{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:ZQDB6A7EB5LTQNKVB43H3OLMX2","short_pith_number":"pith:ZQDB6A7E","schema_version":"1.0","canonical_sha256":"cc061f03e40f573835550f367db96cbeb7e3e84db74b830ffa159d01613051e4","source":{"kind":"arxiv","id":"2207.02098","version":3},"attestation_state":"computed","paper":{"title":"Neural Networks and the Chomsky Hierarchy","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.FL"],"primary_cat":"cs.LG","authors_text":"Anian Ruoss, Chris Cundy, Elliot Catt, Gr\\'egoire Del\\'etang, Joel Veness, Jordi Grau-Moya, Li Kevin Wenliang, Marcus Hutter, Pedro A. Ortega, Shane Legg, Tim Genewein","submitted_at":"2022-07-05T15:06:11Z","abstract_excerpt":"Reliable generalization lies at the heart of safe ML and AI. However, understanding when and how neural networks generalize remains one of the most important unsolved problems in the field. In this work, we conduct an extensive empirical study (20'910 models, 15 tasks) to investigate whether insights from the theory of computation can predict the limits of neural network generalization in practice. We demonstrate that grouping tasks according to the Chomsky hierarchy allows us to forecast whether certain architectures will be able to generalize to out-of-distribution inputs. This includes nega"},"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":"2207.02098","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-07-05T15:06:11Z","cross_cats_sorted":["cs.AI","cs.CL","cs.FL"],"title_canon_sha256":"e2bee7cb5aade9fdc4e465cd85831359de4f12af5b1d6be7bda2739ea4a45f2c","abstract_canon_sha256":"9ea15b61cdfb71aa94964d6f468cf56a31727fca76ab54ae158aa6e99debfadf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:46:17.794096Z","signature_b64":"CigUfVkJNIZd3e9QnbXJx4Opqnm1IaRXhL4S4SxkBJ5wQ2IzlGut95uLlgAOf8usBouEZccwtW7ACBcDfkwaAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cc061f03e40f573835550f367db96cbeb7e3e84db74b830ffa159d01613051e4","last_reissued_at":"2026-07-05T05:46:17.793620Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:46:17.793620Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Neural Networks and the Chomsky Hierarchy","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.FL"],"primary_cat":"cs.LG","authors_text":"Anian Ruoss, Chris Cundy, Elliot Catt, Gr\\'egoire Del\\'etang, Joel Veness, Jordi Grau-Moya, Li Kevin Wenliang, Marcus Hutter, Pedro A. Ortega, Shane Legg, Tim Genewein","submitted_at":"2022-07-05T15:06:11Z","abstract_excerpt":"Reliable generalization lies at the heart of safe ML and AI. However, understanding when and how neural networks generalize remains one of the most important unsolved problems in the field. In this work, we conduct an extensive empirical study (20'910 models, 15 tasks) to investigate whether insights from the theory of computation can predict the limits of neural network generalization in practice. We demonstrate that grouping tasks according to the Chomsky hierarchy allows us to forecast whether certain architectures will be able to generalize to out-of-distribution inputs. This includes nega"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2207.02098","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/2207.02098/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":"2207.02098","created_at":"2026-07-05T05:46:17.793676+00:00"},{"alias_kind":"arxiv_version","alias_value":"2207.02098v3","created_at":"2026-07-05T05:46:17.793676+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2207.02098","created_at":"2026-07-05T05:46:17.793676+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZQDB6A7EB5LT","created_at":"2026-07-05T05:46:17.793676+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZQDB6A7EB5LTQNKV","created_at":"2026-07-05T05:46:17.793676+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZQDB6A7E","created_at":"2026-07-05T05:46:17.793676+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21884","citing_title":"A Verifiable Search Is Not a Learnable Chain-of-Thought","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21070","citing_title":"Towards Understanding Self-Pretraining for Sequence Classification","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01651","citing_title":"On the Spatiotemporal Dynamics of Generalization in Neural Networks","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01651","citing_title":"On the Spatiotemporal Dynamics of Generalization in Neural Networks","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2402.17762","citing_title":"Massive Activations in Large Language Models","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2602.14814","citing_title":"Learning State-Tracking from Code Using Linear RNNs","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05209","citing_title":"Are Flat Minima an Illusion?","ref_index":158,"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":22,"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":23,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02709","citing_title":"Evaluating the Formal Reasoning Capabilities of Large Language Models through Chomsky Hierarchy","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17857","citing_title":"On the Emergence of Syntax by Means of Local Interaction","ref_index":41,"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":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17857","citing_title":"On the Emergence of Syntax by Means of Local Interaction","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2","json":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2.json","graph_json":"https://pith.science/api/pith-number/ZQDB6A7EB5LTQNKVB43H3OLMX2/graph.json","events_json":"https://pith.science/api/pith-number/ZQDB6A7EB5LTQNKVB43H3OLMX2/events.json","paper":"https://pith.science/paper/ZQDB6A7E"},"agent_actions":{"view_html":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2","download_json":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2.json","view_paper":"https://pith.science/paper/ZQDB6A7E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2207.02098&json=true","fetch_graph":"https://pith.science/api/pith-number/ZQDB6A7EB5LTQNKVB43H3OLMX2/graph.json","fetch_events":"https://pith.science/api/pith-number/ZQDB6A7EB5LTQNKVB43H3OLMX2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2/action/storage_attestation","attest_author":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2/action/author_attestation","sign_citation":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2/action/citation_signature","submit_replication":"https://pith.science/pith/ZQDB6A7EB5LTQNKVB43H3OLMX2/action/replication_record"}},"created_at":"2026-07-05T05:46:17.793676+00:00","updated_at":"2026-07-05T05:46:17.793676+00:00"}