{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:NJKVGJG7BSUWMBIOWVEILGMKY4","short_pith_number":"pith:NJKVGJG7","schema_version":"1.0","canonical_sha256":"6a555324df0ca966050eb54885998ac71f1e40c28551465ebf90f4f06ce2b48b","source":{"kind":"arxiv","id":"2505.15816","version":1},"attestation_state":"computed","paper":{"title":"Streamline Without Sacrifice -- Squeeze out Computation Redundancy in LMM","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Lewei Lu, Penghao Wu, Ziwei Liu","submitted_at":"2025-05-21T17:59:52Z","abstract_excerpt":"Large multimodal models excel in multimodal tasks but face significant computational challenges due to excessive computation on visual tokens. Unlike token reduction methods that focus on token-level redundancy, we identify and study the computation-level redundancy on vision tokens to ensure no information loss. Our key insight is that vision tokens from the pretrained vision encoder do not necessarily require all the heavy operations (e.g., self-attention, FFNs) in decoder-only LMMs and could be processed more lightly with proper designs. We designed a series of experiments to discover and p"},"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":"2505.15816","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-21T17:59:52Z","cross_cats_sorted":[],"title_canon_sha256":"84ce63c26f04aa0f95ae090a9e76dcdfaddc29bc9745b9173b6c54cb7ca81609","abstract_canon_sha256":"e4b4cd025fb8194b3ce39d6f0e5a939c9f67a391bc1b91d55c29298f1ae5973f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:58.306183Z","signature_b64":"7oKfblC/pt8njnBhzAhlo9UdOG4lwxkhBXpPuSaxPEgpD7cTUxuTCSOQqinV/uujEMuo4fc2zRop5vcCFindCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6a555324df0ca966050eb54885998ac71f1e40c28551465ebf90f4f06ce2b48b","last_reissued_at":"2026-07-05T11:06:58.305714Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:58.305714Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Streamline Without Sacrifice -- Squeeze out Computation Redundancy in LMM","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Lewei Lu, Penghao Wu, Ziwei Liu","submitted_at":"2025-05-21T17:59:52Z","abstract_excerpt":"Large multimodal models excel in multimodal tasks but face significant computational challenges due to excessive computation on visual tokens. Unlike token reduction methods that focus on token-level redundancy, we identify and study the computation-level redundancy on vision tokens to ensure no information loss. Our key insight is that vision tokens from the pretrained vision encoder do not necessarily require all the heavy operations (e.g., self-attention, FFNs) in decoder-only LMMs and could be processed more lightly with proper designs. We designed a series of experiments to discover and p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.15816","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/2505.15816/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":"2505.15816","created_at":"2026-07-05T11:06:58.305772+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.15816v1","created_at":"2026-07-05T11:06:58.305772+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.15816","created_at":"2026-07-05T11:06:58.305772+00:00"},{"alias_kind":"pith_short_12","alias_value":"NJKVGJG7BSUW","created_at":"2026-07-05T11:06:58.305772+00:00"},{"alias_kind":"pith_short_16","alias_value":"NJKVGJG7BSUWMBIO","created_at":"2026-07-05T11:06:58.305772+00:00"},{"alias_kind":"pith_short_8","alias_value":"NJKVGJG7","created_at":"2026-07-05T11:06:58.305772+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.22138","citing_title":"Efficient Agentic Reasoning Through Self-Regulated Simulative Planning","ref_index":108,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4","json":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4.json","graph_json":"https://pith.science/api/pith-number/NJKVGJG7BSUWMBIOWVEILGMKY4/graph.json","events_json":"https://pith.science/api/pith-number/NJKVGJG7BSUWMBIOWVEILGMKY4/events.json","paper":"https://pith.science/paper/NJKVGJG7"},"agent_actions":{"view_html":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4","download_json":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4.json","view_paper":"https://pith.science/paper/NJKVGJG7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.15816&json=true","fetch_graph":"https://pith.science/api/pith-number/NJKVGJG7BSUWMBIOWVEILGMKY4/graph.json","fetch_events":"https://pith.science/api/pith-number/NJKVGJG7BSUWMBIOWVEILGMKY4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4/action/storage_attestation","attest_author":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4/action/author_attestation","sign_citation":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4/action/citation_signature","submit_replication":"https://pith.science/pith/NJKVGJG7BSUWMBIOWVEILGMKY4/action/replication_record"}},"created_at":"2026-07-05T11:06:58.305772+00:00","updated_at":"2026-07-05T11:06:58.305772+00:00"}