{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:PTDDNMDWYZ5JFATUGIQXFX4LNK","short_pith_number":"pith:PTDDNMDW","schema_version":"1.0","canonical_sha256":"7cc636b076c67a928274322172df8b6abbc2e7a55da687bd73c3c7eba6c7a156","source":{"kind":"arxiv","id":"2507.02279","version":1},"attestation_state":"computed","paper":{"title":"LaCo: Efficient Layer-wise Compression of Visual Tokens for Multimodal Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fandong Meng, Jie Zhou, Juntao Liu, Liqiang Niu, Wenchao Chen","submitted_at":"2025-07-03T03:42:54Z","abstract_excerpt":"Existing visual token compression methods for Multimodal Large Language Models (MLLMs) predominantly operate as post-encoder modules, limiting their potential for efficiency gains. To address this limitation, we propose LaCo (Layer-wise Visual Token Compression), a novel framework that enables effective token compression within the intermediate layers of the vision encoder. LaCo introduces two core components: 1) a layer-wise pixel-shuffle mechanism that systematically merges adjacent tokens through space-to-channel transformations, and 2) a residual learning architecture with non-parametric s"},"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":"2507.02279","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-07-03T03:42:54Z","cross_cats_sorted":[],"title_canon_sha256":"af9a12d00d5bdb8629bd0b4cd1575ff442876ef3fc07d4fc411de90b61f07292","abstract_canon_sha256":"3a270eef3474896c55fd3b987d456010c69a27c8886846c4a3a410fff6ddb0fc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:31:15.080306Z","signature_b64":"QsTRQMj627wKJOmqiXtDNPgk9QKqXnWCpmKE3qU0uWTIJscjdZL8xrgnfZIDPumLytN8UJpuKbSBMC6MXX+lDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7cc636b076c67a928274322172df8b6abbc2e7a55da687bd73c3c7eba6c7a156","last_reissued_at":"2026-07-05T11:31:15.079815Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:31:15.079815Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LaCo: Efficient Layer-wise Compression of Visual Tokens for Multimodal Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fandong Meng, Jie Zhou, Juntao Liu, Liqiang Niu, Wenchao Chen","submitted_at":"2025-07-03T03:42:54Z","abstract_excerpt":"Existing visual token compression methods for Multimodal Large Language Models (MLLMs) predominantly operate as post-encoder modules, limiting their potential for efficiency gains. To address this limitation, we propose LaCo (Layer-wise Visual Token Compression), a novel framework that enables effective token compression within the intermediate layers of the vision encoder. LaCo introduces two core components: 1) a layer-wise pixel-shuffle mechanism that systematically merges adjacent tokens through space-to-channel transformations, and 2) a residual learning architecture with non-parametric s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.02279","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/2507.02279/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":"2507.02279","created_at":"2026-07-05T11:31:15.079875+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.02279v1","created_at":"2026-07-05T11:31:15.079875+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.02279","created_at":"2026-07-05T11:31:15.079875+00:00"},{"alias_kind":"pith_short_12","alias_value":"PTDDNMDWYZ5J","created_at":"2026-07-05T11:31:15.079875+00:00"},{"alias_kind":"pith_short_16","alias_value":"PTDDNMDWYZ5JFATU","created_at":"2026-07-05T11:31:15.079875+00:00"},{"alias_kind":"pith_short_8","alias_value":"PTDDNMDW","created_at":"2026-07-05T11:31:15.079875+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.25343","citing_title":"Toward Native Multimodal Modeling: A Roadmap","ref_index":207,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK","json":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK.json","graph_json":"https://pith.science/api/pith-number/PTDDNMDWYZ5JFATUGIQXFX4LNK/graph.json","events_json":"https://pith.science/api/pith-number/PTDDNMDWYZ5JFATUGIQXFX4LNK/events.json","paper":"https://pith.science/paper/PTDDNMDW"},"agent_actions":{"view_html":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK","download_json":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK.json","view_paper":"https://pith.science/paper/PTDDNMDW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.02279&json=true","fetch_graph":"https://pith.science/api/pith-number/PTDDNMDWYZ5JFATUGIQXFX4LNK/graph.json","fetch_events":"https://pith.science/api/pith-number/PTDDNMDWYZ5JFATUGIQXFX4LNK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK/action/storage_attestation","attest_author":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK/action/author_attestation","sign_citation":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK/action/citation_signature","submit_replication":"https://pith.science/pith/PTDDNMDWYZ5JFATUGIQXFX4LNK/action/replication_record"}},"created_at":"2026-07-05T11:31:15.079875+00:00","updated_at":"2026-07-05T11:31:15.079875+00:00"}