{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:UZRI3G2SSBUOJ37ZBNT3JMJM3C","short_pith_number":"pith:UZRI3G2S","schema_version":"1.0","canonical_sha256":"a6628d9b529068e4eff90b67b4b12cd8ab0e8b9488633ae227085ebc8d980eac","source":{"kind":"arxiv","id":"2305.15033","version":2},"attestation_state":"computed","paper":{"title":"SmartTrim: Adaptive Tokens and Attention Pruning for Efficient Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bing Qin, Dongliang Xu, Haichao Zhu, Jiafeng Liang, Jingchang Chen, Liping Shan, Ming Liu, Qing Yang, Wangchunshu Zhou, Zekun Wang","submitted_at":"2023-05-24T11:18:00Z","abstract_excerpt":"Despite achieving remarkable performance on various vision-language tasks, Transformer-based Vision-Language Models (VLMs) suffer from redundancy in inputs and parameters, significantly hampering their efficiency in real-world applications. Moreover, the degree of redundancy in token representations and model parameters, such as attention heads, varies significantly for different inputs. In light of the challenges, we propose SmartTrim, an adaptive acceleration framework for VLMs, which adjusts the computational overhead per instance. Specifically, we integrate lightweight modules into the ori"},"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":"2305.15033","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-24T11:18:00Z","cross_cats_sorted":[],"title_canon_sha256":"ef0664b61ca4a5aa423974dbcc1d4b86e5a3eb9dccb02cc4cb3d2fc07ee5796e","abstract_canon_sha256":"01c6d824dced8b99330e93d4b5d13454e933de53c12aef44dd8c07f6cf2197da"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:49:12.005282Z","signature_b64":"dV6jUURoies2Cjo4KiOisuSgCrCtalTFam3GQacpiayOlPzZYVo1WSKNL8ITcR9UjKXseQB8ybGUIfPDbRFLBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a6628d9b529068e4eff90b67b4b12cd8ab0e8b9488633ae227085ebc8d980eac","last_reissued_at":"2026-07-05T07:49:12.004797Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:49:12.004797Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SmartTrim: Adaptive Tokens and Attention Pruning for Efficient Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bing Qin, Dongliang Xu, Haichao Zhu, Jiafeng Liang, Jingchang Chen, Liping Shan, Ming Liu, Qing Yang, Wangchunshu Zhou, Zekun Wang","submitted_at":"2023-05-24T11:18:00Z","abstract_excerpt":"Despite achieving remarkable performance on various vision-language tasks, Transformer-based Vision-Language Models (VLMs) suffer from redundancy in inputs and parameters, significantly hampering their efficiency in real-world applications. Moreover, the degree of redundancy in token representations and model parameters, such as attention heads, varies significantly for different inputs. In light of the challenges, we propose SmartTrim, an adaptive acceleration framework for VLMs, which adjusts the computational overhead per instance. Specifically, we integrate lightweight modules into the ori"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.15033","kind":"arxiv","version":2},"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/2305.15033/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":"2305.15033","created_at":"2026-07-05T07:49:12.004857+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.15033v2","created_at":"2026-07-05T07:49:12.004857+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.15033","created_at":"2026-07-05T07:49:12.004857+00:00"},{"alias_kind":"pith_short_12","alias_value":"UZRI3G2SSBUO","created_at":"2026-07-05T07:49:12.004857+00:00"},{"alias_kind":"pith_short_16","alias_value":"UZRI3G2SSBUOJ37Z","created_at":"2026-07-05T07:49:12.004857+00:00"},{"alias_kind":"pith_short_8","alias_value":"UZRI3G2S","created_at":"2026-07-05T07:49:12.004857+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C","json":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C.json","graph_json":"https://pith.science/api/pith-number/UZRI3G2SSBUOJ37ZBNT3JMJM3C/graph.json","events_json":"https://pith.science/api/pith-number/UZRI3G2SSBUOJ37ZBNT3JMJM3C/events.json","paper":"https://pith.science/paper/UZRI3G2S"},"agent_actions":{"view_html":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C","download_json":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C.json","view_paper":"https://pith.science/paper/UZRI3G2S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.15033&json=true","fetch_graph":"https://pith.science/api/pith-number/UZRI3G2SSBUOJ37ZBNT3JMJM3C/graph.json","fetch_events":"https://pith.science/api/pith-number/UZRI3G2SSBUOJ37ZBNT3JMJM3C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C/action/storage_attestation","attest_author":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C/action/author_attestation","sign_citation":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C/action/citation_signature","submit_replication":"https://pith.science/pith/UZRI3G2SSBUOJ37ZBNT3JMJM3C/action/replication_record"}},"created_at":"2026-07-05T07:49:12.004857+00:00","updated_at":"2026-07-05T07:49:12.004857+00:00"}