# Mastering Rate based Curriculum Learning

arXiv: https://arxiv.org/abs/2008.06456 · HTML: https://mlsanity.com/paper/2008.06456/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Aug 2020.

## Built on by (most cited first, 9 of 9)

- [Kolb-Based Experiential Learning for Generalist Agents with Human-Level Kaggle Data Science Performance](https://mlsanity.com/paper/2411.03562/) (2024)
- [Analyzing the Nuances of Transformers' Polynomial Simplification Abilities](https://mlsanity.com/paper/2104.14095/) (2021)
- [Curriculum-based Asymmetric Multi-task Reinforcement Learning](https://mlsanity.com/paper/2211.03352/) (2022)
- [Loss-Guided Auxiliary Agents for Overcoming Mode Collapse in GFlowNets](https://mlsanity.com/paper/2505.15251/) (2025)
- [Communication Enables Cooperation in LLM Agents: A Comparison with Curriculum-Based Approaches](https://mlsanity.com/paper/2510.05748/) (2025)
- [Understanding Curriculum Learning in Policy Optimization for Online Combinatorial Optimization](https://mlsanity.com/paper/2202.05423/) (2022)
- [Curriculum-Augmented GFlowNets For mRNA Sequence Generation](https://mlsanity.com/paper/2510.03811/) (2025)
- [Improved Exploration in GFlownets via Enhanced Epistemic Neural Networks](https://mlsanity.com/paper/2506.16313/) (2025)
- [Frontier Learning: Training LLM Reasoners at the Edge of Capability](https://mlsanity.com/paper/2609.35426/) (2026)

## Abstract (arXiv)

Recent automatic curriculum learning algorithms, and in particular Teacher-Student algorithms, rely on the notion of learning progress, making the assumption that the good next tasks are the ones on which the learner is making the fastest progress or digress. In this work, we first propose a simpler and improved version of these algorithms. We then argue that the notion of learning progress itself has several shortcomings that lead to a low sample efficiency for the learner. We finally propose a new algorithm, based on the notion of mastering rate, that significantly outperforms learning progress-based algorithms.