{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:6O4PUAC2IZ6STKG7RSSOKHCQ2B","short_pith_number":"pith:6O4PUAC2","schema_version":"1.0","canonical_sha256":"f3b8fa005a467d29a8df8ca4e51c50d048ee49ba89d19c2e103bb3fa318cbfed","source":{"kind":"arxiv","id":"2410.15787","version":2},"attestation_state":"computed","paper":{"title":"Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chulhee Yun, Hanseul Cho, Jaeyoung Cha, Srinadh Bhojanapalli","submitted_at":"2024-10-21T08:49:51Z","abstract_excerpt":"Transformers often struggle with length generalization, meaning they fail to generalize to sequences longer than those encountered during training. While arithmetic tasks are commonly used to study length generalization, certain tasks are considered notoriously difficult, e.g., multi-operand addition (requiring generalization over both the number of operands and their lengths) and multiplication (requiring generalization over both operand lengths). In this work, we achieve approximately 2-3x length generalization on both tasks, which is the first such achievement in arithmetic Transformers. We"},"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":"2410.15787","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-10-21T08:49:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"330ba2b93748952fcce2e993acf42c481c5f331e1473b1a4b6e733271ee02f22","abstract_canon_sha256":"24c468832c4a383b8837f147de38567be28118cf81d615ce0cd577f4a2f33f6b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:50:09.816131Z","signature_b64":"2Ud2y9kM0s4GX/rhO+7pIUNdEFhtzvdXcAmUoA9fXP9rIHtcXwKLid52LdvMZbY6kdyKU/1C9aZN2ZKW3qRyAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f3b8fa005a467d29a8df8ca4e51c50d048ee49ba89d19c2e103bb3fa318cbfed","last_reissued_at":"2026-07-05T10:50:09.815656Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:50:09.815656Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chulhee Yun, Hanseul Cho, Jaeyoung Cha, Srinadh Bhojanapalli","submitted_at":"2024-10-21T08:49:51Z","abstract_excerpt":"Transformers often struggle with length generalization, meaning they fail to generalize to sequences longer than those encountered during training. While arithmetic tasks are commonly used to study length generalization, certain tasks are considered notoriously difficult, e.g., multi-operand addition (requiring generalization over both the number of operands and their lengths) and multiplication (requiring generalization over both operand lengths). In this work, we achieve approximately 2-3x length generalization on both tasks, which is the first such achievement in arithmetic Transformers. We"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.15787","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/2410.15787/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":"2410.15787","created_at":"2026-07-05T10:50:09.815713+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.15787v2","created_at":"2026-07-05T10:50:09.815713+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.15787","created_at":"2026-07-05T10:50:09.815713+00:00"},{"alias_kind":"pith_short_12","alias_value":"6O4PUAC2IZ6S","created_at":"2026-07-05T10:50:09.815713+00:00"},{"alias_kind":"pith_short_16","alias_value":"6O4PUAC2IZ6STKG7","created_at":"2026-07-05T10:50:09.815713+00:00"},{"alias_kind":"pith_short_8","alias_value":"6O4PUAC2","created_at":"2026-07-05T10:50:09.815713+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.08022","citing_title":"Globally Optimal Training of Spiking Neural Networks via Parameter Reconstruction","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01651","citing_title":"On the Spatiotemporal Dynamics of Generalization in Neural Networks","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2603.29069","citing_title":"On the Mirage of Long-Range Dependency, with an Application to Integer Multiplication","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08022","citing_title":"Globally Optimal Training of Spiking Neural Networks via Parameter Reconstruction","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B","json":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B.json","graph_json":"https://pith.science/api/pith-number/6O4PUAC2IZ6STKG7RSSOKHCQ2B/graph.json","events_json":"https://pith.science/api/pith-number/6O4PUAC2IZ6STKG7RSSOKHCQ2B/events.json","paper":"https://pith.science/paper/6O4PUAC2"},"agent_actions":{"view_html":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B","download_json":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B.json","view_paper":"https://pith.science/paper/6O4PUAC2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.15787&json=true","fetch_graph":"https://pith.science/api/pith-number/6O4PUAC2IZ6STKG7RSSOKHCQ2B/graph.json","fetch_events":"https://pith.science/api/pith-number/6O4PUAC2IZ6STKG7RSSOKHCQ2B/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B/action/storage_attestation","attest_author":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B/action/author_attestation","sign_citation":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B/action/citation_signature","submit_replication":"https://pith.science/pith/6O4PUAC2IZ6STKG7RSSOKHCQ2B/action/replication_record"}},"created_at":"2026-07-05T10:50:09.815713+00:00","updated_at":"2026-07-05T10:50:09.815713+00:00"}