https://doi.org/10.1214/ss/ 1177009939

    1995

  12. [20]

    D. J. C. MacKay. Information-based objective functions for active data selection.Neural Com- putation, 4(4):590–604, Jul 1992. ISSN 0899-7667. doi:10.1162/neco.1992.4.4.590. Funding by Caltech Fellowship

    1992 doi

  13. [21]

    Houlsby, F

    N. Houlsby, F. Husz ´ar, Z. Ghahramani, and M. Lengyel. Bayesian active learning for classifi- cation and preference learning, 2011. URL https://arxiv.org/abs/1112.5745

    2011 arXiv

  14. [22]

    Sener and S

    O. Sener and S. Savarese. Active learning for convolutional neural networks: A core-set ap- proach. In International Conference on Learning Representations, 2018

    2018

  15. [23]

    Anwar, R

    A. Anwar, R. Gupta, Z. Merchant, S. Ghosh, W. Neiswanger, and J. Thomason. Efficient Evaluation of Multi-Task Robot Policies With Active Experiment Selection, Feb. 2025. URL http://arxiv.org/abs/2502.09829. arXiv:2502.09829 [cs]

    2025 arXiv

  16. [25]

    Gpt-4 technical report, 2024

    OpenAI et al. Gpt-4 technical report, 2024. URL https://arxiv.org/abs/2303.08774

    2024 arXiv

  17. [26]

    Brown et al

    T. Brown et al. Language models are few-shot learners. In Advances in Neural Informa- tion Processing Systems, 2020. URL https://proceedings.neurips.cc/paper_files/ paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf

    2020

  18. [27]

    Hoffmann, S

    J. Hoffmann, S. Borgeaud, A. Mensch, E. Buchatskaya, T. Cai, E. Rutherford, D. de Las Casas, L. A. Hendricks, J. Welbl, A. Clark, T. Hennigan, E. Noland, K. Millican, G. van den Driess- che, B. Damoc, A. Guy, S. Osindero, K. Simonyan, E. Elsen, J. W. Rae, O. Vinyals, and L. Si...

    2022 arXiv

  19. [28]

    Grattafiori et al

    A. Grattafiori et al. The llama 3 herd of models, 2024. URL https://arxiv.org/abs/ 2407.21783

    2024 arXiv

  20. [29]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, and I. Sutskever. Learning transferable visual models from natural language supervision. In Proceedings of the 38th International Conference on Machine Lea...

    2021

  21. [30]

    X. Zhai, A. Kolesnikov, N. Houlsby, and L. Beyer. Scaling vision transformers. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 12104–12113, June 2022

    2022

  22. [31]

    Peebles and S

    W. Peebles and S. Xie. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF international conference on computer vision, pages 4195–4205, 2023

    2023

  23. [32]

    Henighan, J

    T. Henighan, J. Kaplan, M. Katz, M. Chen, C. Hesse, J. Jackson, H. Jun, T. B. Brown, P. Dhari- wal, S. Gray, C. Hallacy, B. Mann, A. Radford, A. Ramesh, N. Ryder, D. M. Ziegler, J. Schul- man, D. Amodei, and S. McCandlish. Scaling laws for autoregressive generative modeling,

  24. [33]

    Sharma, L

    P. Sharma, L. Mohan, L. Pinto, and A. Gupta. Multiple interactions made easy (mime): Large scale demonstrations data for imitation. In Conference on robot learning , pages 906–915. PMLR, 2018

    2018

  25. [34]

    Kalashnikov, A

    D. Kalashnikov, A. Irpan, P. Pastor, J. Ibarz, A. Herzog, E. Jang, D. Quillen, E. Holly, M. Kalakrishnan, V . Vanhoucke, et al. Scalable deep reinforcement learning for vision-based robotic manipulation. In Conference on robot learning, pages 651–673. PMLR, 2018

    2018

  26. [35]

    S. Cabi, S. G. Colmenarejo, A. Novikov, K. Konyushkova, S. Reed, R. Jeong, K. Zolna, Y . Ay- tar, D. Budden, M. Vecerik, O. Sushkov, D. Barker, J. Scholz, M. Denil, N. de Freitas, and Z. Wang. Scaling data-driven robotics with reward sketching and batch reinforcement learn- in...

    2020 arXiv

  27. [36]

    F. Lin, Y . Hu, P. Sheng, C. Wen, J. You, and Y . Gao. Data scaling laws in imitation learning for robotic manipulation, 2024. URL https://arxiv.org/abs/2410.18647

    2024 arXiv

  28. [37]

    Belkhale, Y

    S. Belkhale, Y . Cui, and D. Sadigh. Data quality in imitation learning, 2023. URL https: //arxiv.org/abs/2306.02437

    2023 arXiv

  29. [38]

    H. Liu, S. Nasiriany, L. Zhang, Z. Bao, and Y . Zhu. Robot learning on the job: Human-in- the-loop autonomy and learning during deployment. In Robotics: Science and Systems (RSS), 2023

    2023

  30. [39]

    Y . Cui, S. Karamcheti, R. Palleti, N. Shivakumar, P. Liang, and D. Sadigh. No, to the right: Online language corrections for robotic manipulation via shared autonomy. In Proceedings of the 2023 ACM/IEEE International Conference on Human-Robot Interaction , HRI ’23, page 93–10...

    2023

  31. [40]

    Gandhi, S

    K. Gandhi, S. Karamcheti, M. Liao, and D. Sadigh. Eliciting compatible demonstrations for multi-human imitation learning, 2022. URL https://arxiv.org/abs/2210.08073

    2022 arXiv

  32. [41]

    Majumdar, M

    A. Majumdar, M. Sharma, D. Kalashnikov, S. Singh, P. Sermanet, and V . Sindhwani. Predictive red teaming: Breaking policies without breaking robots, 2025. URL https://arxiv.org/ abs/2502.06575

    2025 arXiv

  33. [42]

    Kaplan, S

    J. Kaplan, S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Rad- ford, J. Wu, and D. Amodei. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361, 2020

    2001 arXiv

  34. [43]

    Clauset, C

    A. Clauset, C. R. Shalizi, and M. E. Newman. Power-law distributions in empirical data.SIAM review, 51(4):661–703, 2009

    2009

  35. [44]

    Salton, A

    G. Salton, A. Wong, and C. S. Yang. A vector space model for automatic indexing. Commun. ACM, 18(11):613–620, Nov. 1975. ISSN 0001-0782. doi:10.1145/361219.361220. URL https://doi.org/10.1145/361219.361220. 13

    1975

  36. [45]

    Mikolov, K

    T. Mikolov, K. Chen, G. Corrado, and J. Dean. Efficient estimation of word representations in vector space, 2013. URL https://arxiv.org/abs/1301.3781

    2013 arXiv

  37. [46]

    S. Tao, F. Xiang, A. Shukla, Y . Qin, X. Hinrichsen, X. Yuan, C. Bao, X. Lin, Y . Liu, T. kai Chan, Y . Gao, X. Li, T. Mu, N. Xiao, A. Gurha, Z. Huang, R. Calandra, R. Chen, S. Luo, and H. Su. Maniskill3: Gpu parallelized robotics simulation and rendering for generalizable emb...

    arXiv

  38. [47]

    C. Chi, S. Feng, Y . Du, Z. Xu, E. Cousineau, B. Burchfiel, and S. Song. Diffusion policy: Visuomotor policy learning via action diffusion. In Proceedings of Robotics: Science and Systems (RSS), 2023

    2023

  39. [48]

    Z. Xue, S. Deng, Z. Chen, Y . Wang, Z. Yuan, and H. Xu. Demogen: Synthetic demonstration generation for data-efficient visuomotor policy learning. arXiv preprint arXiv:2502.16932 , 2025

    2025 arXiv

  40. [49]

    K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition, 2015. URL https://arxiv.org/abs/1512.03385

    2015 arXiv

  41. [50]

    M. Du, S. Nair, D. Sadigh, and C. Finn. Behavior retrieval: Few-shot imitation learning by querying unlabeled datasets. arXiv preprint arXiv:2304.08742, 2023

    2023 arXiv

  42. [51]

    Di Palo and E

    N. Di Palo and E. Johns. On the effectiveness of retrieval, alignment, and replay in manipula- tion. IEEE Robotics and Automation Letters, 9(3):2032–2039, 2024. 14 A Additional Results A.1 Algorithms We present the construction of factored scaling curves and the subsequent dat...

    2024

  43. [2020]

    URL https://arxiv.org/abs/2010.14701

    2010 arXiv

  44. [2024]

    URL https://arxiv.org/abs/2310.08864

    arXiv

Pith tools

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