{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VXWQTAP2EIUJBZUTKWEK4S4D4U","short_pith_number":"pith:VXWQTAP2","schema_version":"1.0","canonical_sha256":"aded0981fa222890e6935588ae4b83e53503a53836c613fad95c418141ea7edd","source":{"kind":"arxiv","id":"2506.15692","version":3},"attestation_state":"computed","paper":{"title":"MLE-STAR: Machine Learning Engineering Agent via Search and Targeted Refinement","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jaehyun Nam, Jiefeng Chen, Jinsung Yoon, Jinwoo Shin, Sercan \\\"O. Ar{\\i}k, Tomas Pfister","submitted_at":"2025-05-27T18:11:25Z","abstract_excerpt":"Agents based on large language models (LLMs) for machine learning engineering (MLE) can automatically implement ML models via code generation. However, existing approaches to build such agents often rely heavily on inherent LLM knowledge and employ coarse exploration strategies that modify the entire code structure at once. This limits their ability to select effective task-specific models and perform deep exploration within specific components, such as experimenting extensively with feature engineering options. To overcome these, we propose MLE-STAR, a novel approach to build MLE agents. MLE-"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2506.15692","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-27T18:11:25Z","cross_cats_sorted":[],"title_canon_sha256":"220b9a35f5c498ea562312421d6ec4e645b6e9566ad38f26c24df3a6a9b65c5c","abstract_canon_sha256":"693a8dab68fb533c1064aac24a9eea9ca648d941cce402863b68177245e9a3ce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:00:28.279428Z","signature_b64":"NM4nPTR11EOwFUAfepvpv1kaZ4sYw/5DkeuydaWqOnpu88qGU1dCwBoO0bxYsxgB37TmASdbqjhr0qV/P7rdCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aded0981fa222890e6935588ae4b83e53503a53836c613fad95c418141ea7edd","last_reissued_at":"2026-07-05T12:00:28.278885Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:00:28.278885Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MLE-STAR: Machine Learning Engineering Agent via Search and Targeted Refinement","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jaehyun Nam, Jiefeng Chen, Jinsung Yoon, Jinwoo Shin, Sercan \\\"O. Ar{\\i}k, Tomas Pfister","submitted_at":"2025-05-27T18:11:25Z","abstract_excerpt":"Agents based on large language models (LLMs) for machine learning engineering (MLE) can automatically implement ML models via code generation. However, existing approaches to build such agents often rely heavily on inherent LLM knowledge and employ coarse exploration strategies that modify the entire code structure at once. This limits their ability to select effective task-specific models and perform deep exploration within specific components, such as experimenting extensively with feature engineering options. To overcome these, we propose MLE-STAR, a novel approach to build MLE agents. MLE-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.15692","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2506.15692/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2506.15692","created_at":"2026-07-05T12:00:28.278945+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.15692v3","created_at":"2026-07-05T12:00:28.278945+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.15692","created_at":"2026-07-05T12:00:28.278945+00:00"},{"alias_kind":"pith_short_12","alias_value":"VXWQTAP2EIUJ","created_at":"2026-07-05T12:00:28.278945+00:00"},{"alias_kind":"pith_short_16","alias_value":"VXWQTAP2EIUJBZUT","created_at":"2026-07-05T12:00:28.278945+00:00"},{"alias_kind":"pith_short_8","alias_value":"VXWQTAP2","created_at":"2026-07-05T12:00:28.278945+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":18,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2604.17406","citing_title":"EvoMaster: A Foundational Evolving Agent Framework for Agentic Science at Scale","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2606.29151","citing_title":"CADENZA: Compiling Natural-Language Intent into Task-Specific Operator DAGs for Semantic Query Processing","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06473","citing_title":"MLEvolve: A Self-Evolving Framework for Automated Machine Learning Algorithm Discovery","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05250","citing_title":"Towards Persistent Case-Based Memory for Autonomous Data Science: A CBR-Augmented R&D-Agent with a Locally Deployable Small Language Model","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31478","citing_title":"One Reflection Is Not Enough: Self-Correcting Autonomous Research via Multi-Hypothesis Failure Attribution","ref_index":105,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29151","citing_title":"CADENZA: Compiling Natural-Language Intent into Task-Specific Operator DAGs for Semantic Query Processing","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30434","citing_title":"LongDS-Bench: On the Failure of Long-Horizon Agentic Data Analysis","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2509.23986","citing_title":"TusoAI: Agentic Optimization for Scientific Methods","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2508.10177","citing_title":"KompeteAI: Accelerated Autonomous Multi-Agent System for End-to-End Pipeline Generation for Machine Learning Problems","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2601.05930","citing_title":"Can We Predict Before Executing Machine Learning Agents?","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2602.07906","citing_title":"AceGRPO: Adaptive Curriculum Enhanced Group Relative Policy Optimization for Autonomous Machine Learning Engineering","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2603.01692","citing_title":"Reasoning as Gradient: Scaling MLE Agents Beyond Tree Search","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13874","citing_title":"GEAR: Genetic AutoResearch for Agentic Code Evolution","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10906","citing_title":"DataMaster: Data-Centric Autonomous AI Research","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10906","citing_title":"DataMaster: Data-Centric Autonomous AI Research","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09791","citing_title":"Pioneer Agent: Continual Improvement of Small Language Models in Production","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04872","citing_title":"Synthetic Sandbox for Training Machine Learning Engineering Agents","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17406","citing_title":"EvoMaster: A Foundational Evolving Agent Framework for Agentic Science at Scale","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U","json":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U.json","graph_json":"https://pith.science/api/pith-number/VXWQTAP2EIUJBZUTKWEK4S4D4U/graph.json","events_json":"https://pith.science/api/pith-number/VXWQTAP2EIUJBZUTKWEK4S4D4U/events.json","paper":"https://pith.science/paper/VXWQTAP2"},"agent_actions":{"view_html":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U","download_json":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U.json","view_paper":"https://pith.science/paper/VXWQTAP2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.15692&json=true","fetch_graph":"https://pith.science/api/pith-number/VXWQTAP2EIUJBZUTKWEK4S4D4U/graph.json","fetch_events":"https://pith.science/api/pith-number/VXWQTAP2EIUJBZUTKWEK4S4D4U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U/action/storage_attestation","attest_author":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U/action/author_attestation","sign_citation":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U/action/citation_signature","submit_replication":"https://pith.science/pith/VXWQTAP2EIUJBZUTKWEK4S4D4U/action/replication_record"}},"created_at":"2026-07-05T12:00:28.278945+00:00","updated_at":"2026-07-05T12:00:28.278945+00:00"}