{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:DSLNLQLD4NWJ63B4G6WLSF7SUV","short_pith_number":"pith:DSLNLQLD","canonical_record":{"source":{"id":"2405.05803","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-05-09T14:38:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e3af14e027746ad8aa9bf0bcecd08333fce10e975bc372a4800ebf5b059d8eb2","abstract_canon_sha256":"69508d0fd17e5078d8ebc77296a82dbcae5795069e06287f679de61930583fc0"},"schema_version":"1.0"},"canonical_sha256":"1c96d5c163e36c9f6c3c37acb917f2a54b2f5bcdcb831cae5de27807630ae170","source":{"kind":"arxiv","id":"2405.05803","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.05803","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"arxiv_version","alias_value":"2405.05803v3","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.05803","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"pith_short_12","alias_value":"DSLNLQLD4NWJ","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"pith_short_16","alias_value":"DSLNLQLD4NWJ63B4","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"pith_short_8","alias_value":"DSLNLQLD","created_at":"2026-07-05T10:05:10Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:DSLNLQLD4NWJ63B4G6WLSF7SUV","target":"record","payload":{"canonical_record":{"source":{"id":"2405.05803","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-05-09T14:38:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e3af14e027746ad8aa9bf0bcecd08333fce10e975bc372a4800ebf5b059d8eb2","abstract_canon_sha256":"69508d0fd17e5078d8ebc77296a82dbcae5795069e06287f679de61930583fc0"},"schema_version":"1.0"},"canonical_sha256":"1c96d5c163e36c9f6c3c37acb917f2a54b2f5bcdcb831cae5de27807630ae170","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:05:10.442608Z","signature_b64":"60RG1IVjemFyXTTVhk5U7xFDaz+OVz0jlmBJ80nq0QrwNaDWq33Gse+hEIxALTuPGWZMNw5GRWqNGhXStXc8BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1c96d5c163e36c9f6c3c37acb917f2a54b2f5bcdcb831cae5de27807630ae170","last_reissued_at":"2026-07-05T10:05:10.442181Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:05:10.442181Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.05803","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:05:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ZGK8tLEMLGRFS2Ap/QVz7vpLog5zyiYUuc+swcgGV1le0xGeZlsg9sV11LJCluq7DQGGtXpoUQipAf4qiYLjAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T18:33:28.380108Z"},"content_sha256":"e3f4a161deeca5d9cc045c6dec5187074d880433d7f800e5d0d986d3b89ba267","schema_version":"1.0","event_id":"sha256:e3f4a161deeca5d9cc045c6dec5187074d880433d7f800e5d0d986d3b89ba267"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:DSLNLQLD4NWJ63B4G6WLSF7SUV","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Boosting Multimodal Large Language Models with Visual Tokens Withdrawal for Rapid Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Luxi Lin, Mingbao Lin, Rongrong Ji, Zhihang Lin","submitted_at":"2024-05-09T14:38:53Z","abstract_excerpt":"Multimodal large language models (MLLMs) demand considerable computations for inference due to the extensive parameters and the additional input tokens needed for visual information representation. Herein, we introduce Visual Tokens Withdrawal (VTW), a plug-and-play module to boost MLLMs for rapid inference. Our approach is inspired by two intriguing phenomena we have observed: (1) the attention sink phenomenon that is prevalent in LLMs also persists in MLLMs, suggesting that initial tokens and nearest tokens receive the majority of attention, while middle vision tokens garner minimal attentio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.05803","kind":"arxiv","version":3},"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/2405.05803/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:05:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"IF/dsXH/zcDIzbY6Stxn6DIVKwaDIQepf5fmbpenjFF2X+g3I7MCM95hEOf2TggzD1q47cDw4ks5MwHP0k8qDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T18:33:28.380620Z"},"content_sha256":"02ab87ef9442703f56bdeb0a429c80a53299d90ff2300f9e13734d864471a9ba","schema_version":"1.0","event_id":"sha256:02ab87ef9442703f56bdeb0a429c80a53299d90ff2300f9e13734d864471a9ba"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DSLNLQLD4NWJ63B4G6WLSF7SUV/bundle.json","state_url":"https://pith.science/pith/DSLNLQLD4NWJ63B4G6WLSF7SUV/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DSLNLQLD4NWJ63B4G6WLSF7SUV/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-07T18:33:28Z","links":{"resolver":"https://pith.science/pith/DSLNLQLD4NWJ63B4G6WLSF7SUV","bundle":"https://pith.science/pith/DSLNLQLD4NWJ63B4G6WLSF7SUV/bundle.json","state":"https://pith.science/pith/DSLNLQLD4NWJ63B4G6WLSF7SUV/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DSLNLQLD4NWJ63B4G6WLSF7SUV/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:DSLNLQLD4NWJ63B4G6WLSF7SUV","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"69508d0fd17e5078d8ebc77296a82dbcae5795069e06287f679de61930583fc0","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-05-09T14:38:53Z","title_canon_sha256":"e3af14e027746ad8aa9bf0bcecd08333fce10e975bc372a4800ebf5b059d8eb2"},"schema_version":"1.0","source":{"id":"2405.05803","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.05803","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"arxiv_version","alias_value":"2405.05803v3","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.05803","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"pith_short_12","alias_value":"DSLNLQLD4NWJ","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"pith_short_16","alias_value":"DSLNLQLD4NWJ63B4","created_at":"2026-07-05T10:05:10Z"},{"alias_kind":"pith_short_8","alias_value":"DSLNLQLD","created_at":"2026-07-05T10:05:10Z"}],"graph_snapshots":[{"event_id":"sha256:02ab87ef9442703f56bdeb0a429c80a53299d90ff2300f9e13734d864471a9ba","target":"graph","created_at":"2026-07-05T10:05:10Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2405.05803/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Multimodal large language models (MLLMs) demand considerable computations for inference due to the extensive parameters and the additional input tokens needed for visual information representation. Herein, we introduce Visual Tokens Withdrawal (VTW), a plug-and-play module to boost MLLMs for rapid inference. Our approach is inspired by two intriguing phenomena we have observed: (1) the attention sink phenomenon that is prevalent in LLMs also persists in MLLMs, suggesting that initial tokens and nearest tokens receive the majority of attention, while middle vision tokens garner minimal attentio","authors_text":"Luxi Lin, Mingbao Lin, Rongrong Ji, Zhihang Lin","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-05-09T14:38:53Z","title":"Boosting Multimodal Large Language Models with Visual Tokens Withdrawal for Rapid Inference"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.05803","kind":"arxiv","version":3},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e3f4a161deeca5d9cc045c6dec5187074d880433d7f800e5d0d986d3b89ba267","target":"record","created_at":"2026-07-05T10:05:10Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"69508d0fd17e5078d8ebc77296a82dbcae5795069e06287f679de61930583fc0","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-05-09T14:38:53Z","title_canon_sha256":"e3af14e027746ad8aa9bf0bcecd08333fce10e975bc372a4800ebf5b059d8eb2"},"schema_version":"1.0","source":{"id":"2405.05803","kind":"arxiv","version":3}},"canonical_sha256":"1c96d5c163e36c9f6c3c37acb917f2a54b2f5bcdcb831cae5de27807630ae170","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1c96d5c163e36c9f6c3c37acb917f2a54b2f5bcdcb831cae5de27807630ae170","first_computed_at":"2026-07-05T10:05:10.442181Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:05:10.442181Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"60RG1IVjemFyXTTVhk5U7xFDaz+OVz0jlmBJ80nq0QrwNaDWq33Gse+hEIxALTuPGWZMNw5GRWqNGhXStXc8BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:05:10.442608Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.05803","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e3f4a161deeca5d9cc045c6dec5187074d880433d7f800e5d0d986d3b89ba267","sha256:02ab87ef9442703f56bdeb0a429c80a53299d90ff2300f9e13734d864471a9ba"],"state_sha256":"31f236f4d6466a75c491ed47746ec717a6595d100e88219eda3eb4bb70aab522"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1rNAU+susXZIHaEevweK9cy5/tfUj5K+rvTmpA6nrc9PEWLKBev5xSVnrOIdETALhDVe9QSYOzI5wmJYW0CUDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T18:33:28.385565Z","bundle_sha256":"e6761efac05e9766dffea486edb6904800bdcb307570305324c259fe86e5173c"}}