{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:GT5RNMXGP6JYECKKD7R6Y66DW7","short_pith_number":"pith:GT5RNMXG","schema_version":"1.0","canonical_sha256":"34fb16b2e67f9382094a1fe3ec7bc3b7d3e11266c7bed96eef877d7456ef0875","source":{"kind":"arxiv","id":"2305.14699","version":2},"attestation_state":"computed","paper":{"title":"Can Transformers Learn to Solve Problems Recursively?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LO","cs.PL"],"primary_cat":"cs.LG","authors_text":"Curt Tigges, Maxim Raginsky, Shizhuo Dylan Zhang, Stella Biderman, Talia Ringer","submitted_at":"2023-05-24T04:08:37Z","abstract_excerpt":"Neural networks have in recent years shown promise for helping software engineers write programs and even formally verify them. While semantic information plays a crucial part in these processes, it remains unclear to what degree popular neural architectures like transformers are capable of modeling that information. This paper examines the behavior of neural networks learning algorithms relevant to programs and formal verification proofs through the lens of mechanistic interpretability, focusing in particular on structural recursion. Structural recursion is at the heart of tasks on which symb"},"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":"2305.14699","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-05-24T04:08:37Z","cross_cats_sorted":["cs.AI","cs.LO","cs.PL"],"title_canon_sha256":"9c6afbea1ecf69389039fe2dbcf332c795155c00245366c2ce619f8b362cd6f1","abstract_canon_sha256":"835f67259e3d0333bab5e09cd672908422609ab6cccc48340b068a7a87808d61"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:24:29.980629Z","signature_b64":"ACBlp22hXEoUili7txpEgwjkcjGY0ftoYt6UhlzTPySSYAphDWQygnIPDUPmIgJMNDeC3Ae733dOZHUBHGxAAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"34fb16b2e67f9382094a1fe3ec7bc3b7d3e11266c7bed96eef877d7456ef0875","last_reissued_at":"2026-07-05T06:24:29.980192Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:24:29.980192Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can Transformers Learn to Solve Problems Recursively?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LO","cs.PL"],"primary_cat":"cs.LG","authors_text":"Curt Tigges, Maxim Raginsky, Shizhuo Dylan Zhang, Stella Biderman, Talia Ringer","submitted_at":"2023-05-24T04:08:37Z","abstract_excerpt":"Neural networks have in recent years shown promise for helping software engineers write programs and even formally verify them. While semantic information plays a crucial part in these processes, it remains unclear to what degree popular neural architectures like transformers are capable of modeling that information. This paper examines the behavior of neural networks learning algorithms relevant to programs and formal verification proofs through the lens of mechanistic interpretability, focusing in particular on structural recursion. Structural recursion is at the heart of tasks on which symb"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.14699","kind":"arxiv","version":2},"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/2305.14699/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":"2305.14699","created_at":"2026-07-05T06:24:29.980246+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.14699v2","created_at":"2026-07-05T06:24:29.980246+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.14699","created_at":"2026-07-05T06:24:29.980246+00:00"},{"alias_kind":"pith_short_12","alias_value":"GT5RNMXGP6JY","created_at":"2026-07-05T06:24:29.980246+00:00"},{"alias_kind":"pith_short_16","alias_value":"GT5RNMXGP6JYECKK","created_at":"2026-07-05T06:24:29.980246+00:00"},{"alias_kind":"pith_short_8","alias_value":"GT5RNMXG","created_at":"2026-07-05T06:24:29.980246+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2502.12717","citing_title":"Learning the symmetric group: large from small","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2310.10631","citing_title":"Llemma: An Open Language Model For Mathematics","ref_index":98,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12809","citing_title":"Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces","ref_index":90,"is_internal_anchor":false},{"citing_arxiv_id":"2403.07974","citing_title":"LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code","ref_index":98,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7","json":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7.json","graph_json":"https://pith.science/api/pith-number/GT5RNMXGP6JYECKKD7R6Y66DW7/graph.json","events_json":"https://pith.science/api/pith-number/GT5RNMXGP6JYECKKD7R6Y66DW7/events.json","paper":"https://pith.science/paper/GT5RNMXG"},"agent_actions":{"view_html":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7","download_json":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7.json","view_paper":"https://pith.science/paper/GT5RNMXG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.14699&json=true","fetch_graph":"https://pith.science/api/pith-number/GT5RNMXGP6JYECKKD7R6Y66DW7/graph.json","fetch_events":"https://pith.science/api/pith-number/GT5RNMXGP6JYECKKD7R6Y66DW7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7/action/storage_attestation","attest_author":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7/action/author_attestation","sign_citation":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7/action/citation_signature","submit_replication":"https://pith.science/pith/GT5RNMXGP6JYECKKD7R6Y66DW7/action/replication_record"}},"created_at":"2026-07-05T06:24:29.980246+00:00","updated_at":"2026-07-05T06:24:29.980246+00:00"}