{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KVJTAX3TXGLJD4XJGF77JN4BDE","short_pith_number":"pith:KVJTAX3T","schema_version":"1.0","canonical_sha256":"5553305f73b99691f2e9317ff4b781192def69c206935685500885ade42112ea","source":{"kind":"arxiv","id":"2506.07138","version":1},"attestation_state":"computed","paper":{"title":"Learning Compact Vision Tokens for Efficient Large Multimodal Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.MM"],"primary_cat":"cs.CV","authors_text":"Chengchao Shen, Hao Tang","submitted_at":"2025-06-08T13:36:06Z","abstract_excerpt":"Large multimodal models (LMMs) suffer significant computational challenges due to the high cost of Large Language Models (LLMs) and the quadratic complexity of processing long vision token sequences. In this paper, we explore the spatial redundancy among vision tokens and shorten the length of vision token sequences for inference acceleration. Specifically, we propose a Spatial Token Fusion (STF) method to learn compact vision tokens for short vision token sequence, where spatial-adjacent tokens are fused into one. Meanwhile, weight-frozen vision encoder can not well adapt to the demand of ext"},"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":"2506.07138","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-06-08T13:36:06Z","cross_cats_sorted":["cs.AI","cs.CL","cs.MM"],"title_canon_sha256":"a1b1865aff22eef91847d30898061bb2bb6fee76e826395176bc3efbebd8ec2e","abstract_canon_sha256":"cf2dc98cdfc434285629fe07f9062b35e8c74f3f133748d5188f67a9e89071ee"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:18:02.261806Z","signature_b64":"2+KLzCkbBGgahDkVAvh/HpY2NhN+r/SmsJlQZLDqNhpd8xdtPYVL6oIKNQEa6Ba+Z+2zceJKiKATM4VwsAsGAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5553305f73b99691f2e9317ff4b781192def69c206935685500885ade42112ea","last_reissued_at":"2026-07-05T11:18:02.261310Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:18:02.261310Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Compact Vision Tokens for Efficient Large Multimodal Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.MM"],"primary_cat":"cs.CV","authors_text":"Chengchao Shen, Hao Tang","submitted_at":"2025-06-08T13:36:06Z","abstract_excerpt":"Large multimodal models (LMMs) suffer significant computational challenges due to the high cost of Large Language Models (LLMs) and the quadratic complexity of processing long vision token sequences. In this paper, we explore the spatial redundancy among vision tokens and shorten the length of vision token sequences for inference acceleration. Specifically, we propose a Spatial Token Fusion (STF) method to learn compact vision tokens for short vision token sequence, where spatial-adjacent tokens are fused into one. Meanwhile, weight-frozen vision encoder can not well adapt to the demand of ext"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.07138","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/2506.07138/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":"2506.07138","created_at":"2026-07-05T11:18:02.261383+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.07138v1","created_at":"2026-07-05T11:18:02.261383+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.07138","created_at":"2026-07-05T11:18:02.261383+00:00"},{"alias_kind":"pith_short_12","alias_value":"KVJTAX3TXGLJ","created_at":"2026-07-05T11:18:02.261383+00:00"},{"alias_kind":"pith_short_16","alias_value":"KVJTAX3TXGLJD4XJ","created_at":"2026-07-05T11:18:02.261383+00:00"},{"alias_kind":"pith_short_8","alias_value":"KVJTAX3T","created_at":"2026-07-05T11:18:02.261383+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.27383","citing_title":"Decompose, Mix, Adapt: A Unified Framework for Parameter-Efficient Neural Network Recombination and Compression","ref_index":67,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE","json":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE.json","graph_json":"https://pith.science/api/pith-number/KVJTAX3TXGLJD4XJGF77JN4BDE/graph.json","events_json":"https://pith.science/api/pith-number/KVJTAX3TXGLJD4XJGF77JN4BDE/events.json","paper":"https://pith.science/paper/KVJTAX3T"},"agent_actions":{"view_html":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE","download_json":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE.json","view_paper":"https://pith.science/paper/KVJTAX3T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.07138&json=true","fetch_graph":"https://pith.science/api/pith-number/KVJTAX3TXGLJD4XJGF77JN4BDE/graph.json","fetch_events":"https://pith.science/api/pith-number/KVJTAX3TXGLJD4XJGF77JN4BDE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE/action/storage_attestation","attest_author":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE/action/author_attestation","sign_citation":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE/action/citation_signature","submit_replication":"https://pith.science/pith/KVJTAX3TXGLJD4XJGF77JN4BDE/action/replication_record"}},"created_at":"2026-07-05T11:18:02.261383+00:00","updated_at":"2026-07-05T11:18:02.261383+00:00"}