{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FHBK7ISUFBZYV7TBMFT7M42EPJ","short_pith_number":"pith:FHBK7ISU","schema_version":"1.0","canonical_sha256":"29c2afa25428738afe616167f673447a5ac9cd35488d5feb54a9732ef49f441b","source":{"kind":"arxiv","id":"2412.02975","version":1},"attestation_state":"computed","paper":{"title":"Theoretical limitations of multi-layer Transformer","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CC","cs.DS"],"primary_cat":"cs.LG","authors_text":"Binghui Peng, Hongxun Wu, Lijie Chen","submitted_at":"2024-12-04T02:37:31Z","abstract_excerpt":"Transformers, especially the decoder-only variants, are the backbone of most modern large language models; yet we do not have much understanding of their expressive power except for the simple $1$-layer case.\n  Due to the difficulty of analyzing multi-layer models, all previous work relies on unproven complexity conjectures to show limitations for multi-layer Transformers. In this work, we prove the first $\\textit{unconditional}$ lower bound against multi-layer decoder-only transformers. For any constant $L$, we prove that any $L$-layer decoder-only transformer needs a polynomial model dimensi"},"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":"2412.02975","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-12-04T02:37:31Z","cross_cats_sorted":["cs.AI","cs.CC","cs.DS"],"title_canon_sha256":"af404ade2cb972bb37e94e5241bc678bbc44af9efb5d2eba3a50fa8ed193b09c","abstract_canon_sha256":"b5f690dab5598e1e6921c1f8b84a78329e7f5ae9edd8e2213e7c8165bfcd61b4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:44:22.871570Z","signature_b64":"JnxR7TR4XgVI0aax/mPrVxXKJljbX/veULTRNw1udLgsDRQx2IkPXrp6TTCjfIRVKDF240cLeUX2Z+soyuX4Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"29c2afa25428738afe616167f673447a5ac9cd35488d5feb54a9732ef49f441b","last_reissued_at":"2026-07-05T09:44:22.871062Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:44:22.871062Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Theoretical limitations of multi-layer Transformer","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CC","cs.DS"],"primary_cat":"cs.LG","authors_text":"Binghui Peng, Hongxun Wu, Lijie Chen","submitted_at":"2024-12-04T02:37:31Z","abstract_excerpt":"Transformers, especially the decoder-only variants, are the backbone of most modern large language models; yet we do not have much understanding of their expressive power except for the simple $1$-layer case.\n  Due to the difficulty of analyzing multi-layer models, all previous work relies on unproven complexity conjectures to show limitations for multi-layer Transformers. In this work, we prove the first $\\textit{unconditional}$ lower bound against multi-layer decoder-only transformers. For any constant $L$, we prove that any $L$-layer decoder-only transformer needs a polynomial model dimensi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.02975","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/2412.02975/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":"2412.02975","created_at":"2026-07-05T09:44:22.871123+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.02975v1","created_at":"2026-07-05T09:44:22.871123+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.02975","created_at":"2026-07-05T09:44:22.871123+00:00"},{"alias_kind":"pith_short_12","alias_value":"FHBK7ISUFBZY","created_at":"2026-07-05T09:44:22.871123+00:00"},{"alias_kind":"pith_short_16","alias_value":"FHBK7ISUFBZYV7TB","created_at":"2026-07-05T09:44:22.871123+00:00"},{"alias_kind":"pith_short_8","alias_value":"FHBK7ISU","created_at":"2026-07-05T09:44:22.871123+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"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":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00183","citing_title":"Agentic Transformers Provably Learn to Search via Reinforcement Learning","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2510.26745","citing_title":"Deep sequence models tend to memorize geometrically; it is unclear why","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2507.12549","citing_title":"The Serial Scaling Hypothesis","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09867","citing_title":"Continuous Latent Contexts Enable Efficient Online Learning in Transformers","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22951","citing_title":"The Power of Power Law: Asymmetry Enables Compositional Reasoning","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ","json":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ.json","graph_json":"https://pith.science/api/pith-number/FHBK7ISUFBZYV7TBMFT7M42EPJ/graph.json","events_json":"https://pith.science/api/pith-number/FHBK7ISUFBZYV7TBMFT7M42EPJ/events.json","paper":"https://pith.science/paper/FHBK7ISU"},"agent_actions":{"view_html":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ","download_json":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ.json","view_paper":"https://pith.science/paper/FHBK7ISU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.02975&json=true","fetch_graph":"https://pith.science/api/pith-number/FHBK7ISUFBZYV7TBMFT7M42EPJ/graph.json","fetch_events":"https://pith.science/api/pith-number/FHBK7ISUFBZYV7TBMFT7M42EPJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ/action/storage_attestation","attest_author":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ/action/author_attestation","sign_citation":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ/action/citation_signature","submit_replication":"https://pith.science/pith/FHBK7ISUFBZYV7TBMFT7M42EPJ/action/replication_record"}},"created_at":"2026-07-05T09:44:22.871123+00:00","updated_at":"2026-07-05T09:44:22.871123+00:00"}