{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:44MKPKEXF7V3ZLZQEV64MBXFNA","short_pith_number":"pith:44MKPKEX","schema_version":"1.0","canonical_sha256":"e718a7a8972febbcaf30257dc606e56839ca1c459706ce2e56b2b26e69e9eeb2","source":{"kind":"arxiv","id":"2507.09331","version":2},"attestation_state":"computed","paper":{"title":"Correcting the LogQ Correction: Revisiting Sampled Softmax for Large-Scale Retrieval","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Artem Matveev, Kirill Khrylchenko, Sergei Liamaev, Sergei Makeev, Vladimir Baikalov","submitted_at":"2025-07-12T16:16:11Z","abstract_excerpt":"Two-tower neural networks are a popular architecture for the retrieval stage in recommender systems. These models are typically trained with a softmax loss over the item catalog. However, in web-scale settings, the item catalog is often prohibitively large, making full softmax infeasible. A common solution is sampled softmax, which approximates the full softmax using a small number of sampled negatives.\n  One practical and widely adopted approach is to use in-batch negatives, where negatives are drawn from items in the current mini-batch. However, this introduces a bias: items that appear more"},"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":"2507.09331","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2025-07-12T16:16:11Z","cross_cats_sorted":[],"title_canon_sha256":"5f407229be0a4216c96b4b78836010b75827afc457eec0226f8c73ff3ed0f1c8","abstract_canon_sha256":"ecde579723b531a7fd823146ed214b36b9917a027853e66b6c92c2c50b42549a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:43:33.482586Z","signature_b64":"8HtHbHZ+mVJY0qKl4DY+wlGgtewRB6FWvgh7j99SKuZNs5vncVlui7EiFW5ajNkzIjf7/YtchR3w2Z4DV81LBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e718a7a8972febbcaf30257dc606e56839ca1c459706ce2e56b2b26e69e9eeb2","last_reissued_at":"2026-07-05T11:43:33.482147Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:43:33.482147Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Correcting the LogQ Correction: Revisiting Sampled Softmax for Large-Scale Retrieval","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Artem Matveev, Kirill Khrylchenko, Sergei Liamaev, Sergei Makeev, Vladimir Baikalov","submitted_at":"2025-07-12T16:16:11Z","abstract_excerpt":"Two-tower neural networks are a popular architecture for the retrieval stage in recommender systems. These models are typically trained with a softmax loss over the item catalog. However, in web-scale settings, the item catalog is often prohibitively large, making full softmax infeasible. A common solution is sampled softmax, which approximates the full softmax using a small number of sampled negatives.\n  One practical and widely adopted approach is to use in-batch negatives, where negatives are drawn from items in the current mini-batch. However, this introduces a bias: items that appear more"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.09331","kind":"arxiv","version":2},"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/2507.09331/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":"2507.09331","created_at":"2026-07-05T11:43:33.482207+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.09331v2","created_at":"2026-07-05T11:43:33.482207+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.09331","created_at":"2026-07-05T11:43:33.482207+00:00"},{"alias_kind":"pith_short_12","alias_value":"44MKPKEXF7V3","created_at":"2026-07-05T11:43:33.482207+00:00"},{"alias_kind":"pith_short_16","alias_value":"44MKPKEXF7V3ZLZQ","created_at":"2026-07-05T11:43:33.482207+00:00"},{"alias_kind":"pith_short_8","alias_value":"44MKPKEX","created_at":"2026-07-05T11:43:33.482207+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/44MKPKEXF7V3ZLZQEV64MBXFNA","json":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA.json","graph_json":"https://pith.science/api/pith-number/44MKPKEXF7V3ZLZQEV64MBXFNA/graph.json","events_json":"https://pith.science/api/pith-number/44MKPKEXF7V3ZLZQEV64MBXFNA/events.json","paper":"https://pith.science/paper/44MKPKEX"},"agent_actions":{"view_html":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA","download_json":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA.json","view_paper":"https://pith.science/paper/44MKPKEX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.09331&json=true","fetch_graph":"https://pith.science/api/pith-number/44MKPKEXF7V3ZLZQEV64MBXFNA/graph.json","fetch_events":"https://pith.science/api/pith-number/44MKPKEXF7V3ZLZQEV64MBXFNA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA/action/storage_attestation","attest_author":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA/action/author_attestation","sign_citation":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA/action/citation_signature","submit_replication":"https://pith.science/pith/44MKPKEXF7V3ZLZQEV64MBXFNA/action/replication_record"}},"created_at":"2026-07-05T11:43:33.482207+00:00","updated_at":"2026-07-05T11:43:33.482207+00:00"}