mlsanityData to 7 Oct 2026

Actor-Curator: Co-adaptive Curriculum Learning via Policy-Improvement Bandits for RL Post-Training

arXiv 2602.20532 ↗First posted 24 Feb 2026cs.LG cs.AI cs.CLcited by 7

Not on the map. Why?

Abstract (arXiv)

Post-training large foundation models with reinforcement learning typically relies on massive and heterogeneous datasets, making effective curriculum learning both critical and challenging. In this work, we propose ACTOR-CURATOR, a scalable and fully automated curriculum learning framework for reinforcement learning post-training of large language models (LLMs). ACTOR-CURATOR learns a neural curator that dynamically selects training problems from large problem banks by directly optimizing for expected policy performance improvement. We formulate problem selection as a non-stationary stochastic bandit problem, derive a principled loss function based on online stochastic mirror descent, and establish regret guarantees under partial feedback. Empirically, ACTOR-CURATOR consistently outperforms uniform sampling and strong curriculum baselines across a wide range of challenging reasoning benchmarks, demonstrating improved training stability and efficiency. Notably, it achieves relative gains of 28.6% on AIME2024 and 30.5% on ARC-1D over the strongest baseline and up to 80% speedup. These results suggest that ACTOR-CURATOR is a powerful and practical approach for scalable LLM post-training.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 32 of its 44 references that are in the map. First posted on arXiv 24 Feb 2026.

Builds on

32 of 44 references.

References · 32

  1. 2017Proximal Policy Optimization Algorithmscited by 14,441
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  3. 2025Qwen3 Technical Reportcited by 8,195
  4. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  6. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,678
  7. 2015Trust Region Policy Optimizationcited by 2,503
  8. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 1,980
  9. 2025Understanding R1-Zero-Like Training: A Critical Perspectivecited by 1,276
  10. 2025Group Sequence Policy Optimizationcited by 788
  11. 2024ReST-MCTS*: LLM Self-Training via Process Reward Guided Tree Searchcited by 338
  12. 2025VAPO: Efficient and Reliable Reinforcement Learning for Advanced Reasoning Taskscited by 281
20 more
  1. 2024Rewarding Progress: Scaling Automated Process Verifiers for LLM Reasoningcited by 267
  2. 2017Automatic Goal Generation for Reinforcement Learning Agentscited by 252
  3. 2017Automated Curriculum Learning for Neural Networkscited by 251
  4. 2017Reverse Curriculum Generation for Reinforcement Learningcited by 185
  5. 2025Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyondcited by 175
  6. 2025Agentic Reinforced Policy Optimizationcited by 166
  7. 2025REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewardscited by 121
  8. 2025Act Only When It Pays: Efficient Reinforcement Learning for LLM Reasoning via Selective Rolloutscited by 108
  9. 2025Self-Evolving Curriculum for LLM Reasoningcited by 107
  10. 2025Curriculum Reinforcement Learning from Easy to Hard Tasks Improves LLM Reasoningcited by 89
  11. 2025AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learningcited by 85
  12. 2023LLMs and the Abstraction and Reasoning Corpus: Successes, Failures, and the Importance of Object-based Representationscited by 81
  13. 2025DUMP: Automated Distribution-Level Curriculum Learning for RL-based LLM Post-trainingcited by 45
  14. 2025Prompt Curriculum Learning for Efficient LLM Post-Trainingcited by 40
  15. 2025SPEED-RL: Faster Training of Reasoning Models via Online Curriculum Learningcited by 27
  16. 2024Scalable Reinforcement Post-Training Beyond Static Human Prompts: Evolving Alignment via Asymmetric Self-Playcited by 21
  17. 2023Proximal Curriculum for Reinforcement Learning Agentscited by 19
  18. 2024Strategist: Self-improvement of LLM Decision Making via Bi-Level Tree Searchcited by 16
  19. 2021Curriculum Design for Teaching via Demonstrations: Theory and Applicationscited by 9
  20. 2025DISC: Dynamic Decomposition Improves LLM Inference Scalingcited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 7.

7 papers.

  1. 2026Learning Agentic Policy from Action Guidancecited by 3
  2. 2026CuSearch: Curriculum Rollout Sampling via Search Depth for Agentic RAGcited by 1
  3. 2026Beyond Solvability: Task Learnability as a Static Prior for LLM RL Post-Trainingcited by 1
  4. 2026Reinforcement Learning for Symbolic Equation Solvingcited by 0
  5. 2026Rollout Efficiency in Reinforcement Learning for Reasoning Large Language Models: A Taxonomy and Future Directionscited by 0
  6. 2026Frontier Learning: Training LLM Reasoners at the Edge of Capabilitycited by 0
  7. 2026Shockingly Simple Self-retrospection Improves Agentic Models Without RLcited by 0