# Meta Automatic Curriculum Learning

arXiv: https://arxiv.org/abs/2011.08463 · HTML: https://mlsanity.com/paper/2011.08463/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Nov 2020.

## Built on by (most cited first, 6 of 6)

- [Curriculum Learning: A Survey](https://mlsanity.com/paper/2101.10382/) (2021)
- [TeachMyAgent: a Benchmark for Automatic Curriculum Learning in Deep RL](https://mlsanity.com/paper/2103.09815/) (2021)
- [Towards Skilled Population Curriculum for Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2302.03429/) (2023)
- [Learning Multi-Objective Curricula for Robotic Policy Learning](https://mlsanity.com/paper/2110.03032/) (2021)
- [Driver Dojo: A Benchmark for Generalizable Reinforcement Learning for Autonomous Driving](https://mlsanity.com/paper/2207.11432/) (2022)
- [Learning to Order: Task Sequencing as In-Context Optimization](https://mlsanity.com/paper/2603.14550/) (2026)

## Abstract (arXiv)

A major challenge in the Deep RL (DRL) community is to train agents able to generalize their control policy over situations never seen in training. Training on diverse tasks has been identified as a key ingredient for good generalization, which pushed researchers towards using rich procedural task generation systems controlled through complex continuous parameter spaces. In such complex task spaces, it is essential to rely on some form of Automatic Curriculum Learning (ACL) to adapt the task sampling distribution to a given learning agent, instead of randomly sampling tasks, as many could end up being either trivial or unfeasible. Since it is hard to get prior knowledge on such task spaces, many ACL algorithms explore the task space to detect progress niches over time, a costly tabula-rasa process that needs to be performed for each new learning agents, although they might have similarities in their capabilities profiles. To address this limitation, we introduce the concept of Meta-ACL, and formalize it in the context of black-box RL learners, i.e. algorithms seeking to generalize curriculum generation to an (unknown) distribution of learners. In this work, we present AGAIN, a first instantiation of Meta-ACL, and showcase its benefits for curriculum generation over classical ACL in multiple simulated environments including procedurally generated parkour environments with learners of varying morphologies. Videos and code are available at https://sites.google.com/view/meta-acl .