Accepted by KDD2026</p>\n","updatedAt":"2026-06-05T03:10:05.926Z","author":{"_id":"68d65e99d862fb5b2cad9d3a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/PLJmhy2-dfQAMkbcfDW3A.png","fullname":"Zherui Yang","name":"yangzhr","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7791436910629272},"editors":["yangzhr"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/PLJmhy2-dfQAMkbcfDW3A.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.03841","authors":[{"_id":"6a22392a3490a593e87b14aa","name":"Zherui Yang","hidden":false},{"_id":"6a22392a3490a593e87b14ab","name":"Fan Liu","hidden":false},{"_id":"6a22392a3490a593e87b14ac","name":"Yansong Ning","hidden":false},{"_id":"6a22392a3490a593e87b14ad","name":"Hao Liu","hidden":false}],"publishedAt":"2026-06-02T00:00:00.000Z","submittedOnDailyAt":"2026-06-05T00:00:00.000Z","title":"EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management","submittedOnDailyBy":{"_id":"68d65e99d862fb5b2cad9d3a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/PLJmhy2-dfQAMkbcfDW3A.png","isPro":false,"fullname":"Zherui Yang","user":"yangzhr","type":"user","name":"yangzhr"},"summary":"Recent progress in Large Language Model (LLM) agents has enabled promising advances in automated data science. However, existing approaches remain fundamentally limited by their static action sets and lack of principled long-horizon context management, hindering their ability to accumulate reusable experience across tasks and operate reliably in multi-stage, iterative data science pipelines. To address these challenges, we introduce EvoDS, a self-evolving autonomous data science agent that learns to expand its skills and adaptively managing long-term context through agentic reinforcement learning. Specifically, EvoDS introduces two key strategies: (1) Autonomous Skill Acquisition (ASA) mechanism, which enables agents to synthesize, validate, and reuse executable skills; and (2) Adaptive Context Compression (ACC) strategy, which treats context management as a learned control problem rather than passive truncation. These strategies are orchestrated within a two-stage multi-agent training scheme, enabling EvoDS to autonomously improve over time. Theoretically, we prove that EvoDS's hierarchical design reduces tool-selection error, and its optimization objective aligns with an information bottleneck principle, ensuring efficient context use. Empirically, EvoDS outperforms state-of-the-art open-source data science agents by an average of 28.9% across four diverse benchmarks while eliminating out-of-token failures. Our code and data are available at https://github.com/usail-hkust/EvoDS.","upvotes":2,"discussionId":"6a22392a3490a593e87b14ae","projectPage":"https://huggingface.co/yangzhr/EvoDS","githubRepo":"https://github.com/usail-hkust/EvoDS","githubRepoAddedBy":"user","ai_summary":"EvoDS introduces a self-evolving autonomous data science agent that enhances its capabilities through skill acquisition and adaptive context management via reinforcement learning.","ai_keywords":["Large Language Model agents","agentic reinforcement learning","autonomous skill acquisition","adaptive context compression","multi-agent training","information bottleneck principle"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":1,"organization":{"_id":"665abecde9121df9e6e43e33","name":"HKUST-GZ2","fullname":"Hong Kong University of Science and Technology(GuangZhou)","avatar":"https://www.gravatar.com/avatar/df3d0f963dc7e28f1d51efe4f494cd32?d=retro&size=100"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"68d65e99d862fb5b2cad9d3a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/PLJmhy2-dfQAMkbcfDW3A.png","isPro":false,"fullname":"Zherui Yang","user":"yangzhr","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"665abecde9121df9e6e43e33","name":"HKUST-GZ2","fullname":"Hong Kong University of Science and Technology(GuangZhou)","avatar":"https://www.gravatar.com/avatar/df3d0f963dc7e28f1d51efe4f494cd32?d=retro&size=100"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.03841.md"}">
EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management
Abstract
EvoDS introduces a self-evolving autonomous data science agent that enhances its capabilities through skill acquisition and adaptive context management via reinforcement learning.
Recent progress in Large Language Model (LLM) agents has enabled promising advances in automated data science. However, existing approaches remain fundamentally limited by their static action sets and lack of principled long-horizon context management, hindering their ability to accumulate reusable experience across tasks and operate reliably in multi-stage, iterative data science pipelines. To address these challenges, we introduce EvoDS, a self-evolving autonomous data science agent that learns to expand its skills and adaptively managing long-term context through agentic reinforcement learning. Specifically, EvoDS introduces two key strategies: (1) Autonomous Skill Acquisition (ASA) mechanism, which enables agents to synthesize, validate, and reuse executable skills; and (2) Adaptive Context Compression (ACC) strategy, which treats context management as a learned control problem rather than passive truncation. These strategies are orchestrated within a two-stage multi-agent training scheme, enabling EvoDS to autonomously improve over time. Theoretically, we prove that EvoDS's hierarchical design reduces tool-selection error, and its optimization objective aligns with an information bottleneck principle, ensuring efficient context use. Empirically, EvoDS outperforms state-of-the-art open-source data science agents by an average of 28.9% across four diverse benchmarks while eliminating out-of-token failures. Our code and data are available at https://github.com/usail-hkust/EvoDS.
Community
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/2606.03841 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.03841 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.03841 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.