{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SY6IBORW4XMZK7KHUK5BMV6LXA","short_pith_number":"pith:SY6IBORW","schema_version":"1.0","canonical_sha256":"963c80ba36e5d9957d47a2ba1657cbb80f3e6fab02ac6cb8ab393fe5ce8110bc","source":{"kind":"arxiv","id":"2607.20125","version":1},"attestation_state":"computed","paper":{"title":"HeadCast: Casting Attention Heads for Efficient Autoregressive Video Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Chengru Song, Jinliang Shen, Kang He, Lianghao Su, Yanbing Jiang, Zheming Li, Ziliang Lai","submitted_at":"2026-07-22T13:29:35Z","abstract_excerpt":"Autoregressive (AR) video diffusion models have become a promising paradigm for long and streaming video synthesis, but the continuously growing Key-Value (KV) cache makes attention the dominant inference cost, especially at high resolution where each frame contributes many tokens. Existing remedies either evict the cache with coarse heuristics that cause inter-frame flickering, or require model re-training. We propose HeadCast, a training-free, plug-and-play acceleration framework built on the observation that a pre-trained AR model's attention heads exhibit stable, heterogeneous behaviors. A"},"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":"2607.20125","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-07-22T13:29:35Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ae0235f0da9b77d33cf61fead9c3f289a22b5ea0703b73127457fff97cf142d2","abstract_canon_sha256":"bb39d530de953e08203476d3ddff15a19f04af196475862420eb2d65552f5383"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-23T01:25:02.746273Z","signature_b64":"dDkK8n1YimJR2wruqtLTez4ZTlRnv+2RxZD1mbPLDzWahZCV2X9RFHBrpFljddmS3hUG9uy8tycZnC2ZPVQqDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"963c80ba36e5d9957d47a2ba1657cbb80f3e6fab02ac6cb8ab393fe5ce8110bc","last_reissued_at":"2026-07-23T01:25:02.745381Z","signature_status":"signed_v1","first_computed_at":"2026-07-23T01:25:02.745381Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HeadCast: Casting Attention Heads for Efficient Autoregressive Video Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Chengru Song, Jinliang Shen, Kang He, Lianghao Su, Yanbing Jiang, Zheming Li, Ziliang Lai","submitted_at":"2026-07-22T13:29:35Z","abstract_excerpt":"Autoregressive (AR) video diffusion models have become a promising paradigm for long and streaming video synthesis, but the continuously growing Key-Value (KV) cache makes attention the dominant inference cost, especially at high resolution where each frame contributes many tokens. Existing remedies either evict the cache with coarse heuristics that cause inter-frame flickering, or require model re-training. We propose HeadCast, a training-free, plug-and-play acceleration framework built on the observation that a pre-trained AR model's attention heads exhibit stable, heterogeneous behaviors. A"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.20125","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/2607.20125/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":"2607.20125","created_at":"2026-07-23T01:25:02.745867+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.20125v1","created_at":"2026-07-23T01:25:02.745867+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.20125","created_at":"2026-07-23T01:25:02.745867+00:00"},{"alias_kind":"pith_short_12","alias_value":"SY6IBORW4XMZ","created_at":"2026-07-23T01:25:02.745867+00:00"},{"alias_kind":"pith_short_16","alias_value":"SY6IBORW4XMZK7KH","created_at":"2026-07-23T01:25:02.745867+00:00"},{"alias_kind":"pith_short_8","alias_value":"SY6IBORW","created_at":"2026-07-23T01:25:02.745867+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/SY6IBORW4XMZK7KHUK5BMV6LXA","json":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA.json","graph_json":"https://pith.science/api/pith-number/SY6IBORW4XMZK7KHUK5BMV6LXA/graph.json","events_json":"https://pith.science/api/pith-number/SY6IBORW4XMZK7KHUK5BMV6LXA/events.json","paper":"https://pith.science/paper/SY6IBORW"},"agent_actions":{"view_html":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA","download_json":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA.json","view_paper":"https://pith.science/paper/SY6IBORW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.20125&json=true","fetch_graph":"https://pith.science/api/pith-number/SY6IBORW4XMZK7KHUK5BMV6LXA/graph.json","fetch_events":"https://pith.science/api/pith-number/SY6IBORW4XMZK7KHUK5BMV6LXA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA/action/storage_attestation","attest_author":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA/action/author_attestation","sign_citation":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA/action/citation_signature","submit_replication":"https://pith.science/pith/SY6IBORW4XMZK7KHUK5BMV6LXA/action/replication_record"}},"created_at":"2026-07-23T01:25:02.745867+00:00","updated_at":"2026-07-23T01:25:02.745867+00:00"}