{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:WDLJVET2QWSGBXY5AYDRJA4IS2","short_pith_number":"pith:WDLJVET2","schema_version":"1.0","canonical_sha256":"b0d69a927a85a460df1d060714838896bad1d00aa9b83a7a1fb52585cd6c7492","source":{"kind":"arxiv","id":"2607.25669","version":1},"attestation_state":"computed","paper":{"title":"OmniDelta: Skill-Driven Budget Allocation for Token Compression in OmniLLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Baolin Zhang, Haoyang Huang, Hongyaoxing Gu, Jun Dai, Jun Zhang, Kang Tan, Lei Chen, Meng Zhang, Mingchen Wang, Shuang Ge, Tianqi Xu, Tianyu Liu, Wenjie Huang, Xinyi Hu, Yikai Fu, Yue Li, Yuhao Shen","submitted_at":"2026-07-28T12:50:31Z","abstract_excerpt":"Emerging Omni-modal Large Language Models (OmniLLMs) enable unified understanding of text, audio, and video, but their long audio-video token sequences introduce substantial memory and inference costs. Existing compression methods mainly focus on selecting important tokens under fixed budgets, leaving the preceding budget-allocation problem underexplored. We show that direct query-to-audio/video similarity is unreliable for inter-modal budget allocation, and that uniform intra-modal budgets can miss key evidence while retaining redundant content. To address these limitations, we propose OmniDe"},"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":"2607.25669","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-07-28T12:50:31Z","cross_cats_sorted":[],"title_canon_sha256":"174614fd0e93c7a7600e77af9edde8a93c77bb767e1874a74715ec467496180d","abstract_canon_sha256":"e82bbd4b6d577ac8bccb7d4946bc3566f07db05f4cd0b9dc4534f265b41b28b0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-29T01:25:55.839915Z","signature_b64":"LtdFZvIN8qWGO5G+GiWKTyXGYY//gbKu4SiYKQetQ1IUL13Z9TFIGNvzsBQGbaS/J2wJxWNGixBho7e4ErVfDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b0d69a927a85a460df1d060714838896bad1d00aa9b83a7a1fb52585cd6c7492","last_reissued_at":"2026-07-29T01:25:55.839028Z","signature_status":"signed_v1","first_computed_at":"2026-07-29T01:25:55.839028Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"OmniDelta: Skill-Driven Budget Allocation for Token Compression in OmniLLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Baolin Zhang, Haoyang Huang, Hongyaoxing Gu, Jun Dai, Jun Zhang, Kang Tan, Lei Chen, Meng Zhang, Mingchen Wang, Shuang Ge, Tianqi Xu, Tianyu Liu, Wenjie Huang, Xinyi Hu, Yikai Fu, Yue Li, Yuhao Shen","submitted_at":"2026-07-28T12:50:31Z","abstract_excerpt":"Emerging Omni-modal Large Language Models (OmniLLMs) enable unified understanding of text, audio, and video, but their long audio-video token sequences introduce substantial memory and inference costs. Existing compression methods mainly focus on selecting important tokens under fixed budgets, leaving the preceding budget-allocation problem underexplored. We show that direct query-to-audio/video similarity is unreliable for inter-modal budget allocation, and that uniform intra-modal budgets can miss key evidence while retaining redundant content. To address these limitations, we propose OmniDe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.25669","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/2607.25669/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":"2607.25669","created_at":"2026-07-29T01:25:55.839471+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.25669v1","created_at":"2026-07-29T01:25:55.839471+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.25669","created_at":"2026-07-29T01:25:55.839471+00:00"},{"alias_kind":"pith_short_12","alias_value":"WDLJVET2QWSG","created_at":"2026-07-29T01:25:55.839471+00:00"},{"alias_kind":"pith_short_16","alias_value":"WDLJVET2QWSGBXY5","created_at":"2026-07-29T01:25:55.839471+00:00"},{"alias_kind":"pith_short_8","alias_value":"WDLJVET2","created_at":"2026-07-29T01:25:55.839471+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/WDLJVET2QWSGBXY5AYDRJA4IS2","json":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2.json","graph_json":"https://pith.science/api/pith-number/WDLJVET2QWSGBXY5AYDRJA4IS2/graph.json","events_json":"https://pith.science/api/pith-number/WDLJVET2QWSGBXY5AYDRJA4IS2/events.json","paper":"https://pith.science/paper/WDLJVET2"},"agent_actions":{"view_html":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2","download_json":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2.json","view_paper":"https://pith.science/paper/WDLJVET2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.25669&json=true","fetch_graph":"https://pith.science/api/pith-number/WDLJVET2QWSGBXY5AYDRJA4IS2/graph.json","fetch_events":"https://pith.science/api/pith-number/WDLJVET2QWSGBXY5AYDRJA4IS2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2/action/storage_attestation","attest_author":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2/action/author_attestation","sign_citation":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2/action/citation_signature","submit_replication":"https://pith.science/pith/WDLJVET2QWSGBXY5AYDRJA4IS2/action/replication_record"}},"created_at":"2026-07-29T01:25:55.839471+00:00","updated_at":"2026-07-29T01:25:55.839471+00:00"}