{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZVLQNJWKIB6RJTS4ES5YUMAPQP","short_pith_number":"pith:ZVLQNJWK","schema_version":"1.0","canonical_sha256":"cd5706a6ca407d14ce5c24bb8a300f83ed25874f7b42b5df5b6f67048d012abc","source":{"kind":"arxiv","id":"2606.07317","version":1},"attestation_state":"computed","paper":{"title":"Gated Bidirectional Linear Attention for Generative Retrieval","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Artem Matveev, Sergei Liamaev, Sergei Makeev, Vladislav Tytskiy","submitted_at":"2026-06-05T14:37:02Z","abstract_excerpt":"In recommender systems, generative retrieval typically uses an encoder-decoder setup: an encoder processes a user interaction history, and an autoregressive decoder then generates recommended items. In large-scale streaming services, active users accumulate very long histories over time. As histories grow, the encoder becomes a major latency bottleneck because softmax attention scales quadratically with sequence length. In our experiments, using bidirectional attention in the encoder substantially improves quality. However, most sub-quadratic attention methods focus on causal attention.\n  We p"},"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":"2606.07317","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2026-06-05T14:37:02Z","cross_cats_sorted":[],"title_canon_sha256":"6b686121d4d43410e59c2f84d50c86a731d61af7f331bf164b4c1f450db49382","abstract_canon_sha256":"1087b54f3e786fa0518e6a14552875b3ceba0b08a58b9cb117a64ecb984374c2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-08T01:05:19.285821Z","signature_b64":"E3WnIuMtIp7cfw4PN76htDN0DfPtE/vie0NNPUAW2+tyz1u8KBfO6Gkc3bB+0UZTDA4SHelmhxJaD8rJKX8ACQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cd5706a6ca407d14ce5c24bb8a300f83ed25874f7b42b5df5b6f67048d012abc","last_reissued_at":"2026-06-08T01:05:19.285073Z","signature_status":"signed_v1","first_computed_at":"2026-06-08T01:05:19.285073Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gated Bidirectional Linear Attention for Generative Retrieval","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Artem Matveev, Sergei Liamaev, Sergei Makeev, Vladislav Tytskiy","submitted_at":"2026-06-05T14:37:02Z","abstract_excerpt":"In recommender systems, generative retrieval typically uses an encoder-decoder setup: an encoder processes a user interaction history, and an autoregressive decoder then generates recommended items. In large-scale streaming services, active users accumulate very long histories over time. As histories grow, the encoder becomes a major latency bottleneck because softmax attention scales quadratically with sequence length. In our experiments, using bidirectional attention in the encoder substantially improves quality. However, most sub-quadratic attention methods focus on causal attention.\n  We p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07317","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/2606.07317/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":"2606.07317","created_at":"2026-06-08T01:05:19.285189+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.07317v1","created_at":"2026-06-08T01:05:19.285189+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07317","created_at":"2026-06-08T01:05:19.285189+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZVLQNJWKIB6R","created_at":"2026-06-08T01:05:19.285189+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZVLQNJWKIB6RJTS4","created_at":"2026-06-08T01:05:19.285189+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZVLQNJWK","created_at":"2026-06-08T01:05:19.285189+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/ZVLQNJWKIB6RJTS4ES5YUMAPQP","json":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP.json","graph_json":"https://pith.science/api/pith-number/ZVLQNJWKIB6RJTS4ES5YUMAPQP/graph.json","events_json":"https://pith.science/api/pith-number/ZVLQNJWKIB6RJTS4ES5YUMAPQP/events.json","paper":"https://pith.science/paper/ZVLQNJWK"},"agent_actions":{"view_html":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP","download_json":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP.json","view_paper":"https://pith.science/paper/ZVLQNJWK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.07317&json=true","fetch_graph":"https://pith.science/api/pith-number/ZVLQNJWKIB6RJTS4ES5YUMAPQP/graph.json","fetch_events":"https://pith.science/api/pith-number/ZVLQNJWKIB6RJTS4ES5YUMAPQP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP/action/storage_attestation","attest_author":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP/action/author_attestation","sign_citation":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP/action/citation_signature","submit_replication":"https://pith.science/pith/ZVLQNJWKIB6RJTS4ES5YUMAPQP/action/replication_record"}},"created_at":"2026-06-08T01:05:19.285189+00:00","updated_at":"2026-06-08T01:05:19.285189+00:00"}