{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IMJPHJWW7CZS6V6GFBFJBC3HCN","short_pith_number":"pith:IMJPHJWW","schema_version":"1.0","canonical_sha256":"4312f3a6d6f8b32f57c6284a908b67134fa7e6b6bf7fc81659dfc574a0151815","source":{"kind":"arxiv","id":"2506.13536","version":1},"attestation_state":"computed","paper":{"title":"What Matters in Learning from Large-Scale Datasets for Robot Manipulation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.RO","authors_text":"Ajay Mandlekar, Danfei Xu, Kuancheng Wang, Matthew Bronars, Nadun Ranawaka Arachchige, Soroush Nasiriany, Vaibhav Saxena, Woo Chul Shin","submitted_at":"2025-06-16T14:25:29Z","abstract_excerpt":"Imitation learning from large multi-task demonstration datasets has emerged as a promising path for building generally-capable robots. As a result, 1000s of hours have been spent on building such large-scale datasets around the globe. Despite the continuous growth of such efforts, we still lack a systematic understanding of what data should be collected to improve the utility of a robotics dataset and facilitate downstream policy learning. In this work, we conduct a large-scale dataset composition study to answer this question. We develop a data generation framework to procedurally emulate com"},"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":"2506.13536","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.RO","submitted_at":"2025-06-16T14:25:29Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"8beb8d453394a183f3ff8e861bafe5b04130dfabf84bd1402949fb017afa82e1","abstract_canon_sha256":"c83a37017ed4c6191e6a05cb7e4ed6ee97358ee32e9f90f64331124ad3cc70b0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:22:21.822532Z","signature_b64":"LzSVpUHwDOj3CZVL/q59hwYDB7M+IcgM9PW82Lf1EiNvUOv0snW5QhJrCueAhgGxY1yg0rDolXfPkzJZYtTkBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4312f3a6d6f8b32f57c6284a908b67134fa7e6b6bf7fc81659dfc574a0151815","last_reissued_at":"2026-07-05T11:22:21.822050Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:22:21.822050Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Matters in Learning from Large-Scale Datasets for Robot Manipulation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.RO","authors_text":"Ajay Mandlekar, Danfei Xu, Kuancheng Wang, Matthew Bronars, Nadun Ranawaka Arachchige, Soroush Nasiriany, Vaibhav Saxena, Woo Chul Shin","submitted_at":"2025-06-16T14:25:29Z","abstract_excerpt":"Imitation learning from large multi-task demonstration datasets has emerged as a promising path for building generally-capable robots. As a result, 1000s of hours have been spent on building such large-scale datasets around the globe. Despite the continuous growth of such efforts, we still lack a systematic understanding of what data should be collected to improve the utility of a robotics dataset and facilitate downstream policy learning. In this work, we conduct a large-scale dataset composition study to answer this question. We develop a data generation framework to procedurally emulate com"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.13536","kind":"arxiv","version":1},"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/2506.13536/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":"2506.13536","created_at":"2026-07-05T11:22:21.822103+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.13536v1","created_at":"2026-07-05T11:22:21.822103+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.13536","created_at":"2026-07-05T11:22:21.822103+00:00"},{"alias_kind":"pith_short_12","alias_value":"IMJPHJWW7CZS","created_at":"2026-07-05T11:22:21.822103+00:00"},{"alias_kind":"pith_short_16","alias_value":"IMJPHJWW7CZS6V6G","created_at":"2026-07-05T11:22:21.822103+00:00"},{"alias_kind":"pith_short_8","alias_value":"IMJPHJWW","created_at":"2026-07-05T11:22:21.822103+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23625","citing_title":"Learning to See While Learning to Act: Diffusion Models for Active Perception in Robot Imitation","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01072","citing_title":"Expanding Spatial and Temporal Context for Robotic Imitation Learning With Scene Graphs","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27724","citing_title":"HumanoidMimicGen: Data Generation for Loco-Manipulation via Whole-Body Planning","ref_index":47,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN","json":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN.json","graph_json":"https://pith.science/api/pith-number/IMJPHJWW7CZS6V6GFBFJBC3HCN/graph.json","events_json":"https://pith.science/api/pith-number/IMJPHJWW7CZS6V6GFBFJBC3HCN/events.json","paper":"https://pith.science/paper/IMJPHJWW"},"agent_actions":{"view_html":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN","download_json":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN.json","view_paper":"https://pith.science/paper/IMJPHJWW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.13536&json=true","fetch_graph":"https://pith.science/api/pith-number/IMJPHJWW7CZS6V6GFBFJBC3HCN/graph.json","fetch_events":"https://pith.science/api/pith-number/IMJPHJWW7CZS6V6GFBFJBC3HCN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN/action/storage_attestation","attest_author":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN/action/author_attestation","sign_citation":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN/action/citation_signature","submit_replication":"https://pith.science/pith/IMJPHJWW7CZS6V6GFBFJBC3HCN/action/replication_record"}},"created_at":"2026-07-05T11:22:21.822103+00:00","updated_at":"2026-07-05T11:22:21.822103+00:00"}