{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:JWHNTNTIYEFUC57S6R7ZWGR52Z","short_pith_number":"pith:JWHNTNTI","schema_version":"1.0","canonical_sha256":"4d8ed9b668c10b4177f2f47f9b1a3dd64936a763e591f51ef6241a2c8681e432","source":{"kind":"arxiv","id":"2110.06176","version":2},"attestation_state":"computed","paper":{"title":"Mention Memory: incorporating textual knowledge into Transformers through entity mention attention","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Fei Sha, Michiel de Jong, Nicholas FitzGerald, William Cohen, Yury Zemlyanskiy","submitted_at":"2021-10-12T17:19:05Z","abstract_excerpt":"Natural language understanding tasks such as open-domain question answering often require retrieving and assimilating factual information from multiple sources. We propose to address this problem by integrating a semi-parametric representation of a large text corpus into a Transformer model as a source of factual knowledge. Specifically, our method represents knowledge with `mention memory', a table of dense vector representations of every entity mention in a corpus. The proposed model - TOME - is a Transformer that accesses the information through internal memory layers in which each entity m"},"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":"2110.06176","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-10-12T17:19:05Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"30c24b7653add93fb2de1922175fed8df64cb93d86c0fc1c1b8d04379f25df5e","abstract_canon_sha256":"6f3fc84d44be1968ffe2d8c34ad1f3e65bd39a40a53f4e9177ea7a732773a47e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:16:15.524301Z","signature_b64":"ojded6ZQvulCf8ncYfSMPkTL25P/mkAwI73kbwbifL/V+ApTcxbgQNYj0ekUu+NvrMxDb4vhYmaFd1RElpn+CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4d8ed9b668c10b4177f2f47f9b1a3dd64936a763e591f51ef6241a2c8681e432","last_reissued_at":"2026-07-05T04:16:15.523775Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:16:15.523775Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mention Memory: incorporating textual knowledge into Transformers through entity mention attention","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Fei Sha, Michiel de Jong, Nicholas FitzGerald, William Cohen, Yury Zemlyanskiy","submitted_at":"2021-10-12T17:19:05Z","abstract_excerpt":"Natural language understanding tasks such as open-domain question answering often require retrieving and assimilating factual information from multiple sources. We propose to address this problem by integrating a semi-parametric representation of a large text corpus into a Transformer model as a source of factual knowledge. Specifically, our method represents knowledge with `mention memory', a table of dense vector representations of every entity mention in a corpus. The proposed model - TOME - is a Transformer that accesses the information through internal memory layers in which each entity m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2110.06176","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/2110.06176/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":"2110.06176","created_at":"2026-07-05T04:16:15.523835+00:00"},{"alias_kind":"arxiv_version","alias_value":"2110.06176v2","created_at":"2026-07-05T04:16:15.523835+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2110.06176","created_at":"2026-07-05T04:16:15.523835+00:00"},{"alias_kind":"pith_short_12","alias_value":"JWHNTNTIYEFU","created_at":"2026-07-05T04:16:15.523835+00:00"},{"alias_kind":"pith_short_16","alias_value":"JWHNTNTIYEFUC57S","created_at":"2026-07-05T04:16:15.523835+00:00"},{"alias_kind":"pith_short_8","alias_value":"JWHNTNTI","created_at":"2026-07-05T04:16:15.523835+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.05431","citing_title":"APE: Faster and Longer Context-Augmented Generation via Adaptive Parallel Encoding","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z","json":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z.json","graph_json":"https://pith.science/api/pith-number/JWHNTNTIYEFUC57S6R7ZWGR52Z/graph.json","events_json":"https://pith.science/api/pith-number/JWHNTNTIYEFUC57S6R7ZWGR52Z/events.json","paper":"https://pith.science/paper/JWHNTNTI"},"agent_actions":{"view_html":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z","download_json":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z.json","view_paper":"https://pith.science/paper/JWHNTNTI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2110.06176&json=true","fetch_graph":"https://pith.science/api/pith-number/JWHNTNTIYEFUC57S6R7ZWGR52Z/graph.json","fetch_events":"https://pith.science/api/pith-number/JWHNTNTIYEFUC57S6R7ZWGR52Z/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z/action/storage_attestation","attest_author":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z/action/author_attestation","sign_citation":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z/action/citation_signature","submit_replication":"https://pith.science/pith/JWHNTNTIYEFUC57S6R7ZWGR52Z/action/replication_record"}},"created_at":"2026-07-05T04:16:15.523835+00:00","updated_at":"2026-07-05T04:16:15.523835+00:00"}