{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7KMPHQDOUYPPUP2OZR3VQ42XWD","short_pith_number":"pith:7KMPHQDO","schema_version":"1.0","canonical_sha256":"fa98f3c06ea61efa3f4ecc77587357b0e63b3e10d0aca2eab71e12610ebd1f7c","source":{"kind":"arxiv","id":"2502.12853","version":1},"attestation_state":"computed","paper":{"title":"S$^2$R: Teaching LLMs to Self-verify and Self-correct via Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Bang Zhang, Cheng Liu, Jia Li, Jiaqi Chen, Nan Du, Peisong Wang, Ruotian Ma, Xingyan Liu, Xin Zhou","submitted_at":"2025-02-18T13:40:22Z","abstract_excerpt":"Recent studies have demonstrated the effectiveness of LLM test-time scaling. However, existing approaches to incentivize LLMs' deep thinking abilities generally require large-scale data or significant training efforts. Meanwhile, it remains unclear how to improve the thinking abilities of less powerful base models. In this work, we introduce S$^2$R, an efficient framework that enhances LLM reasoning by teaching models to self-verify and self-correct during inference. Specifically, we first initialize LLMs with iterative self-verification and self-correction behaviors through supervised fine-tu"},"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":"2502.12853","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-02-18T13:40:22Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"54a4ed2e32c4fe6cc4830c8d8db45a2e119bf9d13328aac6e9eeb5ac313de1dc","abstract_canon_sha256":"8ead46da08533a3cbe5e3024f8855dd8611d8f806215bb5b36407c57878f6507"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:16:25.326660Z","signature_b64":"Ja0Pyy11wg8wW+IEPwNj5JFvUS8POZ6tfKRQ6T9mr90d6vF1plZZ6xSFy/vOwz7kKhsMRZ1yoIbu6uBNNvRUDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fa98f3c06ea61efa3f4ecc77587357b0e63b3e10d0aca2eab71e12610ebd1f7c","last_reissued_at":"2026-07-05T10:16:25.326099Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:16:25.326099Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"S$^2$R: Teaching LLMs to Self-verify and Self-correct via Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Bang Zhang, Cheng Liu, Jia Li, Jiaqi Chen, Nan Du, Peisong Wang, Ruotian Ma, Xingyan Liu, Xin Zhou","submitted_at":"2025-02-18T13:40:22Z","abstract_excerpt":"Recent studies have demonstrated the effectiveness of LLM test-time scaling. However, existing approaches to incentivize LLMs' deep thinking abilities generally require large-scale data or significant training efforts. Meanwhile, it remains unclear how to improve the thinking abilities of less powerful base models. In this work, we introduce S$^2$R, an efficient framework that enhances LLM reasoning by teaching models to self-verify and self-correct during inference. Specifically, we first initialize LLMs with iterative self-verification and self-correction behaviors through supervised fine-tu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.12853","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/2502.12853/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":"2502.12853","created_at":"2026-07-05T10:16:25.326166+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.12853v1","created_at":"2026-07-05T10:16:25.326166+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.12853","created_at":"2026-07-05T10:16:25.326166+00:00"},{"alias_kind":"pith_short_12","alias_value":"7KMPHQDOUYPP","created_at":"2026-07-05T10:16:25.326166+00:00"},{"alias_kind":"pith_short_16","alias_value":"7KMPHQDOUYPPUP2O","created_at":"2026-07-05T10:16:25.326166+00:00"},{"alias_kind":"pith_short_8","alias_value":"7KMPHQDO","created_at":"2026-07-05T10:16:25.326166+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.20745","citing_title":"The Hidden Signal of Verifier Strictness: Controlling and Improving Step-Wise Verification via Selective Latent Steering","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2502.17419","citing_title":"From System 1 to System 2: A Survey of Reasoning Large Language Models","ref_index":200,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19567","citing_title":"Multi-modal Reasoning with LLMs for Visual Semantic Arithmetic","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10228","citing_title":"SVSR: A Self-Verification and Self-Rectification Paradigm for Multimodal Reasoning","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05226","citing_title":"Internalizing Outcome Supervision into Process Supervision: A New Paradigm for Reinforcement Learning for Reasoning","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD","json":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD.json","graph_json":"https://pith.science/api/pith-number/7KMPHQDOUYPPUP2OZR3VQ42XWD/graph.json","events_json":"https://pith.science/api/pith-number/7KMPHQDOUYPPUP2OZR3VQ42XWD/events.json","paper":"https://pith.science/paper/7KMPHQDO"},"agent_actions":{"view_html":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD","download_json":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD.json","view_paper":"https://pith.science/paper/7KMPHQDO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.12853&json=true","fetch_graph":"https://pith.science/api/pith-number/7KMPHQDOUYPPUP2OZR3VQ42XWD/graph.json","fetch_events":"https://pith.science/api/pith-number/7KMPHQDOUYPPUP2OZR3VQ42XWD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD/action/storage_attestation","attest_author":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD/action/author_attestation","sign_citation":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD/action/citation_signature","submit_replication":"https://pith.science/pith/7KMPHQDOUYPPUP2OZR3VQ42XWD/action/replication_record"}},"created_at":"2026-07-05T10:16:25.326166+00:00","updated_at":"2026-07-05T10:16:25.326166+00:00"}