Pith. sign in

REVIEW 29 cited by

Competitive Programming with Large Reasoning Models

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 2502.06807 v2 pith:HF766WXI submitted 2025-02-03 cs.LG cs.AIcs.CL

OpenAI: Ahmed El-Kishky , Alexander Wei , Andre Saraiva , Borys Minaiev , Daniel Selsam , David Dohan , Francis Song , Hunter Lightman
show 16 more authors
Ignasi Clavera Jakub Pachocki Jerry Tworek Lorenz Kuhn Lukasz Kaiser Mark Chen Max Schwarzer Mostafa Rohaninejad Nat McAleese o3 contributors Oleg Mürk Rhythm Garg Rui Shu Szymon Sidor Vineet Kosaraju Wenda Zhou
This is my paper
classification cs.LGcs.AIcs.CL
keywords modelso1-ioireasoningdomain-specificgeneral-purposegoldhand-craftedstrategies
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

We show that reinforcement learning applied to large language models (LLMs) significantly boosts performance on complex coding and reasoning tasks. Additionally, we compare two general-purpose reasoning models - OpenAI o1 and an early checkpoint of o3 - with a domain-specific system, o1-ioi, which uses hand-engineered inference strategies designed for competing in the 2024 International Olympiad in Informatics (IOI). We competed live at IOI 2024 with o1-ioi and, using hand-crafted test-time strategies, placed in the 49th percentile. Under relaxed competition constraints, o1-ioi achieved a gold medal. However, when evaluating later models such as o3, we find that o3 achieves gold without hand-crafted domain-specific strategies or relaxed constraints. Our findings show that although specialized pipelines such as o1-ioi yield solid improvements, the scaled-up, general-purpose o3 model surpasses those results without relying on hand-crafted inference heuristics. Notably, o3 achieves a gold medal at the 2024 IOI and obtains a Codeforces rating on par with elite human competitors. Overall, these results indicate that scaling general-purpose reinforcement learning, rather than relying on domain-specific techniques, offers a robust path toward state-of-the-art AI in reasoning domains, such as competitive programming.

Discussion (0). Sign in to comment.

Forward citations

Cited by 29 Pith papers

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

  1. AllocBench: Measuring Online Tool Allocation Capability in LLM Agents

    cs.LG 2026-07 conditional novelty 7.0 of 10

    LLM agents allocate budgets near-optimally in abstract tasks but lose this ability when the same decisions require writing reusable scripts.

  2. Embarrassingly Simple Self-Distillation Improves Code Generation

    cs.CL 2026-04 conditional novelty 7.0 of 10

    Simple self-distillation—fine-tuning a code model on its own temperature-sampled, truncated outputs—raises LiveCodeBench pass@1 substantially without verifiers, teachers, or RL.

  3. CodeContests+: High-Quality Test Case Generation for Competitive Programming

    cs.SE 2025-06 conditional novelty 7.0 of 10

    A new dataset, CodeContests+, uses an LLM agent system to generate and validate competitive programming test cases, improving evaluation accuracy and reinforcement learning training.

  4. LA-RL: Label-Aware Self-Reflection for Reinforcement Learning in Information Extraction

    cs.CL 2026-07 conditional novelty 6.0 of 10

    Label-aware diagnostic reflection plus two-stage outcome GRPO improves same-backbone IE F1 over SFT, with larger gains under relation-extraction domain shift.

  5. Frontier Language Models Struggle to Copy: Text Can Be Better Viewed in 2D

    cs.CL 2026-07 conditional novelty 6.0 of 10

    2D-RoPE, which arranges text by line breaks into rows and columns, lets Transformers copy strings hundreds of times longer than training lengths, while standard 1D positional encodings fail on the same task.

  6. Seesaw: Accelerating Training by Balancing Learning Rate and Batch Size Scheduling

    cs.LG 2025-10 conditional novelty 6.0 of 10

    When a cosine schedule would halve the learning rate, Seesaw cuts it by √2 and doubles the batch, matching loss curves with ~36% fewer serial steps.

  7. 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.

  8. AR$^2$: Adversarial Reinforcement Learning for Abstract Reasoning in Large Language Models

    cs.CL 2025-08 conditional novelty 6.0 of 10

    Adversarial teacher-student RL on computationally equivalent narrative rewrites of coding problems yields modest but consistent pass@1 gains on competitive programming benchmarks.

  9. SSRL: Self-Search Reinforcement Learning

    cs.CL 2025-08 unverdicted novelty 6.0 of 10

    SSRL, a training pipeline that uses an LLM's own repeated sampling as a search environment for RL, improves question answering without external tools and transfers to real search engines.

  10. StepFun-Prover Preview: Let's Think and Verify Step by Step

    cs.AI 2025-07 conditional novelty 6.0 of 10

    A reinforcement learning pipeline with Lean verifier feedback trains a 32B model that scores 70.0% pass@1 on miniF2F-test, a new state of the art.

  11. Rethinking Verification for LLM Code Generation: From Generation to Testing

    cs.CL 2025-07 reject novelty 6.0 of 10

    SAGA, a human-LLM workflow, generates tests that catch 90.62% of known human bugs on a new benchmark, though those bugs also helped create the tests.

  12. Coding Triangle: How Does Large Language Model Understand Code?

    cs.CL 2025-07 conditional novelty 6.0 of 10

    The Coding Triangle framework evaluates LLMs on editorials, code, and test cases, revealing that models are self-consistent yet lack diversity and that model mixtures improve robustness.

  13. ADRD: LLM-Driven Autonomous Driving Based on Rule-based Decision Systems

    cs.AI 2025-06 conditional novelty 6.0 of 10

    ADRD lets an LLM generate and debug rule-based driving decision trees that, on three highway-env scenarios, outlast a PPO baseline and the DiLu LLM method.

  14. ScaleRTL: Scaling LLMs with Reasoning Data and Test-Time Compute for Accurate RTL Code Generation

    cs.AR 2025-06 conditional novelty 6.0 of 10

    A reasoning LLM fine-tuned on 3.5B tokens of RTL chain-of-thought data, with iterative test-time self-correction, sets new state-of-the-art results on VerilogEval and RTLLM.

  15. HardTests: Synthesizing High-Quality Test Cases for LLM Coding

    cs.CL 2025-05 conditional novelty 6.0 of 10

    HardTestGen generates higher-precision, higher-recall test suites for 47,136 competitive programming problems, improving test precision by 11.3 points and recall by 17.5 points over TACO and CodeContests when judging ...

  16. LlamaRL: A Distributed Asynchronous Reinforcement Learning Framework for Efficient Large-scale LLM Training

    cs.LG 2025-05 conditional novelty 6.0 of 10

    LlamaRL is a distributed asynchronous RL framework that reports up to 10.7x faster RL post-training steps for 405B-parameter LLMs than a co-located synchronous baseline, using off-policy corrections and GPU-direct wei...

  17. Beyond path selection: Better LLMs for Scientific Information Extraction with MimicSFT and Relevance and Rule-induced(R$^2$)GRPO

    cs.CL 2025-05 conditional novelty 6.0 of 10

    MimicSFT plus R2GRPO improves scientific relation extraction in LLMs, beating supervised baselines and showing RLVR can expand reasoning capacity.

  18. Collision- and Reachability-Aware Multi-Robot Control with Grounded LLM Planners

    cs.RO 2025-05 conditional novelty 6.0 of 10

    RLVR fine-tuning teaches small LLMs to reason about reachability and collisions, letting them beat far larger ungrounded LLMs on multi-robot box-moving tasks.

  19. Seeing is Believing, but How Much? A Comprehensive Analysis of Verbalized Calibration in Vision-Language Models

    cs.CV 2025-05 conditional novelty 6.0 of 10

    Current vision-language models are largely miscalibrated when they verbalize confidence, visual reasoning models such as o3 and o4-mini are better calibrated, and Visual Confidence-Aware Prompting reduces ECE on IsoBench.

  20. HomeBench: Evaluating LLMs in Smart Homes with Valid and Invalid Instructions Across Single and Multiple Devices

    cs.CL 2025-05 conditional novelty 6.0 of 10

    HomeBench is a new smart home benchmark that exposes near-zero success rates for top LLMs on invalid multi-device instructions.

  21. Towards Revealing the Effectiveness of Small-Scale Fine-tuning in R1-style Reinforcement Learning

    cs.LG 2025-05 conditional novelty 6.0 of 10

    Re-distilling a model's own RL-trained policy into 1K SFT samples reproduces RL accuracy at a fraction of the compute.

  22. ReCopilot: Reverse Engineering Copilot in Binary Analysis

    cs.CR 2025-05 conditional novelty 6.0 of 10

    ReCopilot is a 7B binary-analysis LLM that reports 13% higher average scores than existing tools on a new six-task benchmark, but the benchmark and artifacts are not released.

  23. Reasoning Before Translation: Enhancing Legal Machine Translation with Structured Reasoning

    cs.CL 2026-07 conditional novelty 5.0 of 10

    On Swiss legal translation, reinforcement learning with a ChrF reward improves small open models more than supervised fine-tuning, but frontier reasoning models still score higher.

  24. Dream-Coder 7B: An Open Diffusion Language Model for Code

    cs.CL 2025-09 conditional novelty 5.0 of 10

    An open 7B diffusion language model for code, adapted from Qwen2.5-Coder, reaches 21.4% pass@1 on LiveCodeBench and shows adaptive, non-left-to-right generation patterns.

  25. Solving Formal Math Problems by Decomposition and Iterative Reflection

    cs.AI 2025-07 conditional novelty 5.0 of 10

    An agent that decomposes Lean 4 goals into subproblems and iteratively repairs proofs achieves a 95.9% pass rate on miniF2F-test using a stock Gemini model.

  26. SafeGenBench: A Benchmark Framework for Security Vulnerability Detection in LLM-Generated Code

    cs.CR 2025-06 conditional novelty 5.0 of 10

    A 558-task benchmark with SAST and LLM judging shows state-of-the-art LLMs generate vulnerability-free code only about 37% of the time in zero-shot settings.

  27. From Struggle (06-2024) to Mastery (02-2025) LLMs Conquer Advanced Algorithm Exams and Pave the Way for Editorial Generation

    cs.CL 2025-06 conditional novelty 5.0 of 10

    The newest LLMs, especially o3-mini, score near the top of an advanced algorithms exam while older models fail, but visual graph problems remain hard.

  28. A Practical Two-Stage Recipe for Mathematical LLMs: Maximizing Accuracy with SFT and Efficiency with Reinforcement Learning

    cs.LG 2025-07 conditional novelty 4.0 of 10

    Ten-epoch SFT then GRPO raises AIME/MATH-500 accuracy while reducing solution length, achieving 8th place on AIMO's private set.

  29. QwenLong-L1: Towards Long-Context Large Reasoning Models with Reinforcement Learning

    cs.CL 2025-05 reject novelty 4.0 of 10

    QwenLong-L1-32B, trained with warm-up SFT plus curriculum-based long-context RL, reaches 70.7 average accuracy on seven DocQA benchmarks, though the evaluation gives it longer context than its proprietary baselines.

Pith tools