{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:PJPBESXD4CCFOXK6W4TL5Y55IO","short_pith_number":"pith:PJPBESXD","schema_version":"1.0","canonical_sha256":"7a5e124ae3e084575d5eb726bee3bd43b984060329b141e2d37333a8027623e1","source":{"kind":"arxiv","id":"2309.03576","version":2},"attestation_state":"computed","paper":{"title":"DropPos: Pre-Training Vision Transformers by Reconstructing Dropped Positions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Haochen Wang, Junsong Fan, Kaiyou Song, Tong Wang, Yuxi Wang, Zhaoxiang Zhang","submitted_at":"2023-09-07T09:12:02Z","abstract_excerpt":"As it is empirically observed that Vision Transformers (ViTs) are quite insensitive to the order of input tokens, the need for an appropriate self-supervised pretext task that enhances the location awareness of ViTs is becoming evident. To address this, we present DropPos, a novel pretext task designed to reconstruct Dropped Positions. The formulation of DropPos is simple: we first drop a large random subset of positional embeddings and then the model classifies the actual position for each non-overlapping patch among all possible positions solely based on their visual appearance. To avoid tri"},"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":"2309.03576","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-09-07T09:12:02Z","cross_cats_sorted":[],"title_canon_sha256":"7dadf0b7ca2ab5b3ad731543896df2f5a667645d2de70a652ec3fa7c5be5af10","abstract_canon_sha256":"358d2232565bce0759fbce28a81f23312674cd4bb5f613a9f5a512b5ff2522f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:53:09.889446Z","signature_b64":"+j1v+aP5II5tHMFcsWPg9YGDRmDh1h8Lwd5oKCLHhDrs4aRJN9uqXuyNtDAHOp+4ulX9DT2ssSoqnIyFVKrFBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7a5e124ae3e084575d5eb726bee3bd43b984060329b141e2d37333a8027623e1","last_reissued_at":"2026-07-05T06:53:09.888974Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:53:09.888974Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DropPos: Pre-Training Vision Transformers by Reconstructing Dropped Positions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Haochen Wang, Junsong Fan, Kaiyou Song, Tong Wang, Yuxi Wang, Zhaoxiang Zhang","submitted_at":"2023-09-07T09:12:02Z","abstract_excerpt":"As it is empirically observed that Vision Transformers (ViTs) are quite insensitive to the order of input tokens, the need for an appropriate self-supervised pretext task that enhances the location awareness of ViTs is becoming evident. To address this, we present DropPos, a novel pretext task designed to reconstruct Dropped Positions. The formulation of DropPos is simple: we first drop a large random subset of positional embeddings and then the model classifies the actual position for each non-overlapping patch among all possible positions solely based on their visual appearance. To avoid tri"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2309.03576","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/2309.03576/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":"2309.03576","created_at":"2026-07-05T06:53:09.889030+00:00"},{"alias_kind":"arxiv_version","alias_value":"2309.03576v2","created_at":"2026-07-05T06:53:09.889030+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2309.03576","created_at":"2026-07-05T06:53:09.889030+00:00"},{"alias_kind":"pith_short_12","alias_value":"PJPBESXD4CCF","created_at":"2026-07-05T06:53:09.889030+00:00"},{"alias_kind":"pith_short_16","alias_value":"PJPBESXD4CCFOXK6","created_at":"2026-07-05T06:53:09.889030+00:00"},{"alias_kind":"pith_short_8","alias_value":"PJPBESXD","created_at":"2026-07-05T06:53:09.889030+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2501.10131","citing_title":"ACE: Anatomically Consistent Embeddings in Composition and Decomposition","ref_index":52,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO","json":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO.json","graph_json":"https://pith.science/api/pith-number/PJPBESXD4CCFOXK6W4TL5Y55IO/graph.json","events_json":"https://pith.science/api/pith-number/PJPBESXD4CCFOXK6W4TL5Y55IO/events.json","paper":"https://pith.science/paper/PJPBESXD"},"agent_actions":{"view_html":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO","download_json":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO.json","view_paper":"https://pith.science/paper/PJPBESXD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2309.03576&json=true","fetch_graph":"https://pith.science/api/pith-number/PJPBESXD4CCFOXK6W4TL5Y55IO/graph.json","fetch_events":"https://pith.science/api/pith-number/PJPBESXD4CCFOXK6W4TL5Y55IO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO/action/storage_attestation","attest_author":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO/action/author_attestation","sign_citation":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO/action/citation_signature","submit_replication":"https://pith.science/pith/PJPBESXD4CCFOXK6W4TL5Y55IO/action/replication_record"}},"created_at":"2026-07-05T06:53:09.889030+00:00","updated_at":"2026-07-05T06:53:09.889030+00:00"}