{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6SDPSS7RI7G4LRGDTEV2H4TFVK","short_pith_number":"pith:6SDPSS7R","schema_version":"1.0","canonical_sha256":"f486f94bf147cdc5c4c3992ba3f265aabfd8da2f66612068632b691e0ee4d57c","source":{"kind":"arxiv","id":"2503.23455","version":1},"attestation_state":"computed","paper":{"title":"Efficient Token Compression for Vision Transformer with Spatial Information Preserved","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Jinyang Guo, Junzhu Mao, Xiansheng Hua, Yang Shen, Yazhou Yao","submitted_at":"2025-03-30T14:23:18Z","abstract_excerpt":"Token compression is essential for reducing the computational and memory requirements of transformer models, enabling their deployment in resource-constrained environments. In this work, we propose an efficient and hardware-compatible token compression method called Prune and Merge. Our approach integrates token pruning and merging operations within transformer models to achieve layer-wise token compression. By introducing trainable merge and reconstruct matrices and utilizing shortcut connections, we efficiently merge tokens while preserving important information and enabling the restoration "},"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":"2503.23455","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-03-30T14:23:18Z","cross_cats_sorted":["cs.MM"],"title_canon_sha256":"128bfdcbeda1fb5186fec02600d6dc0a0a7f2e5dadce15deb94487bd4a57bb06","abstract_canon_sha256":"c3c8671e311e72e06f53338e8faadd82fe9c6f90c04f0996c507c90a9c171412"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:41:36.478256Z","signature_b64":"IH0INCGnVpf62bO64kggVF31dn+7l6VgCtim+HzQTAsjeSDwU/nxtKXII3ygosRqFDcw9opZ6cqHRyN4jHOrAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f486f94bf147cdc5c4c3992ba3f265aabfd8da2f66612068632b691e0ee4d57c","last_reissued_at":"2026-07-05T10:41:36.477781Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:41:36.477781Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Efficient Token Compression for Vision Transformer with Spatial Information Preserved","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Jinyang Guo, Junzhu Mao, Xiansheng Hua, Yang Shen, Yazhou Yao","submitted_at":"2025-03-30T14:23:18Z","abstract_excerpt":"Token compression is essential for reducing the computational and memory requirements of transformer models, enabling their deployment in resource-constrained environments. In this work, we propose an efficient and hardware-compatible token compression method called Prune and Merge. Our approach integrates token pruning and merging operations within transformer models to achieve layer-wise token compression. By introducing trainable merge and reconstruct matrices and utilizing shortcut connections, we efficiently merge tokens while preserving important information and enabling the restoration "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.23455","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/2503.23455/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":"2503.23455","created_at":"2026-07-05T10:41:36.477843+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.23455v1","created_at":"2026-07-05T10:41:36.477843+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.23455","created_at":"2026-07-05T10:41:36.477843+00:00"},{"alias_kind":"pith_short_12","alias_value":"6SDPSS7RI7G4","created_at":"2026-07-05T10:41:36.477843+00:00"},{"alias_kind":"pith_short_16","alias_value":"6SDPSS7RI7G4LRGD","created_at":"2026-07-05T10:41:36.477843+00:00"},{"alias_kind":"pith_short_8","alias_value":"6SDPSS7R","created_at":"2026-07-05T10:41:36.477843+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK","json":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK.json","graph_json":"https://pith.science/api/pith-number/6SDPSS7RI7G4LRGDTEV2H4TFVK/graph.json","events_json":"https://pith.science/api/pith-number/6SDPSS7RI7G4LRGDTEV2H4TFVK/events.json","paper":"https://pith.science/paper/6SDPSS7R"},"agent_actions":{"view_html":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK","download_json":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK.json","view_paper":"https://pith.science/paper/6SDPSS7R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.23455&json=true","fetch_graph":"https://pith.science/api/pith-number/6SDPSS7RI7G4LRGDTEV2H4TFVK/graph.json","fetch_events":"https://pith.science/api/pith-number/6SDPSS7RI7G4LRGDTEV2H4TFVK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK/action/storage_attestation","attest_author":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK/action/author_attestation","sign_citation":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK/action/citation_signature","submit_replication":"https://pith.science/pith/6SDPSS7RI7G4LRGDTEV2H4TFVK/action/replication_record"}},"created_at":"2026-07-05T10:41:36.477843+00:00","updated_at":"2026-07-05T10:41:36.477843+00:00"}