{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XBJ7M3KV2XLYRAPUR5R2JVQNLE","short_pith_number":"pith:XBJ7M3KV","schema_version":"1.0","canonical_sha256":"b853f66d55d5d78881f48f63a4d60d59040c0388c0e476ef58b4083fc63193e4","source":{"kind":"arxiv","id":"2607.12789","version":1},"attestation_state":"computed","paper":{"title":"AVQ-Attention: Adaptive Vector-Quantized Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Amir Habibian, Max Welling, Patrick Forr\\'e, Winfried van den Dool, Yuki M. Asano","submitted_at":"2026-07-14T14:02:22Z","abstract_excerpt":"The $\\mathcal{O}(N^2)$ complexity of attention over $N$ tokens remains a computational bottleneck in transformer models. Vector-Quantized (VQ) attention reduces this to $\\mathcal{O}(MN)$ by representing keys with $M$ codewords, but applies uniform codebook capacity regardless of where attention mass concentrates: high-attention regions of key space may be coarsely approximated while low-attention regions waste representational capacity. We propose Adaptive Vector-Quantized (AVQ) Attention, which adaptively allocates codebook capacity based on attention importance. Starting from a small set of "},"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":"2607.12789","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-07-14T14:02:22Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"ff456b66bb18084c0d850919e5e2e5cef9bdb8416a4600e55a21ea9e4999be1e","abstract_canon_sha256":"6d76d2d9b29edc2bb9b625e861031cf6ddfbaa3b274cd0bf75ae994af21460e1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-15T01:22:18.056427Z","signature_b64":"PnqVyBrL5mJARQR1dNA2bN1tkjdzspU2kyD6eQ3U+xs78SfQOK9tP475KgcDrAl3Ls0c/RqyOZg+B0zL2AlPBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b853f66d55d5d78881f48f63a4d60d59040c0388c0e476ef58b4083fc63193e4","last_reissued_at":"2026-07-15T01:22:18.055581Z","signature_status":"signed_v1","first_computed_at":"2026-07-15T01:22:18.055581Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AVQ-Attention: Adaptive Vector-Quantized Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Amir Habibian, Max Welling, Patrick Forr\\'e, Winfried van den Dool, Yuki M. Asano","submitted_at":"2026-07-14T14:02:22Z","abstract_excerpt":"The $\\mathcal{O}(N^2)$ complexity of attention over $N$ tokens remains a computational bottleneck in transformer models. Vector-Quantized (VQ) attention reduces this to $\\mathcal{O}(MN)$ by representing keys with $M$ codewords, but applies uniform codebook capacity regardless of where attention mass concentrates: high-attention regions of key space may be coarsely approximated while low-attention regions waste representational capacity. We propose Adaptive Vector-Quantized (AVQ) Attention, which adaptively allocates codebook capacity based on attention importance. Starting from a small set of "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.12789","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/2607.12789/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":"2607.12789","created_at":"2026-07-15T01:22:18.056016+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.12789v1","created_at":"2026-07-15T01:22:18.056016+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.12789","created_at":"2026-07-15T01:22:18.056016+00:00"},{"alias_kind":"pith_short_12","alias_value":"XBJ7M3KV2XLY","created_at":"2026-07-15T01:22:18.056016+00:00"},{"alias_kind":"pith_short_16","alias_value":"XBJ7M3KV2XLYRAPU","created_at":"2026-07-15T01:22:18.056016+00:00"},{"alias_kind":"pith_short_8","alias_value":"XBJ7M3KV","created_at":"2026-07-15T01:22:18.056016+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/XBJ7M3KV2XLYRAPUR5R2JVQNLE","json":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE.json","graph_json":"https://pith.science/api/pith-number/XBJ7M3KV2XLYRAPUR5R2JVQNLE/graph.json","events_json":"https://pith.science/api/pith-number/XBJ7M3KV2XLYRAPUR5R2JVQNLE/events.json","paper":"https://pith.science/paper/XBJ7M3KV"},"agent_actions":{"view_html":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE","download_json":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE.json","view_paper":"https://pith.science/paper/XBJ7M3KV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.12789&json=true","fetch_graph":"https://pith.science/api/pith-number/XBJ7M3KV2XLYRAPUR5R2JVQNLE/graph.json","fetch_events":"https://pith.science/api/pith-number/XBJ7M3KV2XLYRAPUR5R2JVQNLE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE/action/storage_attestation","attest_author":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE/action/author_attestation","sign_citation":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE/action/citation_signature","submit_replication":"https://pith.science/pith/XBJ7M3KV2XLYRAPUR5R2JVQNLE/action/replication_record"}},"created_at":"2026-07-15T01:22:18.056016+00:00","updated_at":"2026-07-15T01:22:18.056016+00:00"}