mlsanityData to 14 Oct 2026

Adaptive Ability Decomposing for Unlocking Large Reasoning Model Effective Reinforcement Learning

arXiv 2602.00759 ↗First posted 31 Jan 2026cs.CL cs.AIcited by 1

On the map: LLM reasoning → Reinforcement learning for reasoning

Abstract (arXiv)

Reinforcement learning with verifiable rewards (RLVR) has shown great potential to enhance the reasoning ability of large language models (LLMs). However, due to the limited amount of information provided during the RLVR process, the model can only engage in largely blind exploration, which often results in failure on challenging problems. To provide additional information for the RLVR process without relying on a teacher model, we propose A2D, an Adaptive Ability Decomposing method for enhancing the effectiveness of RLVR. Specifically, we first train a decomposer via RLVR without distillation, enabling it to decompose complex questions into a set of simpler sub-questions. Next, we use this decomposer to annotate sub-questions for each question in the training dataset, and then train the reasoner under RLVR with sub-question guidance. To better understand A2D, we first compare its performance with competitive baselines, showing its effectiveness. Next, we observe that our method functions as a plug-and-play module that can be applied to different RLVR algorithms. Furthermore, we conduct an analysis of the decomposer, revealing how the RLVR process affects its performance and behavior, and which type of guidance is better suited for enhancing the reasoner's exploration and exploitation abilities.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning. Builds on 9 of its 81 references that are in the map. First posted on arXiv 31 Jan 2026.

Builds on

9 of 81 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025Differential Smoothing Mitigates Sharpening and Improves LLM Reasoning25 Nov
  2. 2025Masked-and-Reordered Self-Supervision for Reinforcement Learning from Verifiable Rewards21 Nov

Other references · 7

  1. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,986
  2. 2025The Invisible Leash: Why RLVR May or May Not Escape Its Origincited by 97
  3. 2025Learning to Reason Across Parallel Samples for LLM Reasoningcited by 32
  4. 2025Asymmetric REINFORCE for off-Policy Reinforcement Learning: Balancing positive and negative rewardscited by 29
  5. 2025Stable Gradients for Stable Learning at Scale in Deep Reinforcement Learningcited by 23
  6. 2025Learning from Peers in Reasoning Modelscited by 16
  7. 2025Toward Honest Language Models for Deductive Reasoningcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Low-rank Optimization Trajectories Modeling for LLM RLVR Accelerationcited by 4

Similar, not cited

Close in wording, no citation either way

  1. 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?
  2. 2025Extending RLVR to Open-Ended Tasks via Verifiable Multiple-Choice Reformulation
  3. 2025Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs
  4. 2025The Reasoning Boundary Paradox: How Reinforcement Learning Constrains Language Models
  5. 2026Reinforcement Learning with Verifiable Rewards for Small Search Agents
  6. 2026Rethinking Multiple-Choice Questions for RLVR: Unlocking Potential via Distractor Design
  7. 2026RLVR Training of LLMs Does Not Improve Thinking Ability for General QA: Evaluation Method and a Simple Solution
  8. 2025A Simple "Motivation" Can Enhance Reinforcement Finetuning of Large Reasoning Models