{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YXPHDCXB2IOUHAEQY5GYGELSLF","short_pith_number":"pith:YXPHDCXB","schema_version":"1.0","canonical_sha256":"c5de718ae1d21d438090c74d831172597943f06d48fffe34710c9abbef75a758","source":{"kind":"arxiv","id":"2502.07837","version":2},"attestation_state":"computed","paper":{"title":"RoboBERT: An End-to-end Multimodal Robotic Manipulation Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.RO","authors_text":"Bin Fang, Jianhua Shan, Sheng Liu, Sicheng Wang, Weiheng Wang","submitted_at":"2025-02-11T02:16:59Z","abstract_excerpt":"Embodied intelligence seamlessly integrates vision, language, and action.~However, most multimodal robotic models rely on massive fine-tuning, incurring high time and hardware costs.~To address this, we introduce RoboBERT, an end-to-end multimodal manipulation model built around a novel two-stage training paradigm.~In the first stage, we freeze most of the vision encoder and train with a single \"standard\" instruction phrasing, allowing the model to focus on stable policy learning via a CNN-based diffusion policy.~In the second stage, we unfreeze all modules and inject diverse natural language "},"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":"2502.07837","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-02-11T02:16:59Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"2147c89d3ab96aee747be36cc769e99267dd55c0426fdcc3dfb26155b8fb729e","abstract_canon_sha256":"e6a242088feb4109a82324fd829eb29e8c9699ec9921468abb51bde308ec3571"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:56:38.187674Z","signature_b64":"XonRCaXc9b6lWdIqu2cqYDdPWhy8gBH5XQ89RfGFh5R5E/E0uJ+DuN/yrs63dT8BGzs6bcrwwtg1c/M5LKZOBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5de718ae1d21d438090c74d831172597943f06d48fffe34710c9abbef75a758","last_reissued_at":"2026-07-05T10:56:38.187030Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:56:38.187030Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RoboBERT: An End-to-end Multimodal Robotic Manipulation Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.RO","authors_text":"Bin Fang, Jianhua Shan, Sheng Liu, Sicheng Wang, Weiheng Wang","submitted_at":"2025-02-11T02:16:59Z","abstract_excerpt":"Embodied intelligence seamlessly integrates vision, language, and action.~However, most multimodal robotic models rely on massive fine-tuning, incurring high time and hardware costs.~To address this, we introduce RoboBERT, an end-to-end multimodal manipulation model built around a novel two-stage training paradigm.~In the first stage, we freeze most of the vision encoder and train with a single \"standard\" instruction phrasing, allowing the model to focus on stable policy learning via a CNN-based diffusion policy.~In the second stage, we unfreeze all modules and inject diverse natural language "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.07837","kind":"arxiv","version":2},"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/2502.07837/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":"2502.07837","created_at":"2026-07-05T10:56:38.187112+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.07837v2","created_at":"2026-07-05T10:56:38.187112+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.07837","created_at":"2026-07-05T10:56:38.187112+00:00"},{"alias_kind":"pith_short_12","alias_value":"YXPHDCXB2IOU","created_at":"2026-07-05T10:56:38.187112+00:00"},{"alias_kind":"pith_short_16","alias_value":"YXPHDCXB2IOUHAEQ","created_at":"2026-07-05T10:56:38.187112+00:00"},{"alias_kind":"pith_short_8","alias_value":"YXPHDCXB","created_at":"2026-07-05T10:56:38.187112+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.21046","citing_title":"CogVLA: Cognition-Aligned Vision-Language-Action Model via Instruction-Driven Routing & Sparsification","ref_index":68,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF","json":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF.json","graph_json":"https://pith.science/api/pith-number/YXPHDCXB2IOUHAEQY5GYGELSLF/graph.json","events_json":"https://pith.science/api/pith-number/YXPHDCXB2IOUHAEQY5GYGELSLF/events.json","paper":"https://pith.science/paper/YXPHDCXB"},"agent_actions":{"view_html":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF","download_json":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF.json","view_paper":"https://pith.science/paper/YXPHDCXB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.07837&json=true","fetch_graph":"https://pith.science/api/pith-number/YXPHDCXB2IOUHAEQY5GYGELSLF/graph.json","fetch_events":"https://pith.science/api/pith-number/YXPHDCXB2IOUHAEQY5GYGELSLF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF/action/storage_attestation","attest_author":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF/action/author_attestation","sign_citation":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF/action/citation_signature","submit_replication":"https://pith.science/pith/YXPHDCXB2IOUHAEQY5GYGELSLF/action/replication_record"}},"created_at":"2026-07-05T10:56:38.187112+00:00","updated_at":"2026-07-05T10:56:38.187112+00:00"}