{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:O4CBYOB5WUMWHBKC7EVCG2KQYK","short_pith_number":"pith:O4CBYOB5","schema_version":"1.0","canonical_sha256":"77041c383db519638542f92a236950c29c339641819bb0273efe4ceb9060cf61","source":{"kind":"arxiv","id":"2506.15894","version":1},"attestation_state":"computed","paper":{"title":"Language Models can perform Single-Utterance Self-Correction of Perturbed Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Eddie Kim, Ivan Zhang, Jimin Sun, Sam Silver, Sara Hooker","submitted_at":"2025-06-18T21:35:44Z","abstract_excerpt":"Large Language Models (LLMs) have demonstrated impressive mathematical reasoning capabilities, yet their performance remains brittle to minor variations in problem description and prompting strategy. Furthermore, reasoning is vulnerable to sampling-induced errors which autoregressive models must primarily address using self-correction via additionally-generated tokens. To better understand self-correction capabilities of recent models, we conduct experiments measuring models' ability to self-correct synthetic perturbations introduced into their Chain of Thought (CoT) reasoning. We observe robu"},"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":"2506.15894","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-06-18T21:35:44Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"14f554f3853671c9e28b34a3f27057c84a9a5619ed1c76c5c12e8fea0058cda6","abstract_canon_sha256":"15f241fdde99496b1c118f1b5ab07fe469fdea417d13025c94f2c4e834b7a1dc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:24:26.010272Z","signature_b64":"B6v2GXTZ5b/0Chgj+P8x0d1cX1RJb7/1CERyoDcbMjJfhK/hZ/Sj7KJXXVME0C5MwqcMnc5JRYBraSrawQBQDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"77041c383db519638542f92a236950c29c339641819bb0273efe4ceb9060cf61","last_reissued_at":"2026-07-05T11:24:26.009773Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:24:26.009773Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Language Models can perform Single-Utterance Self-Correction of Perturbed Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Eddie Kim, Ivan Zhang, Jimin Sun, Sam Silver, Sara Hooker","submitted_at":"2025-06-18T21:35:44Z","abstract_excerpt":"Large Language Models (LLMs) have demonstrated impressive mathematical reasoning capabilities, yet their performance remains brittle to minor variations in problem description and prompting strategy. Furthermore, reasoning is vulnerable to sampling-induced errors which autoregressive models must primarily address using self-correction via additionally-generated tokens. To better understand self-correction capabilities of recent models, we conduct experiments measuring models' ability to self-correct synthetic perturbations introduced into their Chain of Thought (CoT) reasoning. We observe robu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.15894","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/2506.15894/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":"2506.15894","created_at":"2026-07-05T11:24:26.009836+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.15894v1","created_at":"2026-07-05T11:24:26.009836+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.15894","created_at":"2026-07-05T11:24:26.009836+00:00"},{"alias_kind":"pith_short_12","alias_value":"O4CBYOB5WUMW","created_at":"2026-07-05T11:24:26.009836+00:00"},{"alias_kind":"pith_short_16","alias_value":"O4CBYOB5WUMWHBKC","created_at":"2026-07-05T11:24:26.009836+00:00"},{"alias_kind":"pith_short_8","alias_value":"O4CBYOB5","created_at":"2026-07-05T11:24:26.009836+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK","json":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK.json","graph_json":"https://pith.science/api/pith-number/O4CBYOB5WUMWHBKC7EVCG2KQYK/graph.json","events_json":"https://pith.science/api/pith-number/O4CBYOB5WUMWHBKC7EVCG2KQYK/events.json","paper":"https://pith.science/paper/O4CBYOB5"},"agent_actions":{"view_html":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK","download_json":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK.json","view_paper":"https://pith.science/paper/O4CBYOB5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.15894&json=true","fetch_graph":"https://pith.science/api/pith-number/O4CBYOB5WUMWHBKC7EVCG2KQYK/graph.json","fetch_events":"https://pith.science/api/pith-number/O4CBYOB5WUMWHBKC7EVCG2KQYK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK/action/storage_attestation","attest_author":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK/action/author_attestation","sign_citation":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK/action/citation_signature","submit_replication":"https://pith.science/pith/O4CBYOB5WUMWHBKC7EVCG2KQYK/action/replication_record"}},"created_at":"2026-07-05T11:24:26.009836+00:00","updated_at":"2026-07-05T11:24:26.009836+00:00"}