{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:HFHG6TPHQWIHHTLC4JF3SEOAY2","short_pith_number":"pith:HFHG6TPH","schema_version":"1.0","canonical_sha256":"394e6f4de7859073cd62e24bb911c0c6904921735b8eaec8d658eebf7dcc1114","source":{"kind":"arxiv","id":"2405.18512","version":1},"attestation_state":"computed","paper":{"title":"Understanding Transformer Reasoning Capabilities via Graph Algorithms","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Anton Tsitsulin, Bahare Fatemi, Bryan Perozzi, Clayton Sanford, Ethan Hall, Jonathan Halcrow, Mehran Kazemi, Vahab Mirrokni","submitted_at":"2024-05-28T18:31:14Z","abstract_excerpt":"Which transformer scaling regimes are able to perfectly solve different classes of algorithmic problems? While tremendous empirical advances have been attained by transformer-based neural networks, a theoretical understanding of their algorithmic reasoning capabilities in realistic parameter regimes is lacking. We investigate this question in terms of the network's depth, width, and number of extra tokens for algorithm execution. Our novel representational hierarchy separates 9 algorithmic reasoning problems into classes solvable by transformers in different realistic parameter scaling regimes"},"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":"2405.18512","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-28T18:31:14Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2c67d4be1ffb5fb95f7c2432f9e145b77af7a69aa8b58d47c0a0a38abb594e12","abstract_canon_sha256":"9c8e81ed61cdd965cd6fc7dcf2177a159232c8e9b8dfd1fc1b44dee12da8f765"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:24:43.769561Z","signature_b64":"LbLfc2lXd6ExVil2//BggZhJxhQFa3CamKiMqsq2gsCyZs5A+LAJJrl9q+ywJvxLsFumu9bdl1Yq7K1/c2g8Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"394e6f4de7859073cd62e24bb911c0c6904921735b8eaec8d658eebf7dcc1114","last_reissued_at":"2026-07-05T08:24:43.769036Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:24:43.769036Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding Transformer Reasoning Capabilities via Graph Algorithms","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Anton Tsitsulin, Bahare Fatemi, Bryan Perozzi, Clayton Sanford, Ethan Hall, Jonathan Halcrow, Mehran Kazemi, Vahab Mirrokni","submitted_at":"2024-05-28T18:31:14Z","abstract_excerpt":"Which transformer scaling regimes are able to perfectly solve different classes of algorithmic problems? While tremendous empirical advances have been attained by transformer-based neural networks, a theoretical understanding of their algorithmic reasoning capabilities in realistic parameter regimes is lacking. We investigate this question in terms of the network's depth, width, and number of extra tokens for algorithm execution. Our novel representational hierarchy separates 9 algorithmic reasoning problems into classes solvable by transformers in different realistic parameter scaling regimes"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18512","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/2405.18512/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":"2405.18512","created_at":"2026-07-05T08:24:43.769098+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.18512v1","created_at":"2026-07-05T08:24:43.769098+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18512","created_at":"2026-07-05T08:24:43.769098+00:00"},{"alias_kind":"pith_short_12","alias_value":"HFHG6TPHQWIH","created_at":"2026-07-05T08:24:43.769098+00:00"},{"alias_kind":"pith_short_16","alias_value":"HFHG6TPHQWIHHTLC","created_at":"2026-07-05T08:24:43.769098+00:00"},{"alias_kind":"pith_short_8","alias_value":"HFHG6TPH","created_at":"2026-07-05T08:24:43.769098+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.22471","citing_title":"Lost in Tokenization: Fundamental Trade-offs in Graph Tokenization for Transformers","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2510.26745","citing_title":"Deep sequence models tend to memorize geometrically; it is unclear why","ref_index":154,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01484","citing_title":"Evaluating LLMs on Large-Scale Graph Property Estimation via Random Walks","ref_index":75,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2","json":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2.json","graph_json":"https://pith.science/api/pith-number/HFHG6TPHQWIHHTLC4JF3SEOAY2/graph.json","events_json":"https://pith.science/api/pith-number/HFHG6TPHQWIHHTLC4JF3SEOAY2/events.json","paper":"https://pith.science/paper/HFHG6TPH"},"agent_actions":{"view_html":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2","download_json":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2.json","view_paper":"https://pith.science/paper/HFHG6TPH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.18512&json=true","fetch_graph":"https://pith.science/api/pith-number/HFHG6TPHQWIHHTLC4JF3SEOAY2/graph.json","fetch_events":"https://pith.science/api/pith-number/HFHG6TPHQWIHHTLC4JF3SEOAY2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2/action/storage_attestation","attest_author":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2/action/author_attestation","sign_citation":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2/action/citation_signature","submit_replication":"https://pith.science/pith/HFHG6TPHQWIHHTLC4JF3SEOAY2/action/replication_record"}},"created_at":"2026-07-05T08:24:43.769098+00:00","updated_at":"2026-07-05T08:24:43.769098+00:00"}