Pith. sign in

REVIEW 14 cited by

rStar2-Agent: Agentic Reasoning Technical Report

Not yet reviewed by Pith; the record is open.

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 2508.20722 v1 pith:J5UDUKID submitted 2025-08-28 cs.CL

rStar2-Agent: Agentic Reasoning Technical Report

classification cs.CL
keywords agenticcodemodelenvironmentreasoningrstar2-agenttrainingadvanced
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We introduce rStar2-Agent, a 14B math reasoning model trained with agentic reinforcement learning to achieve frontier-level performance. Beyond current long CoT, the model demonstrates advanced cognitive behaviors, such as thinking carefully before using Python coding tools and reflecting on code execution feedback to autonomously explore, verify, and refine intermediate steps in complex problem-solving. This capability is enabled through three key innovations that makes agentic RL effective at scale: (i) an efficient RL infrastructure with a reliable Python code environment that supports high-throughput execution and mitigates the high rollout costs, enabling training on limited GPU resources (64 MI300X GPUs); (ii) GRPO-RoC, an agentic RL algorithm with a Resample-on-Correct rollout strategy that addresses the inherent environment noises from coding tools, allowing the model to reason more effectively in a code environment; (iii) An efficient agent training recipe that starts with non-reasoning SFT and progresses through multi-RL stages, yielding advanced cognitive abilities with minimal compute cost. To this end, rStar2-Agent boosts a pre-trained 14B model to state of the art in only 510 RL steps within one week, achieving average pass@1 scores of 80.6% on AIME24 and 69.8% on AIME25, surpassing DeepSeek-R1 (671B) with significantly shorter responses. Beyond mathematics, rStar2-Agent-14B also demonstrates strong generalization to alignment, scientific reasoning, and agentic tool-use tasks. Code and training recipes are available at https://github.com/microsoft/rStar.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 14 Pith papers

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

  1. Teaching Language Models to Think in Code

    cs.CL 2026-05 unverdicted novelty 7.0

    ThinC trains small models to reason primarily in code rather than natural language, outperforming tool-integrated baselines and even larger models on competition math benchmarks.

  2. Fine-Tuning Small Reasoning Models for Quantum Field Theory

    cs.LG 2026-04 unverdicted novelty 7.0

    Small 7B reasoning models were fine-tuned on synthetic and curated QFT problems using RL and SFT, yielding performance gains, error analysis, and public release of data and traces.

  3. Agent Explorative Policy Optimization for Multimodal Agentic Reasoning

    cs.CL 2026-05 unverdicted novelty 6.0

    AXPO addresses the Thinking-Acting Gap in agentic RL training by targeted resampling of tool calls in all-wrong subgroups, delivering +1.8pp gains over GRPO on nine multimodal benchmarks with an 8B model beating a 32B...

  4. Implicit Hierarchical GRPO: Decoupling Tool Invocation from Execution for Tool-Integrated Mathematical Reasoning

    cs.CL 2026-05 unverdicted novelty 6.0

    IH-GRPO introduces implicit hierarchical control via a surrogate loss to decouple tool invocation from execution in LLMs, reporting 1.87-2.53% gains on math reasoning benchmarks.

  5. Teaching Language Models to Think in Code

    cs.CL 2026-05 unverdicted novelty 6.0

    ThinC trains smaller language models to reason entirely in code after minimal NL planning, outperforming tool-integrated baselines and even much larger models on competition math benchmarks.

  6. Teaching Thinking Models to Reason with Tools: A Full-Pipeline Recipe for Tool-Integrated Reasoning

    cs.CL 2026-05 unverdicted novelty 6.0

    A training recipe for tool-integrated reasoning models achieves state-of-the-art open-source results on math benchmarks such as 96.7% and 99.2% on AIME 2025 at 4B and 30B scales by balancing tool-use trajectories and ...

  7. Every Step Counts: Step-Level Credit Assignment for Tool-Integrated Text-to-SQL

    cs.CL 2026-05 unverdicted novelty 6.0

    FineStep adds step-level process rewards and credit assignment to tool-augmented Text-to-SQL, achieving 3.25% higher execution accuracy than GRPO on BIRD while cutting redundant tool calls.

  8. T$^2$PO: Uncertainty-Guided Exploration Control for Stable Multi-Turn Agentic Reinforcement Learning

    cs.AI 2026-05 unverdicted novelty 6.0

    T²PO improves stability and performance in multi-turn agentic RL by using uncertainty dynamics at token and turn levels to guide exploration and avoid wasted rollouts.

  9. JigsawRL: Assembling RL Pipelines for Efficient LLM Post-Training

    cs.LG 2026-04 unverdicted novelty 6.0

    JigsawRL achieves up to 1.85x higher throughput in LLM RL pipelines via pipeline multiplexing, sub-stage graphs, and look-ahead scheduling compared to prior systems.

  10. The Landscape of Agentic Reinforcement Learning for LLMs: A Survey

    cs.AI 2025-09 accept novelty 6.0

    Survey that defines agentic RL for LLMs via POMDPs, introduces a taxonomy of planning/tool-use/memory/reasoning capabilities and domains, and compiles open environments from over 500 papers.

  11. Tool-Aware Optimization with Entropy Guidance for Efficient Agentic Reinforcement Learning

    cs.LG 2026-06 unverdicted novelty 5.0

    TAO-RL improves agentic RL by filtering degenerate trajectories and reshaping advantages with tool-aware entropy bonuses, yielding better performance on reasoning benchmarks.

  12. CP-Agent: A Calibrated Risk-Controlled Agent for Feedback-Driven Competitive Programming

    cs.CL 2026-05 unverdicted novelty 5.0

    CP-Agent improves LLM competitive programming performance via calibrated feedback mechanisms that target false-admission risk, evidence against bad programs, and success hazard.

  13. SHAPE: Stage-aware Hierarchical Advantage via Potential Estimation for LLM Reasoning

    cs.LG 2026-04 unverdicted novelty 5.0

    SHAPE improves average math reasoning accuracy by 3% while cutting token use by 30% through stage-aware hierarchical advantage and entropy-driven token redistribution.

  14. Discovering Millions of Interpretable Features with Sparse Autoencoders

    cs.LG 2026-06 unverdicted novelty 3.0

    Trains and releases SAEs for Qwen3-1.7B/4B/8B models with layer-wise coverage and demonstrates causal steering of refusal via selected features.