{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ZP3UHA5J4RVWIVIP4AATCQVVHA","short_pith_number":"pith:ZP3UHA5J","schema_version":"1.0","canonical_sha256":"cbf74383a9e46b64550fe0013142b53811cad01c42eebe248ce08a1b182555c5","source":{"kind":"arxiv","id":"2409.17954","version":2},"attestation_state":"computed","paper":{"title":"Enhancing elusive clues in knowledge learning by contrasting attention of language models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Jian Gao, Ji Wu, Miao Li, Xiao Zhang","submitted_at":"2024-09-26T15:30:54Z","abstract_excerpt":"Causal language models acquire vast amount of knowledge from general text corpus during pretraining, but the efficiency of knowledge learning is known to be unsatisfactory, especially when learning from knowledge-dense and small-sized corpora. The deficiency can come from long-distance dependencies which are hard to capture by language models, and overfitting to co-occurrence patterns and distracting clues in the training text. To address these issues, the paper proposes a method to enhance knowledge learning during language model pretraining, by enhancing elusive but important clues in text d"},"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":"2409.17954","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-09-26T15:30:54Z","cross_cats_sorted":[],"title_canon_sha256":"52354edb917e0b732a9f602f37404270321e9ec2b50386fdef5a2764b0f6b03b","abstract_canon_sha256":"4639a54d59327acc13580183fbd58816a742b47d8759ba00809ebfe348b0ff49"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:29:23.724428Z","signature_b64":"g7ONaneTZ5G2zLPbzoNQhJVsv1EOLxDZvGFkLdIhxACqsuRZEyVvmiLq7/Ljw/CECUyV21kiME62MHzMOjurAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cbf74383a9e46b64550fe0013142b53811cad01c42eebe248ce08a1b182555c5","last_reissued_at":"2026-07-05T10:29:23.724000Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:29:23.724000Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Enhancing elusive clues in knowledge learning by contrasting attention of language models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Jian Gao, Ji Wu, Miao Li, Xiao Zhang","submitted_at":"2024-09-26T15:30:54Z","abstract_excerpt":"Causal language models acquire vast amount of knowledge from general text corpus during pretraining, but the efficiency of knowledge learning is known to be unsatisfactory, especially when learning from knowledge-dense and small-sized corpora. The deficiency can come from long-distance dependencies which are hard to capture by language models, and overfitting to co-occurrence patterns and distracting clues in the training text. To address these issues, the paper proposes a method to enhance knowledge learning during language model pretraining, by enhancing elusive but important clues in text d"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.17954","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/2409.17954/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":"2409.17954","created_at":"2026-07-05T10:29:23.724058+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.17954v2","created_at":"2026-07-05T10:29:23.724058+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.17954","created_at":"2026-07-05T10:29:23.724058+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZP3UHA5J4RVW","created_at":"2026-07-05T10:29:23.724058+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZP3UHA5J4RVWIVIP","created_at":"2026-07-05T10:29:23.724058+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZP3UHA5J","created_at":"2026-07-05T10:29:23.724058+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.09475","citing_title":"ResidualDroppath: Enhancing Feature Reuse over Residual Connections","ref_index":28,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA","json":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA.json","graph_json":"https://pith.science/api/pith-number/ZP3UHA5J4RVWIVIP4AATCQVVHA/graph.json","events_json":"https://pith.science/api/pith-number/ZP3UHA5J4RVWIVIP4AATCQVVHA/events.json","paper":"https://pith.science/paper/ZP3UHA5J"},"agent_actions":{"view_html":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA","download_json":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA.json","view_paper":"https://pith.science/paper/ZP3UHA5J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.17954&json=true","fetch_graph":"https://pith.science/api/pith-number/ZP3UHA5J4RVWIVIP4AATCQVVHA/graph.json","fetch_events":"https://pith.science/api/pith-number/ZP3UHA5J4RVWIVIP4AATCQVVHA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA/action/storage_attestation","attest_author":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA/action/author_attestation","sign_citation":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA/action/citation_signature","submit_replication":"https://pith.science/pith/ZP3UHA5J4RVWIVIP4AATCQVVHA/action/replication_record"}},"created_at":"2026-07-05T10:29:23.724058+00:00","updated_at":"2026-07-05T10:29:23.724058+00:00"}