{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XUCYXJON7OCYLGAOMFKVDEZQOC","short_pith_number":"pith:XUCYXJON","schema_version":"1.0","canonical_sha256":"bd058ba5cdfb8585980e615551933070b597a69c3aa1ec763a900cc44ccb26ea","source":{"kind":"arxiv","id":"2606.08302","version":1},"attestation_state":"computed","paper":{"title":"HACK++: Towards More Effective Head-Aware Key-Value Compression for Efficient Visual Autoregressive Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hang Guo, Mingbao Lin, Weiyao Lin, Wen Fei, Youru Lv, Yuchen Jiang, Ziran Qin","submitted_at":"2026-06-06T18:58:26Z","abstract_excerpt":"Visual Autoregressive (VAR) models adopt a next-scale prediction paradigm, offering high-quality generation with substantially fewer decoding steps. However, existing VAR models suffer from significant attention complexity and severe memory overhead due to the accumulation of key-value (KV) caches across scales. In this paper, we tackle this challenge by introducing KV cache compression into the next-scale paradigm. We begin with an in-depth analysis of VAR attention and observe that attention heads can be stably divided into two functionally distinct categories: Contextual Heads focus on main"},"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":"2606.08302","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-06-06T18:58:26Z","cross_cats_sorted":[],"title_canon_sha256":"a3591a4b3ce694bdea61cf78229fe315be2008205775dd269c44ebc0ce5b32c5","abstract_canon_sha256":"dca943ac00fdc9e5116d20675300594926569dc9456ae76b602eb9da4725e636"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:33.018268Z","signature_b64":"QQv7LQy6PGsAbl2VvLqF6PS1fE1pTTbeFpMhf2adRHNxF+EBPQ8tvymw8Us6zZVWyIjyAHuBbH4rp7J7NVd4Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bd058ba5cdfb8585980e615551933070b597a69c3aa1ec763a900cc44ccb26ea","last_reissued_at":"2026-06-09T01:05:33.017809Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:33.017809Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HACK++: Towards More Effective Head-Aware Key-Value Compression for Efficient Visual Autoregressive Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hang Guo, Mingbao Lin, Weiyao Lin, Wen Fei, Youru Lv, Yuchen Jiang, Ziran Qin","submitted_at":"2026-06-06T18:58:26Z","abstract_excerpt":"Visual Autoregressive (VAR) models adopt a next-scale prediction paradigm, offering high-quality generation with substantially fewer decoding steps. However, existing VAR models suffer from significant attention complexity and severe memory overhead due to the accumulation of key-value (KV) caches across scales. In this paper, we tackle this challenge by introducing KV cache compression into the next-scale paradigm. We begin with an in-depth analysis of VAR attention and observe that attention heads can be stably divided into two functionally distinct categories: Contextual Heads focus on main"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08302","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/2606.08302/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":"2606.08302","created_at":"2026-06-09T01:05:33.017876+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.08302v1","created_at":"2026-06-09T01:05:33.017876+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08302","created_at":"2026-06-09T01:05:33.017876+00:00"},{"alias_kind":"pith_short_12","alias_value":"XUCYXJON7OCY","created_at":"2026-06-09T01:05:33.017876+00:00"},{"alias_kind":"pith_short_16","alias_value":"XUCYXJON7OCYLGAO","created_at":"2026-06-09T01:05:33.017876+00:00"},{"alias_kind":"pith_short_8","alias_value":"XUCYXJON","created_at":"2026-06-09T01:05:33.017876+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/XUCYXJON7OCYLGAOMFKVDEZQOC","json":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC.json","graph_json":"https://pith.science/api/pith-number/XUCYXJON7OCYLGAOMFKVDEZQOC/graph.json","events_json":"https://pith.science/api/pith-number/XUCYXJON7OCYLGAOMFKVDEZQOC/events.json","paper":"https://pith.science/paper/XUCYXJON"},"agent_actions":{"view_html":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC","download_json":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC.json","view_paper":"https://pith.science/paper/XUCYXJON","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.08302&json=true","fetch_graph":"https://pith.science/api/pith-number/XUCYXJON7OCYLGAOMFKVDEZQOC/graph.json","fetch_events":"https://pith.science/api/pith-number/XUCYXJON7OCYLGAOMFKVDEZQOC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC/action/storage_attestation","attest_author":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC/action/author_attestation","sign_citation":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC/action/citation_signature","submit_replication":"https://pith.science/pith/XUCYXJON7OCYLGAOMFKVDEZQOC/action/replication_record"}},"created_at":"2026-06-09T01:05:33.017876+00:00","updated_at":"2026-06-09T01:05:33.017876+00:00"}