{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:LEFWJXBQJENDMU7DKPLDBBPGEN","short_pith_number":"pith:LEFWJXBQ","schema_version":"1.0","canonical_sha256":"590b64dc30491a3653e353d63085e62359dbc5fe1c6459bf291124819f2fe28e","source":{"kind":"arxiv","id":"2105.11115","version":3},"attestation_state":"computed","paper":{"title":"Self-Attention Networks Can Process Bounded Hierarchical Languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.FL"],"primary_cat":"cs.CL","authors_text":"Binghui Peng, Christos Papadimitriou, Karthik Narasimhan, Shunyu Yao","submitted_at":"2021-05-24T06:42:58Z","abstract_excerpt":"Despite their impressive performance in NLP, self-attention networks were recently proved to be limited for processing formal languages with hierarchical structure, such as $\\mathsf{Dyck}_k$, the language consisting of well-nested parentheses of $k$ types. This suggested that natural language can be approximated well with models that are too weak for formal languages, or that the role of hierarchy and recursion in natural language might be limited. We qualify this implication by proving that self-attention networks can process $\\mathsf{Dyck}_{k, D}$, the subset of $\\mathsf{Dyck}_{k}$ with dept"},"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":"2105.11115","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-05-24T06:42:58Z","cross_cats_sorted":["cs.AI","cs.FL"],"title_canon_sha256":"3ddce62828993f9dd43acb26db0ad56b05fca19ace34e0b2f200fb34358cd8d3","abstract_canon_sha256":"85b3905032e879caf75f67660f0248287e8c7ca0de99f4922cc17a49ec47a25a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:50:03.281791Z","signature_b64":"3GGXW5TyohZP34sfkeZH3sx3JEC5o+cuJP8MVa9Qp5SZ+NZOSRj23sy6+HW2M0LMOJV6gGI2LtaBTjin0OWxCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"590b64dc30491a3653e353d63085e62359dbc5fe1c6459bf291124819f2fe28e","last_reissued_at":"2026-07-05T05:50:03.281340Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:50:03.281340Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Attention Networks Can Process Bounded Hierarchical Languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.FL"],"primary_cat":"cs.CL","authors_text":"Binghui Peng, Christos Papadimitriou, Karthik Narasimhan, Shunyu Yao","submitted_at":"2021-05-24T06:42:58Z","abstract_excerpt":"Despite their impressive performance in NLP, self-attention networks were recently proved to be limited for processing formal languages with hierarchical structure, such as $\\mathsf{Dyck}_k$, the language consisting of well-nested parentheses of $k$ types. This suggested that natural language can be approximated well with models that are too weak for formal languages, or that the role of hierarchy and recursion in natural language might be limited. We qualify this implication by proving that self-attention networks can process $\\mathsf{Dyck}_{k, D}$, the subset of $\\mathsf{Dyck}_{k}$ with dept"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2105.11115","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/2105.11115/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":"2105.11115","created_at":"2026-07-05T05:50:03.281392+00:00"},{"alias_kind":"arxiv_version","alias_value":"2105.11115v3","created_at":"2026-07-05T05:50:03.281392+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2105.11115","created_at":"2026-07-05T05:50:03.281392+00:00"},{"alias_kind":"pith_short_12","alias_value":"LEFWJXBQJEND","created_at":"2026-07-05T05:50:03.281392+00:00"},{"alias_kind":"pith_short_16","alias_value":"LEFWJXBQJENDMU7D","created_at":"2026-07-05T05:50:03.281392+00:00"},{"alias_kind":"pith_short_8","alias_value":"LEFWJXBQ","created_at":"2026-07-05T05:50:03.281392+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.10019","citing_title":"The two clocks and the innovation window: When and how generative models learn rules","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05923","citing_title":"The UNDO Flip-Flop: A Controlled Probe for Reversible Semantic State Management in State Space Model","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17857","citing_title":"On the Emergence of Syntax by Means of Local Interaction","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN","json":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN.json","graph_json":"https://pith.science/api/pith-number/LEFWJXBQJENDMU7DKPLDBBPGEN/graph.json","events_json":"https://pith.science/api/pith-number/LEFWJXBQJENDMU7DKPLDBBPGEN/events.json","paper":"https://pith.science/paper/LEFWJXBQ"},"agent_actions":{"view_html":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN","download_json":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN.json","view_paper":"https://pith.science/paper/LEFWJXBQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2105.11115&json=true","fetch_graph":"https://pith.science/api/pith-number/LEFWJXBQJENDMU7DKPLDBBPGEN/graph.json","fetch_events":"https://pith.science/api/pith-number/LEFWJXBQJENDMU7DKPLDBBPGEN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN/action/storage_attestation","attest_author":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN/action/author_attestation","sign_citation":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN/action/citation_signature","submit_replication":"https://pith.science/pith/LEFWJXBQJENDMU7DKPLDBBPGEN/action/replication_record"}},"created_at":"2026-07-05T05:50:03.281392+00:00","updated_at":"2026-07-05T05:50:03.281392+00:00"}