Pith. sign in

REVIEW 4 major objections 5 minor 43 references

ImagineBench: Evaluating Reinforcement Learning with Large Language Model Rollouts

T0 review · 4 major / 5 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 The paper introduces ImagineBench, the first comprehensive benchmark for reinforcement learning with LLM-imaginary rollouts, and reports that existing offline RL algorithms reach only 35.44% success on hard unseen tasks versus 64.37% when…

desk verdict A genuinely useful benchmark resource for RL from LLM rollouts, but the headline algorithm-gap claim is confounded by low rollout quality and needs a quality-filtered control before it will convince anyone. read the letter →

arxiv 2505.10010 v1 pith:PBBJDEA6 submitted 2025-05-15 cs.LG

Jing-Cheng Pang , Kaiyuan Li , Yidi Wang , Si-Hang Yang , Shengyi Jiang , Yang Yu This is my paper · ORCID
classification cs.LG
keywords imaginaryrolloutsreinforcementlearninglargelanguagemodelsofflineRLbenchmarklanguage-conditionedpolicytaskgeneralization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper introduces ImagineBench, which it calls the first comprehensive benchmark for reinforcement learning with LLM-imaginary rollouts: synthetic trajectories an LLM generates after being fine-tuned on real environment data. The benchmark supplies paired real and imaginary datasets covering locomotion, robotic manipulation, and navigation, with natural-language instructions at four difficulty levels. Using six offline RL algorithms plus SAC, the paper finds that adding imaginary rollouts generally improves performance on unseen tasks but leaves a clear gap: 35.44% success on hard tasks versus 64.37% when the same methods train on real rollouts. The central claim is that RLIM is promising but lacks both a standard evaluation resource and algorithms suited to hybrid real-imaginary data, and that ImagineBench supplies the former to accelerate work on the latter.

What carries the argument

The load-bearing object is the RLIM pipeline itself. An LLM is fine-tuned on real rollout-instruction pairs through three supervised tasks, namely dynamics prediction, rollout explanation, and rollout generation, and is then prompted with a goal-oriented prompt to produce imaginary rollouts for novel goals. ImagineBench wraps this pipeline in standardized assets: real and imaginary datasets per environment, natural-language instructions, success-rate evaluation functions, and a four-level task taxonomy of training, rephrasing, easy, and hard tasks. The taxonomy is what makes the benchmark's claim testable: it separates preserved skills from genuine generalization and forces harder tasks to require composition of behaviors.

What would settle it

Replace the Llama-2-7b generator with a different fine-tuned LLM, or add a simple filter that discards imaginary rollouts failing the paper's own consistency and legality checks, and rerun the hard-task benchmarks. If hard-task success rises to the 64.37% real-rollout level under either change, the reported 35.44% figure is driven by rollout generation quality rather than by a limitation of existing offline RL algorithms.

Watch

Extended reading notes

Core claim

The central discovery is that reinforcement learning from imaginary rollouts works well enough to be worth benchmarking, but existing offline RL algorithms are not yet good at using LLM-imaginary rollouts. On unseen tasks, policies trained with imaginary rollouts outperform policies trained without them in most environments, yet they reach only 35.44% average success on hard tasks compared with 64.37% for policies trained on real rollouts collected for the same novel goals. The paper also reports that imaginary-rollout quality degrades with task difficulty: in BabyAI, goal consistency falls from 88.0% on rephrased goals to 25.8% on hard goals, while transition correctness and legality stay above roughly 65%. This quality degradation, together with the performance gap, motivates the paper's agenda: better offline RL algorithms for imaginary data, filtering of low-quality rollouts, fast online adaptation, and multi-modal imagination.

Load-bearing premise

The load-bearing premise is that LLM-imaginary rollouts produced by the paper's specific Llama-2-7b fine-tuning recipe are representative of LLM-imaginary rollouts in general; if those rollouts are unusually good or bad, the measured algorithm gap is an artifact of that pipeline rather than a property of offline RL algorithms.

Editorial extensions

If this is right

  • Adding LLM-imaginary rollouts to real data improves success on unseen tasks for most offline RL algorithms, so LLM-generated experience is a usable source of synthetic training data.
  • Existing offline RL algorithms leave a large gap on hard tasks, so algorithms designed specifically for hybrid real-imaginary data are needed rather than simply more data.
  • The benchmark's hierarchical task levels provide a common protocol for comparing future RLIM methods.
  • The quality statistics imply that imaginary rollouts are most reliable for rephrasing tasks and degrade as goals require composition, pointing to rollout filtering as a concrete algorithmic lever.
  • The public datasets lower the computational cost of developing new RLIM algorithms.

Reading between the lines

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

  • The rollout-quality statistics are reported only for BabyAI; if the same consistency and legality measurements were made in Meta-world, LIBERO, and MuJoCo, the hard-task gap might shrink or grow, so the 35.44% versus 64.37% figure should be read as tied to the Llama-2-7b fine-tuning pipeline.
  • A testable extension is to use the benchmark to compare different LLM backbones or fine-tuning recipes; if a stronger LLM raises hard-task success toward the real-rollout level, the bottleneck is imagination quality rather than offline RL algorithm design.
  • The paper leaves implicit the natural next step of turning ImagineBench into an online-adaptation benchmark, where a policy trained on imaginary rollouts is fine-tuned with a small budget of real interactions to directly measure the fast-adaptation direction it lists as future work.
  • Because the benchmark uses vectorized state spaces, its instructions may not transfer to pixel observations; extending RLIM to vision-language models would require new state encoders and evaluation functions that the current benchmark does not provide.
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

4 major / 5 minor

Summary. The paper introduces ImagineBench, a benchmark for offline reinforcement learning with LLM-imaginary rollouts. It provides real and synthetic datasets across five environments (MuJoCo, Meta-world, CLEVR-Robot, BabyAI, LIBERO), organizes tasks into Training, Rephrasing, Easy, and Hard levels, and evaluates seven offline RL algorithms (BC, CQL, BCQ, TD3+BC, PRDC, COMBO, SAC) with and without imaginary rollouts. The central empirical claim is that existing offline RL algorithms underutilize LLM-imaginary rollouts, as evidenced by a 35.44% success rate on hard tasks versus 64.37% when training on real rollouts for the same tasks. The authors also report rollout quality statistics, outline future directions, and state that code and datasets will be released.

Significance. The benchmark is a potentially valuable community resource: it targets an emerging research direction, covers diverse domains, includes language-conditioned tasks with graded difficulty, and provides a standardized comparison protocol. The release of code and datasets, if complete and reproducible, would lower the barrier for future RLIM research. However, the headline conclusion that existing algorithms are 'suboptimal' at leveraging imaginary rollouts is currently under-supported: the quality statistics are collected only for BabyAI, and the headline gap is confounded by dropout-quality differences rather than isolating algorithmic design. With additional per-environment quality measurements and quality-filtered control experiments, the benchmark's central finding could become a solid contribution; as written, the evidence is suggestive but not conclusive.

major comments (4)
  1. [Sec. 5.2, Table 2] Quality statistics (consistency, transition correctness, dynamics legality) are reported only for BabyAI, yet the abstract and Section 5.4 draw conclusions about all five environments. Please provide analogous statistics for MuJoCo, Meta-world, CLEVR-Robot, and LIBERO, or explicitly argue why these metrics are not needed for those environments; without this, the headline average gap may simply reflect where LLM rollouts are poor rather than a general algorithmic limitation.
  2. [Sec. 5.4, Fig. 6] The headline comparison (35.44% vs. 64.37% on hard tasks) compares training on noisy LLM rollouts with training on oracle real rollouts for the same novel tasks. Because the imaginary rollouts are known to be only 25.8% goal-consistent (Table 2), the observed gap does not isolate an algorithmic limitation. Please add a control that trains with a quality-filtered subset of imaginary rollouts (e.g., only goal-consistent rollouts) or with synthetic rollouts of matched quality, to distinguish data-quality effects from algorithmic insufficiency.
  3. [Sec. 5.3, Fig. 5] The claim that 'policies augmented with LLM-generated imaginary rollouts exhibit consistently higher performance on novel tasks' is not supported by the displayed bars; many pairs show w/ IR below w/o IR (e.g., BabyAI training, CQL on Meta-world rephrasing, BC on CLEVR-Robot easy). Please provide a quantitative summary across all tasks and algorithms, including error bars and significance tests, and temper or qualify the claim accordingly.
  4. [Sec. 5.4, Fig. 6] The 'Real' condition is not an algorithm baseline but an oracle that has access to real rollouts for the novel tasks; it is useful as an upper bound but should be described as such. The current presentation implies it is a competitor, which conflates the benchmark's evaluation question with a data-availability assumption.
minor comments (5)
  1. [Throughout] Please fix typographical and formatting errors, including 'encodeing' (Sec. 4.1), 'culmulative' (Sec. 3), 'CLRVR-Robot' (Table 1), and '1,3000' (Table 1).
  2. [Sec. 4.3] The evaluation metrics ('5cm positional accuracy', '85% semantic consistency') are environment-specific free parameters; please justify these thresholds or cite prior conventions so that results are reproducible.
  3. [Fig. 5] The figure contains stray '/uni000000...' tokens in the subfigure captions; these should be removed in the camera-ready version.
  4. [Sec. 5.1] The sentence 'Each training batch uniformly samples equal proportions of data from the real and LLM-imaginary datasets' should specify whether this holds across difficulty levels and how the real dataset's size is matched to the imaginary dataset's size.
  5. [Sec. 2] Related work covers KALM, URI, AgentTrek, and InCLET; consider also discussing recent benchmarks for language-conditioned RL to clarify the novelty of ImagineBench relative to those efforts.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is an empirical benchmark study whose headline comparisons are measured, not derived from the benchmark's own definitions.

full rationale

ImagineBench does not claim a formal derivation or first-principles prediction; it constructs datasets, runs standard offline RL baselines with and without LLM-imaginary rollouts, and measures success rates against environment-based evaluation functions. The 35.44% vs. 64.37% hard-task gap is an empirical comparison between training on imaginary rollouts and training on real rollouts, not a quantity that is fitted or defined into existence. The imaginary-rollout generation pipeline follows the authors' prior KALM paper [Pang et al., 2024b], but that self-citation supplies the data-generation procedure, not the benchmark's conclusions; the benchmark results could in principle have shown imaginary rollouts to be as good as or better than real rollouts. The quality statistics in Tab. 2 cover only BabyAI, and the hard-task comparison mixes dataset-source differences with algorithm limitations, but these are evidence-scope and confounding concerns about whether the stated conclusion is well supported, not circularity. No load-bearing argument reduces to its own input or to an unverified self-citation.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

No new physical or mathematical entities are introduced; the benchmark, datasets, and task-level taxonomy are resources, not postulates. The only hand-set quantities are evaluation thresholds affecting reported success rates.

free parameters (1)
  • Success thresholds per environment (5cm positional accuracy; 85% semantic consistency for HalfCheetah) = 5cm; 85%; others not specified
    Hand-chosen evaluation criteria; they define what counts as a success and therefore directly determine all reported success-rate numbers.
assumptions (3)
  • domain assumption Fine-tuned LLM rollouts are sufficiently faithful proxies for real environment experience
    The whole benchmark premise and the transfer conclusion rest on this; only BabyAI quality is measured (Sec. 5.2, Tab 2).
  • domain assumption The pre-collected offline datasets (Meta-world, CLEVR) and expert policies (BabyAI rule-based, LIBERO BC, MuJoCo SAC) provide valid real-rollout distributions
    Real data quality is taken as ground truth and used to fine-tune the LLM and as the evaluation reference; no independent verification is provided (Sec. 4.2).
  • domain assumption Success rate computed over the final five checkpoints is a stable and comparable metric
    The paper averages the final five checkpoints over three random seeds, but no confidence intervals or significance tests are reported (Sec. 5.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of ImagineBench: Evaluating Reinforcement Learning with Large Language Model Rollouts." pith.science (2026). https://pith.science/paper/PBBJDEA6

@misc{pith2026250510010,
  author       = {Pith},
  title        = {Pith review of: ImagineBench: Evaluating Reinforcement Learning with Large Language Model Rollouts},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PBBJDEA6}},
  note         = {Machine review of arXiv:2505.10010}
}
read the original abstract

A central challenge in reinforcement learning (RL) is its dependence on extensive real-world interaction data to learn task-specific policies. While recent work demonstrates that large language models (LLMs) can mitigate this limitation by generating synthetic experience (noted as imaginary rollouts) for mastering novel tasks, progress in this emerging field is hindered due to the lack of a standard benchmark. To bridge this gap, we introduce ImagineBench, the first comprehensive benchmark for evaluating offline RL algorithms that leverage both real rollouts and LLM-imaginary rollouts. The key features of ImagineBench include: (1) datasets comprising environment-collected and LLM-imaginary rollouts; (2) diverse domains of environments covering locomotion, robotic manipulation, and navigation tasks; and (3) natural language task instructions with varying complexity levels to facilitate language-conditioned policy learning. Through systematic evaluation of state-of-the-art offline RL algorithms, we observe that simply applying existing offline RL algorithms leads to suboptimal performance on unseen tasks, achieving 35.44% success rate in hard tasks in contrast to 64.37% of method training on real rollouts for hard tasks. This result highlights the need for algorithm advancements to better leverage LLM-imaginary rollouts. Additionally, we identify key opportunities for future research: including better utilization of imaginary rollouts, fast online adaptation and continual learning, and extension to multi-modal tasks. Our code is publicly available at https://github.com/LAMDA-RL/ImagineBench.

Figures

Figures reproduced from arXiv: 2505.10010 by the authors.

Figure 1
Figure 1. We benchmark the problem of RL with LLM-imaginary rollouts. The LLM is fine-tuned to generate imaginary [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of ImagineBench. Three key features of our benchmark: (1) datasets of both real and LLM-imaginary [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the generation of LLM-imaginary rollouts. The LLM is first fine-tuned with the environment [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Examples of the LLM-imaginary rollouts for novel goals. The figures are obtained by rendering the states in [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Success rate bars of different methods on various levels of goals. The x-axis denotes the offline RL algorithm, [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Comparison of training with LLM-imaginary rollouts and real environmental rollouts on novel tasks. ‘Real’ [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 37 canonical work pages

  1. [1]

    Atari, Inc. (1977). Atari 2600 home video computer system. Gaming console

    work page 1977

  2. [2]

    Brockman, G., Cheung, V., Pettersson, L., Schneider, J., Schulman, J., Tang, J., and Zaremba, W. (2016). Openai gym. arXiv , abs/1606.01540

    arXiv 2016

  3. [3]

    Chen, L., Lu, K., Rajeswaran, A., Lee, K., Grover, A., Laskin, M., Abbeel, P., Srinivas, A., and Mordatch, I. (2021). Decision transformer: Reinforcement learning via sequence modeling. In NeurIPS

    work page 2021

  4. [4]

    Chen, X.-H., Wang, Z., Du, Y., Jiang, S., Fang, M., Yu, Y., and Wang, J. (2024). Policy learning from tutorial books via understanding, rehearsing and introspecting. In NeurIPS

    work page 2024

  5. [5]

    H., and Bengio, Y

    Chevalier-Boisvert, M., Bahdanau, D., Lahlou, S., Willems, L., Saharia, C., Nguyen, T. H., and Bengio, Y. (2019). Babyai: A platform to study the sample efficiency of grounded language learning. In ICLR

    work page 2019

  6. [6]

    Devlin, J., Chang, M., Lee, K., and Toutanova, K. (2019). BERT: pre-training of deep bidirectional transformers for language understanding. In NAACL-HLT

    work page 2019

  7. [7]

    Fu, J., Kumar, A., Nachum, O., Tucker, G., and Levine, S. (2020). D4RL: datasets for deep data-driven reinforcement learning. arXiv , abs/2004.07219

    arXiv 2020

  8. [8]

    and Gu, S

    Fujimoto, S. and Gu, S. S. (2021). A minimalist approach to offline reinforcement learning. In NeurIPS

    work page 2021

Show all 43 references
  1. [9]

    Fujimoto, S., Meger, D., and Precup, D. (2019). Off-policy deep reinforcement learning without exploration. In ICML

    2019

  2. [10]

    Fujimoto, S., van Hoof, H., and Meger, D. (2018). Addressing function approximation error in actor-critic methods. In ICML

    2018

  3. [11]

    S., Mankowitz, D

    Gulcehre, C., Wang, Z., Novikov, A., Paine, T., G \'o mez, S., Zolna, K., Agarwal, R., Merel, J. S., Mankowitz, D. J., Paduraru, C., et al. (2020). Rl unplugged: A suite of benchmarks for offline reinforcement learning. In NeurIPS

    2020

  4. [12]

    Haarnoja, T., Zhou, A., Hartikainen, K., Tucker, G., Ha, S., Tan, J., Kumar, V., Zhu, H., Gupta, A., Abbeel, P., and Levine, S. (2018). Soft actor-critic algorithms and applications. arXiv , abs/1812.05905

    2018 arXiv

  5. [13]

    Do as I can, not as I say: Grounding language in robotic affordances

    Ichter, B., Brohan, A., and et al (2022). Do as I can, not as I say: Grounding language in robotic affordances. In CoRL

    2022

  6. [14]

    Kidambi, R., Rajeswaran, A., Netrapalli, P., and Joachims, T. (2020). Morel: Model-based offline reinforcement learning. In NeurIPS

    2020

  7. [15]

    Kingma, D. P. and Ba, J. (2015). Adam: A method for stochastic optimization. In ICLR

    2015

  8. [16]

    Kumar, A., Fu, J., Soh, M., Tucker, G., and Levine, S. (2019). Stabilizing off-policy q-learning via bootstrapping error reduction. In NeurIPS

    2019

  9. [17]

    Kumar, A., Zhou, A., Tucker, G., and Levine, S. (2020). Conservative q-learning for offline reinforcement learning. In NeurIPS

    2020

  10. [18]

    Levine, S., Kumar, A., Tucker, G., and Fu, J. (2020). Offline reinforcement learning: Tutorial, review, and perspectives on open problems. arXiv , abs/2005.01643

    2020 arXiv

  11. [19]

    Liu, B., Zhu, Y., Gao, C., Feng, Y., Liu, Q., Zhu, Y., and Stone, P. (2023). Libero: Benchmarking knowledge transfer for lifelong robot learning. Advances in Neural Information Processing Systems , 36:44776--44791

    2023

  12. [20]

    Liu, H., Zhu, Y., Kato, K., Tsukahara, A., Kondo, I., Aoyama, T., and Hasegawa, Y. (2024). Enhancing the llm-based robot manipulation through human-robot collaboration. IEEE Robotics and Automation Letters , 9(8):6904--6911

    2024

  13. [21]

    Luo, F., Xu, T., Lai, H., Chen, X., Zhang, W., and Yu, Y. (2024). A survey on model-based reinforcement learning. Science China Information Science , 67(2)

    2024

  14. [22]

    Lyu, J., Xu, K., Xu, J., Yan, M., Yang, J., Zhang, Z., Bai, C., Lu, Z., and Li, X. (2024). ODRL: A benchmark for off-dynamics reinforcement learning. In NeurIPS

    2024

  15. [23]

    A., Veness, J., Bellemare, M

    Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M. A., Fidjeland, A., Ostrovski, G., Petersen, S., Beattie, C., Sadik, A., Antonoglou, I., King, H., Kumaran, D., Wierstra, D., Legg, S., and Hassabis, D. (2015). Human-le...

    2015

  16. [24]

    Pang, J., Wang, P., Li, K., Chen, X., Xu, J., Zhang, Z., and Yu, Y. (2024a). Language model self-improvement by reinforcement learning contemplation. In ICLR

    2024

  17. [25]

    Pang, J., Yang, S., Chen, X., Yang, X., Yu, Y., Ma, M., Guo, Z., Yang, H., and Huang, B. (2023a). Object-oriented option framework for robotics manipulation in clutter. In IROS

    2023

  18. [26]

    Pang, J., Yang, S., Li, K., Zhang, J., Chen, X., Tang, N., and Yu, Y. (2024b). KALM: knowledgeable agents by offline reinforcement learning from large language model rollouts. In NeurIPS

    2024

  19. [27]

    Pang, J., Yang, X., Yang, S., Chen, X., and Yu, Y. (2023b). Natural language instruction-following with task-related language development and translation. In NeurIPS

    2023

  20. [28]

    Qin, R., Zhang, X., Gao, S., Chen, X., Li, Z., Zhang, W., and Yu, Y. (2022). NeoRL : A near real-world benchmark for offline reinforcement learning. In Koyejo, S., Mohamed, S., Agarwal, A., Belgrave, D., Cho, K., and Oh, A., editors, NeurIPS

    2022

  21. [29]

    Ran, Y., Li, Y.-C., Zhang, F., Zhang, Z., and Yu, Y. (2023). Policy regularization with dataset constraint for offline reinforcement learning. In ICML

    2023

  22. [30]

    Research, G. (2019). CLEVR-Robot environment. https://github.com/google-research/clevr_robot_env

    2019

  23. [31]

    and Imai, M

    Seno, T. and Imai, M. (2022). d3rlpy: An offline deep reinforcement learning library. Journal of Machine Learning Research , 23(315):1--20

    2022

  24. [32]

    Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., Dieleman, S., Grewe, D., Nham, J., Kalchbrenner, N., Sutskever, I., Lillicrap, T. P., Leach, M., Kavukcuoglu, K., Graepel, T....

    2016

  25. [33]

    Sutton, R. S. and Barto, A. G. (1998). Reinforcement learning: An introduction. IEEE Trans. Neural Networks , 9(5):1054--1054

    1998

  26. [34]

    P., and Riedmiller, M

    Tassa, Y., Doron, Y., Muldal, A., Erez, T., Li, Y., de Las Casas, D., Budden, D., Abdolmaleki, A., Merel, J., Lefrancq, A., Lillicrap, T. P., and Riedmiller, M. A. (2018). Deepmind control suite. arXiv , abs/1801.00690

    2018 arXiv

  27. [35]

    Team, P. (2021). OfflineRL : A collection of offline reinforcement learning algorithms. https://github.com/polixir/OfflineRL?tab=readme-ov-file

    2021

  28. [36]

    Todorov, E., Erez, T., and Tassa, Y. (2012). Mujoco: A physics engine for model-based control. In IROS

    2012

  29. [37]

    LLaMA 2: Open foundation and fine-tuned chat models

    Touvron, H., Martin, L., and et al (2023). LLaMA 2: Open foundation and fine-tuned chat models. arXiv , abs/2307.09288

    2023 arXiv

  30. [38]

    u l c ehre, C ., Wang, Z., Pfaff, T., Wu, Y., Ring, R., Yogatama, D., W \

    Vinyals, O., Babuschkin, I., Czarnecki, W. M., Mathieu, M., Dudzik, A., Chung, J., Choi, D. H., Powell, R., Ewalds, T., Georgiev, P., Oh, J., Horgan, D., Kroiss, M., Danihelka, I., Huang, A., Sifre, L., Cai, T., Agapiou, J. P., Jaderberg, M., Vezhnevets, A. S., Leblond, R., Po...

    2019

  31. [39]

    Wang, P.-Y., Pang, J.-C., Wang, C.-Y., Liu, X.-H., Liu, T.-S., Yang, S.-H., Qian, H., and Yu, Y. (2025). InCLET: in-context learning from language models can improve embodied instruction-following. In AAMAS

    2025

  32. [40]

    Xu, Y., Lu, D., Shen, Z., Wang, J., Wang, Z., Mao, Y., Xiong, C., and Yu, T. (2025). Agenttrek: Agent trajectory synthesis via guiding replay with web tutorials. In ICLR

    2025

  33. [41]

    Yu, T., Kumar, A., Rafailov, R., Rajeswaran, A., Levine, S., and Finn, C. (2021). COMBO: conservative offline model-based policy optimization. In NeurIPS

    2021

  34. [42]

    Yu, T., Quillen, D., He, Z., Julian, R., Hausman, K., Finn, C., and Levine, S. (2019). Meta-world: A benchmark and evaluation for multi-task and meta reinforcement learning. In CoRL

    2019

  35. [43]

    Y., Levine, S., Finn, C., and Ma, T

    Yu, T., Thomas, G., Yu, L., Ermon, S., Zou, J. Y., Levine, S., Finn, C., and Ma, T. (2020). MOPO: model-based offline policy optimization. In NeurIPS

    2020

Pith tools

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