Pith. sign in

REVIEW 3 major objections 4 minor 33 references

From Few to More: Large-scale Dynamic Multiagent Curriculum Learning

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read Training agents on small battles first, then scaling up, beats from-scratch multiagent RL

desk verdict The DyAN architecture is the real contribution; the curriculum idea is plausible, but the SOTA claim and MAgent numbers need a tighter budget and baseline story. read the letter →

arxiv 1909.02790 v2 pith:QNLPPW2F submitted 2019-09-06 cs.AI cs.MA

classification cs.AIcs.MA
keywords curriculumlearningmultiagentreinforcementlarge-scalesystemsknowledgetransfergraphneuralnetworkdynamicagent-numberStarCraftIIMAgent
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that large-scale multiagent reinforcement learning can be made tractable by curriculum training: agents first learn in a battle with few combatants, then the roster is grown until the full-scale task is reached. The authors propose three ways to carry knowledge across stages: reusing replay samples from earlier tasks, distilling earlier policies or Q-values into the current learner, and reloading the previous network as initialization for the next. Because the observation size changes when the number of agents changes, they design a Dynamic Agent-number Network (DyAN) that embeds each other-agent observation and aggregates them, so one network accepts any roster size. In StarCraft II and MAgent, curriculum-trained versions of IQL, VDN, PPO, A2C, and ACER outperform their from-scratch counterparts, with model reload and sum aggregation delivering the largest gains.

What carries the argument

The load-bearing mechanism is Dynamic Agent-number Network (DyAN), a network that splits each agent's observation into a fixed-size part for its own features and environment information and a variable-size set of observations of other agents, embeds each other-agent observation, then aggregates the embeddings with a permutation-invariant operator (sum, max, or mean). This aggregation makes the network indifferent to the number of agents in view, which is what allows curriculum stages of different battle sizes to share one architecture and enables Model Reload between stages. The paper treats this GNN aggregation as an approximate implementation of the Semantic Mapping Function from Definition 1, and reports that sum aggregation produces the clearest semantic clusters in t-SNE embeddings and the best final performance.

What would settle it

A decisive test is to extract observations from 5-vs-5, 10-vs-10, and 15-vs-15 battles that share identical local semantics, such as exactly one teammate in view, and check whether DyAN's sum-aggregated embeddings form one cluster while different-semantics observations form another; if same-semantics points do not cluster, the semantic-mapping property that Model Reload depends on fails. A controlled ablation that removes only Model Reload while keeping the same curriculum would then reveal whether the reported gains come from cross-size transfer or from some other curriculum effect.

Watch

Extended reading notes

Core claim

The central claim is that a dynamic multiagent curriculum, in which agents learn on a small scenario and progressively face more agents, greatly improves final performance on large-scale multiagent tasks, and that the improvement depends on a network whose input size can vary. The paper formalizes a semantic-mapping assumption: observations from environments with different agent counts can be projected into a common latent space in which semantically similar observations sit close together, and it builds DyAN to realize that mapping with a graph-neural-network aggregation over per-agent observations. With Model Reload, the trained network from one curriculum stage initializes the next stage directly. Empirically, on a 15-immortals-versus-15-immortals StarCraft II battle and on 50-versus-50 battles in MAgent, curriculum-trained agents reach nonzero win rates and higher survival and kill counts, whereas from-scratch baselines lag substantially, in some cases never leaving zero win rate.

Load-bearing premise

The approach rests on the idea that what an agent sees in a small battle is semantically the same as what it sees in a large battle, so a network trained on the small setting can be reused in the larger one.

Editorial extensions

If this is right

  • Curriculum-trained IQL and VDN reach nonzero win rates on the 15-vs-15 StarCraft II task in the reported training budget, while from-scratch IQL stays at zero win rate.
  • Model Reload beats both Buffer Reuse and Curriculum Distillation across the tested tasks, so reusing the trained network rather than reusing data or distilling behavior carries the largest part of the curriculum benefit.
  • Sum aggregation in DyAN outperforms max and mean on both benchmarks, which the paper links to sum producing better semantic separation of same-meaning states across different-sized scenarios.
  • The curriculum benefit is not specific to one algorithm family: IQL, VDN, PPO, A2C, and ACER all improve on MAgent when trained with DyMA-CL rather than from scratch.
  • Both StarCraft II and MAgent results support the paper's claim that increasing agent count during training is an effective route into large-scale multiagent learning problems.

Reading between the lines

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

  • If the semantic-mapping assumption holds generally, DyAN could serve as a backbone for any multiagent task where roster size changes mid-episode, such as allies dying or reinforcements arriving, not only for curricula with fixed growing stages.
  • The reported superiority of sum aggregation suggests that for these combat tasks, counting and total-mass features over nearby units matter more than picking out specific individuals; a direct test would compare sum against attention-based aggregation on tasks where a particular enemy's identity is crucial.
  • Because the paper fixes total training steps rather than reporting steps-to-threshold, whether curriculum training also reduces wall-clock compute for reaching a given win rate is an open question that a time-to-threshold comparison would settle.
  • The curriculum schedule is manually designed; if the benefit scales with schedule quality, automatic curriculum generation over agent counts could yield further gains on even larger arenas.
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 / 4 minor

Summary. The paper proposes DyMA-CL, a curriculum learning framework for large-scale multiagent reinforcement learning that starts from small agent-count scenarios and progressively increases the number of agents. Three transfer mechanisms are introduced: Buffer Reuse, Curriculum Distillation, and Model Reload. To support Model Reload across changing input dimensions, the authors design DyAN, a GNN-based network that aggregates per-agent observations. Experiments on StarCraft II (IQL, VDN) and MAgent (IQL, PPO, A2C, ACER) report that curriculum training with DyAN improves final performance over learning-from-scratch baselines, with Model Reload performing best. The paper also analyzes aggregation operators (SUM/MEAN/MAX) via t-SNE visualizations.

Significance. If the empirical claims hold, the paper offers a practical recipe for scaling multiagent RL via agent-count curricula and a network design that makes parameter transfer feasible across tasks with varying input sizes. The consistent improvement across five algorithms and two environments is a useful existence proof, and the comparison of transfer mechanisms gives actionable insight. The manuscript is clearly presented and the experimental scope is substantial. The central limitation is that the headline comparison may conflate curriculum benefit with extra environment interactions in MAgent (and possibly VDN in StarCraft II); the paper also overstates 'state-of-the-art' comparisons. These are addressable by additional experimental controls.

major comments (3)
  1. [MAgent simulation and supplementary materials] The total environment steps for DyMA-CL on MAgent are 7,500 + 4,500 + 1,500 + 750 + 10,000 = 24,250 steps, whereas the paper only reports that the target task is trained 'for 1e+4 steps' and never states the budget of the 'w/o CL' baselines. If the baselines receive only the 10,000 target-task steps, the Table 1 gains (e.g., IQL Max survivors 20.35±4.87 vs 0.54±2.65) could be due to the 2.4× larger environment budget rather than curriculum transfer. The authors should either match total steps for the baselines or explicitly report the baseline budget and re-run at matched budget.
  2. [Abstract and Introduction] The claim that DyMA-CL 'greatly improves ... compared with state-of-the-art deep reinforcement learning approaches' is not supported by the baselines, which are IQL, VDN, PPO, A2C, and ACER trained from scratch. These are standard baselines, but not state-of-the-art for large-scale multiagent learning; no comparison is made with QMIX, mean-field MARL, or other curriculum/transfer methods. The claim should be qualified to 'from-scratch deep RL baselines' or the comparison set should be expanded.
  3. [Dynamic Number Agent Network, Definition 1] The semantic mapping function Φ(·) is formalized by an inequality, but the paper does not provide a quantitative test of this property on held-out states. The only evidence is the qualitative t-SNE visualization in Figure 7. Since Model Reload transfers parameters through the same GNN embeddings, the authors should validate the semantic mapping with a numerical measure (e.g., nearest-neighbor distances between semantic classes across scenarios) or explicitly state that the mapping is an assumption that is not independently verified.
minor comments (4)
  1. [Abstract] The phrase 'state dimension varies across curricula,, and existing' contains a double comma and a grammatical break; please fix the typo.
  2. [Figures 5 and 6] The StarCraft II results are reported without error bars or confidence intervals; please state the number of random seeds and include variance information in the plots.
  3. [Equation (2)] The displayed formula for the distillation loss is ambiguous in its typesetting (the division between the softmax terms is not clearly expressed); please rewrite the KL expression in a clearer form.
  4. [References] The reference list contains duplicate entries for Narvekar et al. 2017a and 2017b, which appear to be the same paper; please check and merge.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DyMA-CL is an empirical curriculum-training method whose reported gains are measured against external baselines, and no central claim reduces to a fitted input or a self-citation chain.

full rationale

The paper's central claim is an empirical one: curriculum training with DyMA-CL and the DyAN architecture improves large-scale multiagent learning relative to learning from scratch. The curriculum schedule, transfer mechanisms, and aggregation choices are design decisions validated by experiments, not quantities fitted to the target result. No equation in the paper defines a prediction in terms of the outcome it is said to predict, and no fitted parameter is renamed as a prediction. The formal Definition 1 (Semantic Mapping Function) is an assumption about the existence of a useful latent mapping; it is used to motivate the DyAN architecture, but the performance of that architecture is tested against external baselines rather than derived from the definition. The t-SNE analysis in Figure 7 is a post-hoc visualization of learned embeddings, not a circular justification. There are no load-bearing self-citations: the references are to standard external works on curriculum learning, multiagent RL, GNNs, and the benchmark environments, and no uniqueness theorem or architecture constraint is imported from the authors' own prior work. The paper explicitly acknowledges that the curriculum is manually designed and that automatic curriculum generation is future work; this is a limitation of autonomy, not circularity. The MAgent comparison may raise an experimental-budget question, because the text does not state whether the from-scratch baselines received the same total number of environment steps as the curriculum condition, but that is a potential confound in the empirical evaluation, not a circularity in the derivation. Overall, the derivation chain is self-contained against external empirical benchmarks, so the appropriate circularity score is 0.

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

The central claim rests on domain assumptions about the structure of multiagent observations (partial observability and state semanticity) rather than on a new physical entity. The free parameters are all training-design choices, not fitted physical constants, and they are either manually set or selected after observing performance. The standard POSG/RL framework is assumed as background.

free parameters (3)
  • curriculum schedule (training steps per task) = 1.5e6 steps per task (SC2); 7500/4500/1500/750/1e4 steps (MAgent)
    The number of steps spent on each curriculum task is chosen by hand and is acknowledged in the Discussion as a critical factor that the paper does not ablate.
  • distillation temperature omega = not specified
    Used in Equation (2) for the Curriculum Distillation transfer mechanism; no value or sensitivity analysis is reported in the paper.
  • GNN aggregation operator = SUM selected as best
    MAX, MEAN, SUM are compared and SUM is reported as best; the paper's final claims emphasize SUM results, a model selection made after seeing the outcomes without a significance test.
assumptions (3)
  • domain assumption Property 3: State Semanticity
    Assumes states of different dimensions from different curricula can be mapped to a shared latent space by a function Phi; this underpins DyAN and Model Reload (Section 'Dynamic Multiagent Curriculum Learning').
  • domain assumption Property 1: Partial Observability
    Assumes local observations in a large-scale scenario reduce to small-scale scenarios, justifying the curriculum ordering (Section 'Large-scale Multiagent Systems').
  • standard math Standard RL/POSG formulation
    Uses standard partially observable stochastic games and deep RL losses without modification (Background section).

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Few to More: Large-scale Dynamic Multiagent Curriculum Learning." pith.science (2026). https://pith.science/paper/QNLPPW2F

@misc{pith2026190902790,
  author       = {Pith},
  title        = {Pith review of: From Few to More: Large-scale Dynamic Multiagent Curriculum Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QNLPPW2F}},
  note         = {Machine review of arXiv:1909.02790}
}
read the original abstract

A lot of efforts have been devoted to investigating how agents can learn effectively and achieve coordination in multiagent systems. However, it is still challenging in large-scale multiagent settings due to the complex dynamics between the environment and agents and the explosion of state-action space. In this paper, we design a novel Dynamic Multiagent Curriculum Learning (DyMA-CL) to solve large-scale problems by starting from learning on a multiagent scenario with a small size and progressively increasing the number of agents. We propose three transfer mechanisms across curricula to accelerate the learning process. Moreover, due to the fact that the state dimension varies across curricula,, and existing network structures cannot be applied in such a transfer setting since their network input sizes are fixed. Therefore, we design a novel network structure called Dynamic Agent-number Network (DyAN) to handle the dynamic size of the network input. Experimental results show that DyMA-CL using DyAN greatly improves the performance of large-scale multiagent learning compared with state-of-the-art deep reinforcement learning approaches. We also investigate the influence of three transfer mechanisms across curricula through extensive simulations.

Figures

Figures reproduced from arXiv: 1909.02790 by the authors.

Figure 1
Figure 1. An example of DyMA-CL in StarCraft II. past few years, especially regarding its application to RL. The Curriculum Learning, consists of defining a set of source tasks and training the agent on each of them individually before progressing to learning on the full task. One major direction of applying CL to RL focuses on how to deal with increasingly complicated tasks. Andreas et al. (2017) used curriculum learning to … view at source ↗
Figure 2
Figure 2. An illustration of DyMA-CL using different transfer mechanisms. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Mapping original states to the semantic state space. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The network structure of DyAN. Dynamic Number Agent Network As mentioned above, each kind of transfer mechanisms can￾not directly be used in our DyMA-CL, since the number of agents varies across curricula and the dimension of each agent i’s observation o i t at each st…
Figure 5
Figure 5. Figure 5: Average win rate of IQL and VDN on DyMA-CL. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: The performance of VDN on a 15I scenario. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Embedding analysis for different aggregation mechanisms. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: An illustration of action range on MAgent. [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: An illustration of the network structure of DyAN for StarCraft II. [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 30 canonical work pages

  1. [1]

    Agarwal, A.; Kumar, S.; and Sycara, K. P. 2019. Learning transferable cooperative behavior in multi-agent teams. CoRR abs/1906.01202

  2. [2]

    Andreas, J.; Klein, D.; and Levine, S. 2017. Modular multitask reinforcement learning with policy sketches. In Proceedings of ICML , 166--175

  3. [3]

    Arnekvist, I.; Kragic, D.; and Stork, J. A. 2019. VPE: variational policy embedding for transfer reinforcement learning. In Proceedings of ICRA , 36--42

  4. [4]

    Bengio, Y.; Louradour, J.; Collobert, R.; and Weston, J. 2009. Curriculum learning. In Proceedings of ICML , 41--48

  5. [5]

    Busoniu, L.; Babuska, R.; and Schutter, B. D. 2008. A comprehensive survey of multiagent reinforcement learning. IEEE T SYST MAN CY C 38(2):156--172

  6. [6]

    Chen, Y.; Zhou, M.; Wen, Y.; Yang, Y.; Su, Y.; Zhang, W.; Zhang, D.; Wang, J.; and Liu, H. 2018. Factorized q-learning for large-scale multi-agent systems. arXiv preprint arXiv:1809.03738

  7. [7]

    Claus, C., and Boutilier, C. 1998. The dynamics of reinforcement learning in cooperative multiagent systems. AAAI/IAAI 1998:746--752

  8. [8]

    A.; Bernstein, D

    Hansen, E. A.; Bernstein, D. S.; and Zilberstein, S. 2004. Dynamic programming for partially observable stochastic games. In Proceedings of AAAI , 709--715

Show all 33 references
  1. [9]

    Z.; and Gruslys, A

    Hester, T.; Vecer \' k, M.; Pietquin, O.; Lanctot, M.; Schaul, T.; Piot, B.; Horgan, D.; Quan, J.; Sendonaris, A.; Osband, I.; Dulac - Arnold, G.; Agapiou, J.; Leibo, J. Z.; and Gruslys, A. 2018. Deep q-learning from demonstrations. In Proceedings of AAAI , 3223--3230

  2. [10]

    A.; Matthey, L.; Burgess, C.; Pritzel, A.; Botvinick, M.; Blundell, C.; and Lerchner, A

    Higgins, I.; Pal, A.; Rusu, A. A.; Matthey, L.; Burgess, C.; Pritzel, A.; Botvinick, M.; Blundell, C.; and Lerchner, A. 2017. DARLA: improving zero-shot transfer in reinforcement learning. In Proceedings of ICML , 1480--1490

  3. [11]

    Jiang, J., and Lu, Z. 2018. Learning attentional communication for multi-agent cooperation. In Proceedings of NeurIPS , 7254--7264

  4. [12]

    Littman, M. L. 1994. Markov games as a framework for multi-agent reinforcement learning. In Proceedings of ICML , 157--163

  5. [13]

    A.; Veness, J.; Bellemare, M

    Mnih, V.; Kavukcuoglu, K.; Silver, D.; Rusu, A. A.; Veness, J.; Bellemare, M. G.; Graves, A.; Riedmiller, M.; Fidjeland, A. K.; Ostrovski, G.; et al. 2015. Human-level control through deep reinforcement learning. Nature 518(7540):529

  6. [14]

    P.; Mirza, M.; Graves, A.; Lillicrap, T.; Harley, T.; Silver, D.; and Kavukcuoglu, K

    Mnih, V.; Badia, A. P.; Mirza, M.; Graves, A.; Lillicrap, T.; Harley, T.; Silver, D.; and Kavukcuoglu, K. 2016. Asynchronous methods for deep reinforcement learning. In Proceedings of ICML , 1928--1937

  7. [15]

    Narvekar, S., and Stone, P. 2019. Learning curriculum policies for reinforcement learning. In Proceedings of AAMAS , 25--33

  8. [16]

    Narvekar, S.; Sinapov, J.; Leonetti, M.; and Stone, P. 2016. Source task creation for curriculum learning. In Proceedings of AAMAS , 566--574

  9. [17]

    Narvekar, S.; Sinapov, J.; and Stone, P. 2017a. Autonomous task sequencing for customized curriculum design in reinforcement learning. In Proceedings of IJCAI , 2536--2542

  10. [18]

    Narvekar, S.; Sinapov, J.; and Stone, P. 2017b. Autonomous task sequencing for customized curriculum design in reinforcement learning. In Proceedings of IJCAI , 2536--2542

  11. [19]

    S.; Farquhar, G.; Foerster, J.; and Whiteson, S

    Rashid, T.; Samvelyan, M.; Witt, C. S.; Farquhar, G.; Foerster, J.; and Whiteson, S. 2018. Qmix: Monotonic value function factorisation for deep multi-agent reinforcement learning. In Proceedings of ICML , 4292--4301

  12. [20]

    A.; Colmenarejo, S

    Rusu, A. A.; Colmenarejo, S. G.; G \" u l c ehre, C .; Desjardins, G.; Kirkpatrick, J.; Pascanu, R.; Mnih, V.; Kavukcuoglu, K.; and Hadsell, R. 2016. Policy distillation. In Proceedings of ICLR

  13. [21]

    S.; Farquhar, G.; Nardelli, N.; Rudner, T

    Samvelyan, M.; Rashid, T.; de Witt, C. S.; Farquhar, G.; Nardelli, N.; Rudner, T. G.; Hung, C.-M.; Torr, P. H.; Foerster, J.; and Whiteson, S. 2019. The starcraft multi-agent challenge. arXiv preprint arXiv:1902.04043

  14. [22]

    Schulman, J.; Wolski, F.; Dhariwal, P.; Radford, A.; and Klimov, O. 2017. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347

  15. [23]

    Singh, A.; Jain, T.; and Sukhbaatar, S. 2019. Individualized controlled continuous communication model for multiagent cooperative and competitive tasks. In Proceedings of ICLR

  16. [24]

    M.; Zambaldi, V

    Sunehag, P.; Lever, G.; Gruslys, A.; Czarnecki, W. M.; Zambaldi, V. F.; Jaderberg, M.; Lanctot, M.; Sonnerat, N.; Leibo, J. Z.; Tuyls, K.; and Graepel, T. 2018. Value-decomposition networks for cooperative multi-agent learning based on team reward. In Proceedings of AAMAS , 2085--2087

  17. [25]

    S., and Barto, A

    Sutton, R. S., and Barto, A. G. 2018. Reinforcement learning: An introduction . MIT press

  18. [26]

    Tampuu, A.; Matiisen, T.; Kodelja, D.; Kuzovkin, I.; Korjus, K.; Aru, J.; Aru, J.; and Vicente, R. 2017. Multiagent cooperation and competition with deep reinforcement learning. PLOS ONE 12:1--15

  19. [27]

    Wang, Z.; Bapst, V.; Heess, N.; Mnih, V.; Munos, R.; Kavukcuoglu, K.; and de Freitas, N. 2017. Sample efficient actor-critic with experience replay. In Proceedings of ICLR

  20. [28]

    Wattenberg, M.; Vi \'e gas, F.; and Johnson, I. 2016. How to use t-sne effectively. Distill 1(10):e2

  21. [29]

    Wu, Y., and Tian, Y. 2017. Training agent for first-person shooter game with actor-critic curriculum learning. In Proceedings of ICLR

  22. [30]

    Xu, K.; Hu, W.; Leskovec, J.; and Jegelka, S. 2019. How powerful are graph neural networks? In Proceedings of ICLR

  23. [31]

    Yang, Y.; Luo, R.; Li, M.; Zhou, M.; Zhang, W.; and Wang, J. 2018a. Mean field multi-agent reinforcement learning. In Proceedings of ICML , volume 80, 5571--5580

  24. [32]

    Yang, Y.; Yu, L.; Bai, Y.; Wen, Y.; Zhang, W.; and Wang, J. 2018b. A study of AI population dynamics with million-agent reinforcement learning. In Proceedings of AAMAS , 2133--2135

  25. [33]

    Zheng, L.; Yang, J.; Cai, H.; Zhou, M.; Zhang, W.; Wang, J.; and Yu, Y. 2018. Magent: A many-agent reinforcement learning platform for artificial collective intelligence. In Proceedings of AAAI

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.