{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4VVWGLACV6VZKCXWWY7QYB5W2X","short_pith_number":"pith:4VVWGLAC","schema_version":"1.0","canonical_sha256":"e56b632c02afab950af6b63f0c07b6d5da59ef4b6dfb07d67b52622db16ae97c","source":{"kind":"arxiv","id":"2405.16674","version":3},"attestation_state":"computed","paper":{"title":"Limits of Deep Learning: Sequence Modeling through the Lens of Complexity Theory","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CC","cs.LO"],"primary_cat":"cs.LG","authors_text":"Aurelio Sulser, Davide Scaramuzza, Federico Sold\\'a, Nikola Zubi\\'c","submitted_at":"2024-05-26T19:33:23Z","abstract_excerpt":"Despite their successes, deep learning models struggle with tasks requiring complex reasoning and function composition. We present a theoretical and empirical investigation into the limitations of Structured State Space Models (SSMs) and Transformers in such tasks. We prove that one-layer SSMs cannot efficiently perform function composition over large domains without impractically large state sizes, and even with Chain-of-Thought prompting, they require a number of steps that scale unfavorably with the complexity of the function composition. Also, the language of a finite-precision SSM is with"},"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":"2405.16674","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-26T19:33:23Z","cross_cats_sorted":["cs.CC","cs.LO"],"title_canon_sha256":"93a1f2910efe52e4cc039f214c17d3d7d169aa801c42ec2fb592b75e4206f365","abstract_canon_sha256":"d814ba09b5885559a08c940127e80d0f957b1ce3c40ce0d0200c005a232e2a04"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:22:00.094787Z","signature_b64":"+39XN6fXnxiYaNxY1KIfdPGP16KQ+F7D55/dpw/tH+Rg8JlMOLbX18Oum0CtrIJZn/0uqDtrcZn6MiM0xBaCCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e56b632c02afab950af6b63f0c07b6d5da59ef4b6dfb07d67b52622db16ae97c","last_reissued_at":"2026-07-05T10:22:00.094242Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:22:00.094242Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Limits of Deep Learning: Sequence Modeling through the Lens of Complexity Theory","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CC","cs.LO"],"primary_cat":"cs.LG","authors_text":"Aurelio Sulser, Davide Scaramuzza, Federico Sold\\'a, Nikola Zubi\\'c","submitted_at":"2024-05-26T19:33:23Z","abstract_excerpt":"Despite their successes, deep learning models struggle with tasks requiring complex reasoning and function composition. We present a theoretical and empirical investigation into the limitations of Structured State Space Models (SSMs) and Transformers in such tasks. We prove that one-layer SSMs cannot efficiently perform function composition over large domains without impractically large state sizes, and even with Chain-of-Thought prompting, they require a number of steps that scale unfavorably with the complexity of the function composition. Also, the language of a finite-precision SSM is with"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.16674","kind":"arxiv","version":3},"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/2405.16674/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":"2405.16674","created_at":"2026-07-05T10:22:00.094312+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.16674v3","created_at":"2026-07-05T10:22:00.094312+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.16674","created_at":"2026-07-05T10:22:00.094312+00:00"},{"alias_kind":"pith_short_12","alias_value":"4VVWGLACV6VZ","created_at":"2026-07-05T10:22:00.094312+00:00"},{"alias_kind":"pith_short_16","alias_value":"4VVWGLACV6VZKCXW","created_at":"2026-07-05T10:22:00.094312+00:00"},{"alias_kind":"pith_short_8","alias_value":"4VVWGLAC","created_at":"2026-07-05T10:22:00.094312+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01532","citing_title":"Rethinking the Role of Positional Encoding: Sliding-Window Transformers without PE Remain Turing Complete","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2506.06941","citing_title":"The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X","json":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X.json","graph_json":"https://pith.science/api/pith-number/4VVWGLACV6VZKCXWWY7QYB5W2X/graph.json","events_json":"https://pith.science/api/pith-number/4VVWGLACV6VZKCXWWY7QYB5W2X/events.json","paper":"https://pith.science/paper/4VVWGLAC"},"agent_actions":{"view_html":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X","download_json":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X.json","view_paper":"https://pith.science/paper/4VVWGLAC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.16674&json=true","fetch_graph":"https://pith.science/api/pith-number/4VVWGLACV6VZKCXWWY7QYB5W2X/graph.json","fetch_events":"https://pith.science/api/pith-number/4VVWGLACV6VZKCXWWY7QYB5W2X/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X/action/storage_attestation","attest_author":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X/action/author_attestation","sign_citation":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X/action/citation_signature","submit_replication":"https://pith.science/pith/4VVWGLACV6VZKCXWWY7QYB5W2X/action/replication_record"}},"created_at":"2026-07-05T10:22:00.094312+00:00","updated_at":"2026-07-05T10:22:00.094312+00:00"}