Pith. sign in

REVIEW 10 cited by

Self-rewarding correction for mathematical reasoning

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2502.19613 v1 pith:X3EKF6HQ submitted 2025-02-26 cs.AI cs.LG

Wei Xiong , Hanning Zhang , Chenlu Ye , Lichang Chen , Nan Jiang , Tong Zhang This is my paper · ORCID
classification cs.AIcs.LG
keywords modelsreasoningself-rewardingself-correctionoutputsallowsapproachdata
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

We study self-rewarding reasoning large language models (LLMs), which can simultaneously generate step-by-step reasoning and evaluate the correctness of their outputs during the inference time-without external feedback. This integrated approach allows a single model to independently guide its reasoning process, offering computational advantages for model deployment. We particularly focus on the representative task of self-correction, where models autonomously detect errors in their responses, revise outputs, and decide when to terminate iterative refinement loops. To enable this, we propose a two-staged algorithmic framework for constructing self-rewarding reasoning models using only self-generated data. In the first stage, we employ sequential rejection sampling to synthesize long chain-of-thought trajectories that incorporate both self-rewarding and self-correction mechanisms. Fine-tuning models on these curated data allows them to learn the patterns of self-rewarding and self-correction. In the second stage, we further enhance the models' ability to assess response accuracy and refine outputs through reinforcement learning with rule-based signals. Experiments with Llama-3 and Qwen-2.5 demonstrate that our approach surpasses intrinsic self-correction capabilities and achieves performance comparable to systems that rely on external reward models.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 10 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. ReSum: Synergizing LLM Reasoning and Summarization with Reinforcement Learning

    cs.AI 2026-06 unverdicted novelty 7.0 of 10

    ReSum trains LLMs via RLVR to self-summarize reasoning trajectories, yielding 4% average performance gains and 18.6% shorter rollouts through contrastive rollout branches.

  2. Cognitive Episodes in LLM Reasoning Traces Enable Interpretable Human Item Difficulty Prediction

    cs.CL 2026-06 unverdicted novelty 6.5 of 10

    Structured cognitive-episode features from LRM reasoning traces, combined with item semantics, improve human item-difficulty prediction and show harder items drive more implementation-centered, iterative solving.

  3. rePIRL: Learn PRM with Inverse RL for LLM Reasoning

    cs.LG 2026-02 unverdicted novelty 6.0 of 10

    rePIRL learns token-level process rewards for LLM reasoning via a guided-cost-learning-style IRL objective, and shows these rewards improve reasoning policies on math/coding benchmarks.

  4. CPMobius: Iterative Coach-Player Reasoning for Data-Free Reinforcement Learning

    cs.CL 2026-02 conditional novelty 6.0 of 10

    CPMobius uses iterative coach-player reinforcement learning to improve mathematical reasoning in LLMs without external training data, yielding +4.9 average accuracy gains on Qwen2.5-Math-7B-Instruct.

  5. Breaking the Self-Confirming Loop: Diagnosing and Mitigating Systemic Reward Bias in Self-Rewarding RL

    cs.LG 2025-10 conditional novelty 6.0 of 10

    Self-rewarding RL can be stabilized by ensembling multiple policy models' majority-vote rewards, reaching within 3.6% of verifiable-reward RL on math benchmarks.

  6. PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier

    cs.CL 2025-06 conditional novelty 6.0 of 10

    A new multi-turn reinforcement learning framework trains a single LLM to both solve math problems and verify its own solutions, revising only when its verifier finds a mistake.

  7. Consistent Paths Lead to Truth: Self-Rewarding Reinforcement Learning for LLM Reasoning

    cs.AI 2025-06 conditional novelty 6.0 of 10

    CoVo trains LLMs with a self-generated reward based on the consistency and volatility of intermediate reasoning states, matching supervised RL performance without external labels.

  8. Scaling Test-time Compute for LLM Agents

    cs.AI 2025-06 conditional novelty 5.0 of 10

    On the GAIA benchmark, Best-of-N sampling with list-wise answer selection gives the largest agent gains, and selective (score-triggered) reflection beats reflection at every step.

  9. Boosting LLM Reasoning via Spontaneous Self-Correction

    cs.AI 2025-06 conditional novelty 5.0 of 10

    SPOC trains LLMs to interleave self-verification and solution attempts in a single pass, reporting gains on math benchmarks, though most gains come from stronger first attempts.

  10. MoDoMoDo: Multi-Domain Data Mixtures for Multimodal LLM Reinforcement Learning

    cs.CV 2025-05 conditional novelty 5.0 of 10

    Multi-domain RLVR data mixing, guided by a quadratic surrogate fitted to 11 pilot runs, improves a Qwen2-VL-2B model's out-of-distribution accuracy by about 5 points over uniform mixing.

Pith tools