{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BYZ3D2CZTOMPV2ZRAURLUXDAUX","short_pith_number":"pith:BYZ3D2CZ","schema_version":"1.0","canonical_sha256":"0e33b1e8599b98faeb310522ba5c60a5cb0e3ef0f634ce25e51a8907b6cb20fe","source":{"kind":"arxiv","id":"2411.15453","version":1},"attestation_state":"computed","paper":{"title":"Enhancing Instruction-Following Capability of Visual-Language Models by Reducing Image Redundancy","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Bo Wang, Jian Jia, Kun Gai, Peng Jiang, Quan Chen, Shengyuan Liu, Te Yang, Weisong Zhao, Xiangyu Zhu, Yanhua Cheng, Yan Li, Zhen Lei","submitted_at":"2024-11-23T05:03:32Z","abstract_excerpt":"Large Language Models (LLMs) have strong instruction-following capability to interpret and execute tasks as directed by human commands. Multimodal Large Language Models (MLLMs) have inferior instruction-following ability compared to LLMs. However, there is a significant gap in the instruction-following capabilities between the MLLMs and LLMs. In this study, we conduct a pilot experiment, which demonstrates that spatially down-sampling visual tokens significantly enhances the instruction-following capability of MLLMs. This is attributed to the substantial redundancy in visual modality. However,"},"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":"2411.15453","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2024-11-23T05:03:32Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"29a7fb8ed76a6b966ddc85bd26e6776247d2cf475af33c008baf12cb073170fc","abstract_canon_sha256":"073ae11094c03b6ee9e60255923613c44da634904d30c087643d3d11955eacd0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:39:35.111642Z","signature_b64":"w8prK7eSTJqzqMyEemaM8hQ+YsnT9uaIssAJE+JfUJw08mRdnK/vcfvNXP2SkkFiRl8FQSUrWgzCnXQLaOhwBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0e33b1e8599b98faeb310522ba5c60a5cb0e3ef0f634ce25e51a8907b6cb20fe","last_reissued_at":"2026-07-05T09:39:35.111156Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:39:35.111156Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Enhancing Instruction-Following Capability of Visual-Language Models by Reducing Image Redundancy","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Bo Wang, Jian Jia, Kun Gai, Peng Jiang, Quan Chen, Shengyuan Liu, Te Yang, Weisong Zhao, Xiangyu Zhu, Yanhua Cheng, Yan Li, Zhen Lei","submitted_at":"2024-11-23T05:03:32Z","abstract_excerpt":"Large Language Models (LLMs) have strong instruction-following capability to interpret and execute tasks as directed by human commands. Multimodal Large Language Models (MLLMs) have inferior instruction-following ability compared to LLMs. However, there is a significant gap in the instruction-following capabilities between the MLLMs and LLMs. In this study, we conduct a pilot experiment, which demonstrates that spatially down-sampling visual tokens significantly enhances the instruction-following capability of MLLMs. This is attributed to the substantial redundancy in visual modality. However,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.15453","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/2411.15453/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":"2411.15453","created_at":"2026-07-05T09:39:35.111219+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.15453v1","created_at":"2026-07-05T09:39:35.111219+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.15453","created_at":"2026-07-05T09:39:35.111219+00:00"},{"alias_kind":"pith_short_12","alias_value":"BYZ3D2CZTOMP","created_at":"2026-07-05T09:39:35.111219+00:00"},{"alias_kind":"pith_short_16","alias_value":"BYZ3D2CZTOMPV2ZR","created_at":"2026-07-05T09:39:35.111219+00:00"},{"alias_kind":"pith_short_8","alias_value":"BYZ3D2CZ","created_at":"2026-07-05T09:39:35.111219+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.15694","citing_title":"GoVector: An I/O-Efficient Caching Strategy for High-Dimensional Vector Nearest Neighbor Search","ref_index":16,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX","json":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX.json","graph_json":"https://pith.science/api/pith-number/BYZ3D2CZTOMPV2ZRAURLUXDAUX/graph.json","events_json":"https://pith.science/api/pith-number/BYZ3D2CZTOMPV2ZRAURLUXDAUX/events.json","paper":"https://pith.science/paper/BYZ3D2CZ"},"agent_actions":{"view_html":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX","download_json":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX.json","view_paper":"https://pith.science/paper/BYZ3D2CZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.15453&json=true","fetch_graph":"https://pith.science/api/pith-number/BYZ3D2CZTOMPV2ZRAURLUXDAUX/graph.json","fetch_events":"https://pith.science/api/pith-number/BYZ3D2CZTOMPV2ZRAURLUXDAUX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX/action/storage_attestation","attest_author":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX/action/author_attestation","sign_citation":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX/action/citation_signature","submit_replication":"https://pith.science/pith/BYZ3D2CZTOMPV2ZRAURLUXDAUX/action/replication_record"}},"created_at":"2026-07-05T09:39:35.111219+00:00","updated_at":"2026-07-05T09:39:35.111219+00:00"}