mlsanityData to 7 Oct 2026

IRDS: Interpretable RLVR Data Selection via Verifier-Coupled Sparse Autoencoder Coverage

arXiv 2605.28247 ↗First posted 27 May 2026cs.LG cs.AIcited by 1

On the map: LLM reasoning → Reinforcement learning for reasoning

Abstract (arXiv)

Reinforcement learning with verifiable rewards (RLVR) has become a key technique for en- hancing LLM reasoning, yet its data ineffi- ciency remains a major bottleneck. Existing methods address this problem only partially, each missing at least one of subset-level cov- erage, verifier signal use, or interpretability. To address this gap, we present IRDS (Inter- pretable RLVR Data Selection), which selects RLVR training instances on a sparse autoen- coder (SAE) cluster basis so the selection itself is auditable on recognizable problem motifs. To select instances the model both fails on and can still learn from, we introduce a verifier- coupled coverage objective on the SAE basis and solve it by greedy log-determinant max- imization. Experiments on three instruction- tuned models and six math reasoning bench- marks show that IRDS achieves the highest overall accuracy, exceeding the strongest base- line by +3.9/+4.0 pp on the two Qwen models and by +0.5 pp on Llama-3.1-8B, while run- ning an order of magnitude cheaper than the trajectory-based baseline.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning. Builds on 21 of its 35 references that are in the map. First posted on arXiv 27 May 2026.

Builds on

21 of 35 references.

Closest ancestors · 6

Same direction, newest first

  1. 2026Learn More with Less: Uncertainty Consistency Guided Query Selection for RLVR30 Jan
  2. 2025Data-Efficient RLVR via Off-Policy Influence Guidance30 Oct
  3. 2025Towards High Data Efficiency in Reinforcement Learning with Verifiable Reward1 Sep
  4. 2025Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay5 Jun
  5. 2025LIMR: Less is More for RL Scaling17 Feb
  6. 2023Let's Verify Step by Step31 May

Other references · 15

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  2. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 1,980
  3. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 1,906
  4. 2022Solving Quantitative Reasoning Problems with Language Modelscited by 1,359
  5. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,359
  6. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,099
  7. 2024LESS: Selecting Influential Data for Targeted Instruction Tuningcited by 514
  8. 2023TRAK: Attributing Model Behavior at Scalecited by 273
  9. 2020ReClor: A Reading Comprehension Dataset Requiring Logical Reasoningcited by 211
  10. 2024OpenMathInstruct-2: Accelerating AI for Math with Massive Open-Source Instruction Datacited by 194
  11. 2024Perplexed by Perplexity: Perplexity-Based Data Pruning With Small Reference Modelscited by 83
  12. 2025Prompt Curriculum Learning for Efficient LLM Post-Trainingcited by 40
3 more
  1. 2026Do Sparse Autoencoders Identify Reasoning Features in Language Models?cited by 11
  2. 2026Prioritized Replay for RL Post-trainingcited by 4
  3. 2026Rethinking Representativeness and Diversity in Dynamic Data Selectioncited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Per-Matrix Optimality Is Not Enough: Three-Level Optimization for Low-Rank LLM Compressioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Single-Rollout Hidden-State Dynamics for Training-Free RLVR Data Selection
  2. 2026Smart Picks in the Dark: Towards Efficient RLVR for Reasoning via Tracing Metacognitive Pivots
  3. 2025RLPR: Extrapolating RLVR to General Domains without Verifiers
  4. 2025Generalization of RLVR Using Causal Reasoning as a Testbed
  5. 2025TraPO: A Semi-Supervised Reinforcement Learning Framework for Boosting LLM Reasoning
  6. 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?
  7. 2025Limits of Generalization in RLVR: Two Case Studies in Mathematical Reasoning
  8. 2026Boosting LLM Exploration via Weak-Model Guidance in RLVR