{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NIVC4BKDEVTXFYRQP2QKGUARMI","short_pith_number":"pith:NIVC4BKD","schema_version":"1.0","canonical_sha256":"6a2a2e0543256772e2307ea0a350116232fcb3963d1da5421ee45186fc6f9035","source":{"kind":"arxiv","id":"2606.07577","version":1},"attestation_state":"computed","paper":{"title":"OmniMem: Perturbation-aware Memory Compression for Streaming Audio-Visual LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV","cs.SD","eess.AS"],"primary_cat":"cs.AI","authors_text":"Chao Zhang, Guangzhi Sun, Yixuan Li, Yudong Yang","submitted_at":"2026-05-26T19:12:13Z","abstract_excerpt":"Audio-visual large language models (LLMs) hold strong promise for long-form video understanding, yet their long-video inference is fundamentally limited by the linear growth of video tokens and key-value (KV) caches. We present OmniMem, a memory-efficient streaming framework designed specifically for audio-visual LLMs. Unlike existing compression methods that treat all tokens uniformly, OmniMem introduces a modality-aware memory allocation strategy that separately manages visual and audio contexts, addressing the severe token imbalance between the two modalities. OmniMem further preserves info"},"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":"2606.07577","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-26T19:12:13Z","cross_cats_sorted":["cs.CV","cs.SD","eess.AS"],"title_canon_sha256":"cc55d665b8fa68c2d379f30007ee67a897b876aa0245867ffdb33481d8e1769c","abstract_canon_sha256":"e63a7397adf921b934661652e88878760bca9479d3114ff504e9e94606f34ca9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T00:04:43.566619Z","signature_b64":"IFk8Ordd0EHq+t6PufIpUnl2lDnahMgM1T+nJBCre3qX0LfEGWEZFh4ITlnE/QAYeR09xAUAPa7o/88QiuBpDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6a2a2e0543256772e2307ea0a350116232fcb3963d1da5421ee45186fc6f9035","last_reissued_at":"2026-06-09T00:04:43.565949Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T00:04:43.565949Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"OmniMem: Perturbation-aware Memory Compression for Streaming Audio-Visual LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV","cs.SD","eess.AS"],"primary_cat":"cs.AI","authors_text":"Chao Zhang, Guangzhi Sun, Yixuan Li, Yudong Yang","submitted_at":"2026-05-26T19:12:13Z","abstract_excerpt":"Audio-visual large language models (LLMs) hold strong promise for long-form video understanding, yet their long-video inference is fundamentally limited by the linear growth of video tokens and key-value (KV) caches. We present OmniMem, a memory-efficient streaming framework designed specifically for audio-visual LLMs. Unlike existing compression methods that treat all tokens uniformly, OmniMem introduces a modality-aware memory allocation strategy that separately manages visual and audio contexts, addressing the severe token imbalance between the two modalities. OmniMem further preserves info"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07577","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/2606.07577/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":"2606.07577","created_at":"2026-06-09T00:04:43.566053+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.07577v1","created_at":"2026-06-09T00:04:43.566053+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07577","created_at":"2026-06-09T00:04:43.566053+00:00"},{"alias_kind":"pith_short_12","alias_value":"NIVC4BKDEVTX","created_at":"2026-06-09T00:04:43.566053+00:00"},{"alias_kind":"pith_short_16","alias_value":"NIVC4BKDEVTXFYRQ","created_at":"2026-06-09T00:04:43.566053+00:00"},{"alias_kind":"pith_short_8","alias_value":"NIVC4BKD","created_at":"2026-06-09T00:04:43.566053+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/NIVC4BKDEVTXFYRQP2QKGUARMI","json":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI.json","graph_json":"https://pith.science/api/pith-number/NIVC4BKDEVTXFYRQP2QKGUARMI/graph.json","events_json":"https://pith.science/api/pith-number/NIVC4BKDEVTXFYRQP2QKGUARMI/events.json","paper":"https://pith.science/paper/NIVC4BKD"},"agent_actions":{"view_html":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI","download_json":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI.json","view_paper":"https://pith.science/paper/NIVC4BKD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.07577&json=true","fetch_graph":"https://pith.science/api/pith-number/NIVC4BKDEVTXFYRQP2QKGUARMI/graph.json","fetch_events":"https://pith.science/api/pith-number/NIVC4BKDEVTXFYRQP2QKGUARMI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI/action/storage_attestation","attest_author":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI/action/author_attestation","sign_citation":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI/action/citation_signature","submit_replication":"https://pith.science/pith/NIVC4BKDEVTXFYRQP2QKGUARMI/action/replication_record"}},"created_at":"2026-06-09T00:04:43.566053+00:00","updated_at":"2026-06-09T00:04:43.566053+00:00"}