{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TCKAWUFM5XT4D4KL5LCXWWHZQF","short_pith_number":"pith:TCKAWUFM","schema_version":"1.0","canonical_sha256":"98940b50acede7c1f14beac57b58f981719a1b90811677f6c265cde73d244356","source":{"kind":"arxiv","id":"2606.13125","version":1},"attestation_state":"computed","paper":{"title":"Select and Improve: Understanding the Mechanics of Post-Training for Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Akshay Krishnamurthy, Audrey Huang, Nived Rajaraman","submitted_at":"2026-06-11T09:51:35Z","abstract_excerpt":"Reinforcement learning has rapidly emerged as a key component in the training of reasoning and coding models, yet it remains poorly understood from a mechanistic perspective. We study how and through what underlying processes capabilities are acquired or enhanced via reinforcement learning post-training. Our analysis, based on controlled math reasoning experiments with Qwen-2.5-1.5B, reveals two core mechanisms: strategy selection and strategy improvement. Our results highlight the role of SFT data and reinforcement learning data in activating these mechanisms, in particular showing how superv"},"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":"2606.13125","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-11T09:51:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9d020ae80ca738df6fb6b68254d7fbe6ae229bc2c5f58e80186502df46b3e636","abstract_canon_sha256":"0017fd9eac80f58cad592ba99b2e7c5fd888b0ff94ca3938dc1774ae97f9b216"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-12T01:09:41.636884Z","signature_b64":"Tt2Pz1X5Hm/9TRaRfpLdc1L1YKGyDiQouamMn2VO7qytNmhrWQEoxx6HKn5b3Nbo8yajQ8cks7tS0JMbeQeQCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"98940b50acede7c1f14beac57b58f981719a1b90811677f6c265cde73d244356","last_reissued_at":"2026-06-12T01:09:41.635994Z","signature_status":"signed_v1","first_computed_at":"2026-06-12T01:09:41.635994Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Select and Improve: Understanding the Mechanics of Post-Training for Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Akshay Krishnamurthy, Audrey Huang, Nived Rajaraman","submitted_at":"2026-06-11T09:51:35Z","abstract_excerpt":"Reinforcement learning has rapidly emerged as a key component in the training of reasoning and coding models, yet it remains poorly understood from a mechanistic perspective. We study how and through what underlying processes capabilities are acquired or enhanced via reinforcement learning post-training. Our analysis, based on controlled math reasoning experiments with Qwen-2.5-1.5B, reveals two core mechanisms: strategy selection and strategy improvement. Our results highlight the role of SFT data and reinforcement learning data in activating these mechanisms, in particular showing how superv"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.13125","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/2606.13125/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":"2606.13125","created_at":"2026-06-12T01:09:41.636141+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.13125v1","created_at":"2026-06-12T01:09:41.636141+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.13125","created_at":"2026-06-12T01:09:41.636141+00:00"},{"alias_kind":"pith_short_12","alias_value":"TCKAWUFM5XT4","created_at":"2026-06-12T01:09:41.636141+00:00"},{"alias_kind":"pith_short_16","alias_value":"TCKAWUFM5XT4D4KL","created_at":"2026-06-12T01:09:41.636141+00:00"},{"alias_kind":"pith_short_8","alias_value":"TCKAWUFM","created_at":"2026-06-12T01:09:41.636141+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/TCKAWUFM5XT4D4KL5LCXWWHZQF","json":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF.json","graph_json":"https://pith.science/api/pith-number/TCKAWUFM5XT4D4KL5LCXWWHZQF/graph.json","events_json":"https://pith.science/api/pith-number/TCKAWUFM5XT4D4KL5LCXWWHZQF/events.json","paper":"https://pith.science/paper/TCKAWUFM"},"agent_actions":{"view_html":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF","download_json":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF.json","view_paper":"https://pith.science/paper/TCKAWUFM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.13125&json=true","fetch_graph":"https://pith.science/api/pith-number/TCKAWUFM5XT4D4KL5LCXWWHZQF/graph.json","fetch_events":"https://pith.science/api/pith-number/TCKAWUFM5XT4D4KL5LCXWWHZQF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF/action/storage_attestation","attest_author":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF/action/author_attestation","sign_citation":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF/action/citation_signature","submit_replication":"https://pith.science/pith/TCKAWUFM5XT4D4KL5LCXWWHZQF/action/replication_record"}},"created_at":"2026-06-12T01:09:41.636141+00:00","updated_at":"2026-06-12T01:09:41.636141+00:00"}