We propose Atomic Decomposition and Recombination (ADR), a novel framework that generates verifiable code tasks via decomposition into atomic elements and controlled recombination, thereby enabling the generation of genuinely novel and challenging verifiable code tasks.</p>\n","updatedAt":"2026-06-05T03:05:11.688Z","author":{"_id":"5fc052241160c47d1d438556","avatarUrl":"/avatars/f508707fc92f1b42f7897b12b727754c.svg","fullname":"Boxi Cao","name":"Bowieee","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":5,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8383479118347168},"editors":["Bowieee"],"editorAvatarUrls":["/avatars/f508707fc92f1b42f7897b12b727754c.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.31058","authors":[{"_id":"6a223cb63490a593e87b14cf","name":"Jiasheng Zheng","hidden":false},{"_id":"6a223cb63490a593e87b14d0","name":"Boxi Cao","hidden":false},{"_id":"6a223cb63490a593e87b14d1","name":"Boxi Yu","hidden":false},{"_id":"6a223cb63490a593e87b14d2","name":"Yuzhong Zhang","hidden":false},{"_id":"6a223cb63490a593e87b14d3","name":"Jialun Cao","hidden":false},{"_id":"6a223cb63490a593e87b14d4","name":"Yaojie Lu","hidden":false},{"_id":"6a223cb63490a593e87b14d5","name":"Hongyu Lin","hidden":false},{"_id":"6a223cb63490a593e87b14d6","name":"Xianpei Han","hidden":false},{"_id":"6a223cb63490a593e87b14d7","name":"Le Sun","hidden":false}],"publishedAt":"2026-05-29T00:00:00.000Z","submittedOnDailyAt":"2026-06-05T00:00:00.000Z","title":"Combinatorial Synthesis: Scaling Code RLVR via Atomic Decomposition and Recombination","submittedOnDailyBy":{"_id":"5fc052241160c47d1d438556","avatarUrl":"/avatars/f508707fc92f1b42f7897b12b727754c.svg","isPro":false,"fullname":"Boxi Cao","user":"Bowieee","type":"user","name":"Bowieee"},"summary":"Reinforcement Learning with Verifiable Rewards (RLVR) has recently emerged as the cornerstone for shaping the remarkable coding abilities of Large Language Models (LLMs). However, the scalability of RLVR is severely constrained by the scarcity of sufficiently challenging verifiable code tasks that target near the model's edge of competence. Prior studies often rely on heuristic seed expansions for data synthesis, which severely limits both novelty and difficulty. Consequently, the training value of such data fails to scale proportionally with the size of its synthesis. To this end, we propose Atomic Decomposition and Recombination (ADR), a novel framework that generates verifiable code tasks via decomposition into atomic elements and controlled recombination, thereby enabling the generation of genuinely novel and challenging verifiable code tasks. Experiments and analysis demonstrate that ADR achieves superior originality, difficulty, diversity, and test quality over existing baselines, and consistently delivers greater improvements in code ability across RLVR in diverse downstream domains, including algorithmic programming, tool usage, and data science. Our work sheds light on a new paradigm for novel code task synthesis and scalable RLVR training.","upvotes":1,"discussionId":"6a223cb63490a593e87b14d8","githubRepo":"https://github.com/icip-cas/ADR","githubRepoAddedBy":"user","ai_summary":"Atomic Decomposition and Recombination (ADR) framework generates novel and challenging verifiable code tasks for scalable reinforcement learning with verifiable rewards in large language models.","ai_keywords":["Reinforcement Learning with Verifiable Rewards","Large Language Models","code task synthesis","atomic decomposition","recombination","verifiable code tasks","scalability","heuristic seed expansion","algorithmic programming","tool usage","data science"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":3,"organization":{"_id":"66b2d691e2d6c59a4067505c","name":"ICIP","fullname":"ICIP","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6216496a9b34d2fb49144599/8xuyYJM4haNE67IzonZN9.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"5fc052241160c47d1d438556","avatarUrl":"/avatars/f508707fc92f1b42f7897b12b727754c.svg","isPro":false,"fullname":"Boxi Cao","user":"Bowieee","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"66b2d691e2d6c59a4067505c","name":"ICIP","fullname":"ICIP","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6216496a9b34d2fb49144599/8xuyYJM4haNE67IzonZN9.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.31058.md"}">
Combinatorial Synthesis: Scaling Code RLVR via Atomic Decomposition and Recombination
Abstract
Atomic Decomposition and Recombination (ADR) framework generates novel and challenging verifiable code tasks for scalable reinforcement learning with verifiable rewards in large language models.
Reinforcement Learning with Verifiable Rewards (RLVR) has recently emerged as the cornerstone for shaping the remarkable coding abilities of Large Language Models (LLMs). However, the scalability of RLVR is severely constrained by the scarcity of sufficiently challenging verifiable code tasks that target near the model's edge of competence. Prior studies often rely on heuristic seed expansions for data synthesis, which severely limits both novelty and difficulty. Consequently, the training value of such data fails to scale proportionally with the size of its synthesis. To this end, we propose Atomic Decomposition and Recombination (ADR), a novel framework that generates verifiable code tasks via decomposition into atomic elements and controlled recombination, thereby enabling the generation of genuinely novel and challenging verifiable code tasks. Experiments and analysis demonstrate that ADR achieves superior originality, difficulty, diversity, and test quality over existing baselines, and consistently delivers greater improvements in code ability across RLVR in diverse downstream domains, including algorithmic programming, tool usage, and data science. Our work sheds light on a new paradigm for novel code task synthesis and scalable RLVR training.
Community
We propose Atomic Decomposition and Recombination (ADR), a novel framework that generates verifiable code tasks via decomposition into atomic elements and controlled recombination, thereby enabling the generation of genuinely novel and challenging verifiable code tasks.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2605.31058 in a model README.md to link it from this page.
Cite arxiv.org/abs/2605.31058 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2605.31058 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.