{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BD4CA7UCRQWZXARRH6YZARMM37","short_pith_number":"pith:BD4CA7UC","schema_version":"1.0","canonical_sha256":"08f8207e828c2d9b82313fb190458cdff6957091d8baefa5f12cbbf8902722bd","source":{"kind":"arxiv","id":"2504.16083","version":2},"attestation_state":"computed","paper":{"title":"MMInference: Accelerating Pre-filling for Long-Context VLMs via Modality-Aware Permutation Sparse Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Amir H. Abdi, Chengruidong Zhang, Dongsheng Li, Huiqiang Jiang, Jianfeng Gao, Lili Qiu, Qianhui Wu, Surin Ahn, Xufang Luo, Yucheng Li, Yuqing Yang","submitted_at":"2025-04-22T17:59:51Z","abstract_excerpt":"The integration of long-context capabilities with visual understanding unlocks unprecedented potential for Vision Language Models (VLMs). However, the quadratic attention complexity during the pre-filling phase remains a significant obstacle to real-world deployment. To overcome this limitation, we introduce MMInference (Multimodality Million tokens Inference), a dynamic sparse attention method that accelerates the prefilling stage for long-context multi-modal inputs. First, our analysis reveals that the temporal and spatial locality of video input leads to a unique sparse pattern, the Grid pa"},"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":"2504.16083","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-04-22T17:59:51Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ab6152512468ae8dca721827e03239e9f8954945907acebc1f5b10ac6245ceba","abstract_canon_sha256":"e6b370823d9cf204b315507c6c08ece1743880723e91b22cbe8c8fbc1a713625"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:08:19.792463Z","signature_b64":"l75wnqp9Ei0pI5v5ODY/snKqNzhN3UMsPb9qtsibSZrSoy8A/MDnIWg2ZOg/58PP2w9d1omHMN75ydDBYxuoBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"08f8207e828c2d9b82313fb190458cdff6957091d8baefa5f12cbbf8902722bd","last_reissued_at":"2026-07-05T11:08:19.791957Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:08:19.791957Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MMInference: Accelerating Pre-filling for Long-Context VLMs via Modality-Aware Permutation Sparse Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Amir H. Abdi, Chengruidong Zhang, Dongsheng Li, Huiqiang Jiang, Jianfeng Gao, Lili Qiu, Qianhui Wu, Surin Ahn, Xufang Luo, Yucheng Li, Yuqing Yang","submitted_at":"2025-04-22T17:59:51Z","abstract_excerpt":"The integration of long-context capabilities with visual understanding unlocks unprecedented potential for Vision Language Models (VLMs). However, the quadratic attention complexity during the pre-filling phase remains a significant obstacle to real-world deployment. To overcome this limitation, we introduce MMInference (Multimodality Million tokens Inference), a dynamic sparse attention method that accelerates the prefilling stage for long-context multi-modal inputs. First, our analysis reveals that the temporal and spatial locality of video input leads to a unique sparse pattern, the Grid pa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.16083","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/2504.16083/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":"2504.16083","created_at":"2026-07-05T11:08:19.792009+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.16083v2","created_at":"2026-07-05T11:08:19.792009+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.16083","created_at":"2026-07-05T11:08:19.792009+00:00"},{"alias_kind":"pith_short_12","alias_value":"BD4CA7UCRQWZ","created_at":"2026-07-05T11:08:19.792009+00:00"},{"alias_kind":"pith_short_16","alias_value":"BD4CA7UCRQWZXARR","created_at":"2026-07-05T11:08:19.792009+00:00"},{"alias_kind":"pith_short_8","alias_value":"BD4CA7UC","created_at":"2026-07-05T11:08:19.792009+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19258","citing_title":"CABLE: Cloud-Assisted Bandwidth-efficient LMM-based Encoding for V2X Systems","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30325","citing_title":"Veda: Scalable Video Diffusion via Distilled Sparse Attention","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2505.15269","citing_title":"LiveVLM: Efficient Online Video Understanding via Streaming-Oriented KV Cache and Retrieval","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2407.11550","citing_title":"Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08077","citing_title":"AdaSpark: Adaptive Sparsity for Efficient Long-Video Understanding","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37","json":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37.json","graph_json":"https://pith.science/api/pith-number/BD4CA7UCRQWZXARRH6YZARMM37/graph.json","events_json":"https://pith.science/api/pith-number/BD4CA7UCRQWZXARRH6YZARMM37/events.json","paper":"https://pith.science/paper/BD4CA7UC"},"agent_actions":{"view_html":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37","download_json":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37.json","view_paper":"https://pith.science/paper/BD4CA7UC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.16083&json=true","fetch_graph":"https://pith.science/api/pith-number/BD4CA7UCRQWZXARRH6YZARMM37/graph.json","fetch_events":"https://pith.science/api/pith-number/BD4CA7UCRQWZXARRH6YZARMM37/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37/action/storage_attestation","attest_author":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37/action/author_attestation","sign_citation":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37/action/citation_signature","submit_replication":"https://pith.science/pith/BD4CA7UCRQWZXARRH6YZARMM37/action/replication_record"}},"created_at":"2026-07-05T11:08:19.792009+00:00","updated_at":"2026-07-05T11:08:19.792009+00:00"}