{"id":"arxiv:2511.07317","title":"RLVE: Scaling Up Reinforcement Learning for Language Models with Adaptive Verifiable Environments","date":"2025-11-10","section":"Post-training","quality":0.372291,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/","source_url":"https://arxiv.org/abs/2511.07317","text_available":true,"authors":["Zhiyuan Zeng","Hamish Ivison","Yiping Wang","Lifan Yuan"],"outgoing":[{"source":"arxiv:2511.07317","target":"acl:2022.emnlp-main.340","evidence":"title","id":"acl:2022.emnlp-main.340","title":"Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/e43a06571b7e8d89abd5f20f8f520c8bc3d3eb32ffb156cb7fc9e0469fc14b0a/"},{"source":"arxiv:2511.07317","target":"arxiv:1901.01753","evidence":"id","id":"arxiv:1901.01753","title":"Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/a9790a1ede488fbbb2d7b4cdb97b884a5a1e21357b72c5e3419a83ea629d5a4b/"},{"source":"arxiv:2511.07317","target":"arxiv:2212.09741","evidence":"title","id":"arxiv:2212.09741","title":"One Embedder, Any Task: Instruction-Finetuned Text Embeddings","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b162d4a59ad6b25fd180e566615cf3dc2cd3a6c505f0421e15fd036236ad2ecd/"},{"source":"arxiv:2511.07317","target":"arxiv:2402.03300","evidence":"id","id":"arxiv:2402.03300","title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/636fb871904bc06f742cfdea069796f9e9665f95bd58dc63260eaff5154ef985/"},{"source":"arxiv:2511.07317","target":"arxiv:2503.14476","evidence":"id","id":"arxiv:2503.14476","title":"DAPO: An Open-Source LLM Reinforcement Learning System at Scale","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/548e934b912ef84691ace63bf1cfd670e7b0257d9d6986dde926f569e7b2f8b7/"},{"source":"arxiv:2511.07317","target":"arxiv:2510.13786","evidence":"id","id":"arxiv:2510.13786","title":"The Art of Scaling Reinforcement Learning Compute for LLMs","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/72580bcba3d9b32b2f3e83d074f626234b26dc18126edd22a5eacef927cf87da/"}],"incoming":[{"source":"arxiv:2512.13961","target":"arxiv:2511.07317","evidence":"title","id":"arxiv:2512.13961","title":"Olmo 3","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8daa11a9729da9879a7e95c15d4eef3bab8579d2885a14102ed09db111221802/"},{"source":"arxiv:2608.19197","target":"arxiv:2511.07317","evidence":"id","id":"arxiv:2608.19197","title":"SPADE: Self-Play in Adaptive Synthetic Executable Environments","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c524376d6034b5d79661d6f1a444591404f4a250bcb0790d802c05e02f8249df/"}],"text_parts":[{"start":0,"end":15401,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html"},{"start":15401,"end":30680,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html"},{"start":30680,"end":45964,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html"},{"start":45964,"end":60527,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-4.html"},{"start":60527,"end":71633,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-5.html"}],"contents":[{"title":"RLVE : Scaling Up Reinforcement Learning for Language Models with Adaptive Verifiable Environments","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"Abstract","offset":102,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"Keywords:","offset":1413,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"1 Introduction","offset":1431,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"2 Methodology","offset":7070,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"2.1 Verifiable Environment","offset":7656,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"2.2 RL with Adaptive Verifiable Environments","offset":9930,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"2.3 RL Algorithm","offset":13219,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"3 RLVE-Gym : A Suite of 400 Environments Created through Environment Engineering","offset":14465,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"3.1 Environments as Pedagogical Tools","offset":14887,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-1.html#text"},{"title":"3.2 Verification via Environment Advantages","offset":16284,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"3.3 Designing Configurable Difficulty","offset":18258,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"4 Analyzing Components of RLVE","offset":19244,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"4.1 Adaptivity for Unstalled and Efficient Learning","offset":20555,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"4.2 Environment Scaling as a Key Driver of Generalizable Reasoning Capabilities","offset":24494,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"5 Scaling Up RL Training with RLVE","offset":26185,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"5.1 Scaling beyond Data Saturation","offset":27097,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"5.2 Compute-Efficient Scaling","offset":28825,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"6 Related Work","offset":30604,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"RL training of LMs on procedurally generated data.","offset":30623,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-2.html#text"},{"title":"Adaptive data for LM training.","offset":31722,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"7 Discussion on Future Work","offset":33098,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Scaling up RL with adaptive non-verifiable environments.","offset":33130,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Model-based automatic environment engineering.","offset":33825,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Acknowledgements","offset":34909,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Impact Statement","offset":36242,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Appendix A Details of RLVE","offset":36463,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Appendix B Details of RLVE-Gym","offset":36775,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"B.1 Representative Sources of Verifiable Environments with Example Environments in RLVE-Gym","offset":36810,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Verifiable environments inspired by programming competition problems.","offset":37160,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Verifiable environments of mathematical operations.","offset":38590,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Verifiable environments of optimization problems.","offset":40325,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Verifiable environments of classical algorithmic problems.","offset":41688,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Verifiable environments of logical puzzles.","offset":42722,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Verifiable environments of NP-complete problems.","offset":43975,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"B.2 Full List of 400 Verifiable Environments from RLVE-Gym","offset":45658,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-3.html#text"},{"title":"Appendix C RL Training Details","offset":56986,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-4.html#text"},{"title":"Appendix D Evaluation Details","offset":59124,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-4.html#text"},{"title":"Appendix E Details of Training Environment Collection","offset":62367,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-5.html#text"},{"title":"Instructions for reporting errors","offset":70669,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/part-5.html#text"}],"text_code_points":71633,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c0c18cb0ae7fe76a7239f38ce819c7ac588c1184dd27c5672f868ff4e2e26605/source.md"}