{"id":"arxiv:2502.06807","title":"Competitive Programming with Large Reasoning Models","date":"2025-02-03","section":"Reasoning and the \"physics\" of language models","quality":-0.255672,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/","source_url":"https://arxiv.org/abs/2502.06807","text_available":true,"authors":[" OpenAI"," :","Ahmed El-Kishky","Alexander Wei"],"outgoing":[],"incoming":[{"source":"arxiv:2601.20802","target":"arxiv:2502.06807","evidence":"id","id":"arxiv:2601.20802","title":"Reinforcement Learning via Self-Distillation","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6073ec3d17b9a932a68e79ae078928f4635e40fc79cfcbc31709715d1f1e1ffa/"}],"text_parts":[{"start":0,"end":15412,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html"},{"start":15412,"end":31412,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html"},{"start":31412,"end":31496,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-3.html"}],"contents":[{"title":"Competitive Programming with Large Reasoning Models","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"Abstract","offset":55,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"1 Introduction","offset":1404,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"2 OpenAI o1","offset":4833,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"2.1 CodeForces Benchmark","offset":5914,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"3 OpenAI o1-ioi","offset":7361,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"3.1 Coding RL Fine-tuning","offset":8109,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"3.2 o1-ioi Test-time Strategy","offset":8903,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"Problem formulation","offset":9182,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"Clustering","offset":9722,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"Reranking","offset":10535,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"Submission","offset":11148,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"3.3 CodeForces Benchmark","offset":11881,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"3.4 IOI 2024 Live Competition","offset":12710,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"4 OpenAI o3","offset":13976,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"4.1 CodeForces Benchmark","offset":14668,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-1.html#text"},{"title":"4.2 IOI 2024 Benchmark","offset":16120,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Sampling Approach.","offset":16767,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Results.","offset":17625,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"5 Software Engineering Evaluations","offset":18795,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"5.1 HackerRank Astra","offset":19315,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"5.2 SWE-Bench Verified","offset":20897,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"6 Conclusion","offset":22729,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Appendix A Authorship, credit attribution, and acknowledgments","offset":23397,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Data Preparation:","offset":23464,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"IOI Model Training:","offset":23572,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Sampling Infrastructure:","offset":23637,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Test-time Strategy:","offset":23730,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Evaluation:","offset":23867,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Leadership:","offset":23944,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"o3 Model Development","offset":24019,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Acknowledgments:","offset":24073,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Appendix B Additional CodeForces Details","offset":24965,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"B.1 Data","offset":25250,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"B.2 Grading","offset":25923,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"B.3 Thinking Time","offset":26791,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"B.4 Estimated Rating","offset":28072,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"B.5 Percentile performance","offset":29052,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"B.6 Per Problem Breakdown","offset":29797,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"Appendix C IOI Submissions","offset":29828,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.1 Nile","offset":29971,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.2 Message","offset":30058,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.3 Tree","offset":30212,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.3.1 Submission 1","offset":30301,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.3.2 Submission 2","offset":30428,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.4 Hieroglyphs","offset":30509,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.4.1 Submission 1","offset":30611,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.4.2 Submission 2","offset":30752,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.5 Mosaic","offset":30848,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.5.1 Submission 1","offset":30941,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.5.2 Submission 2","offset":31079,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.6 Sphinx","offset":31184,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.6.1 Submission 1","offset":31279,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-2.html#text"},{"title":"C.6.2 Submission 2","offset":31412,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/part-3.html#text"}],"text_code_points":31496,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d1e6e9343805a99e3ba9032540f71dd4fc3436db1425ff988c9639e0c3641c8c/source.md"}