{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:CKWX33GBOIO45WIXQOIWAX2ZAA","short_pith_number":"pith:CKWX33GB","schema_version":"1.0","canonical_sha256":"12ad7decc1721dced9178391605f59002787b79ae4c0af05b1514e483ca52e01","source":{"kind":"arxiv","id":"2412.01951","version":2},"attestation_state":"computed","paper":{"title":"Self-Improvement in Language Models: The Sharpening Mechanism","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG","stat.ML"],"primary_cat":"cs.AI","authors_text":"Adam Block, Akshay Krishnamurthy, Audrey Huang, Cyril Zhang, Dhruv Rohatgi, Dylan J. Foster, Jordan T. Ash, Max Simchowitz","submitted_at":"2024-12-02T20:24:17Z","abstract_excerpt":"Recent work in language modeling has raised the possibility of self-improvement, where a language models evaluates and refines its own generations to achieve higher performance without external feedback. It is impossible for this self-improvement to create information that is not already in the model, so why should we expect that this will lead to improved capabilities? We offer a new perspective on the capabilities of self-improvement through a lens we refer to as sharpening. Motivated by the observation that language models are often better at verifying response quality than they are at gene"},"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":"2412.01951","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-12-02T20:24:17Z","cross_cats_sorted":["cs.CL","cs.LG","stat.ML"],"title_canon_sha256":"f6f00eb84115be9200f8dacd9d1852b04f56151dbcc4c16bb3c334b583283e05","abstract_canon_sha256":"d3f01a1848a198a061249030b33b5786fa46661ed5d3eedc96eb9de561464233"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:44:17.841325Z","signature_b64":"D/7Q5oQ8hd9OvTCeXI/NKBlf2s2mW2+XhVT34Xzm4W+05i4XT3paPoUlJNttgw4bGobXr+zCcXpNqED1FTpGCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"12ad7decc1721dced9178391605f59002787b79ae4c0af05b1514e483ca52e01","last_reissued_at":"2026-07-05T09:44:17.840853Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:44:17.840853Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Improvement in Language Models: The Sharpening Mechanism","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG","stat.ML"],"primary_cat":"cs.AI","authors_text":"Adam Block, Akshay Krishnamurthy, Audrey Huang, Cyril Zhang, Dhruv Rohatgi, Dylan J. Foster, Jordan T. Ash, Max Simchowitz","submitted_at":"2024-12-02T20:24:17Z","abstract_excerpt":"Recent work in language modeling has raised the possibility of self-improvement, where a language models evaluates and refines its own generations to achieve higher performance without external feedback. It is impossible for this self-improvement to create information that is not already in the model, so why should we expect that this will lead to improved capabilities? We offer a new perspective on the capabilities of self-improvement through a lens we refer to as sharpening. Motivated by the observation that language models are often better at verifying response quality than they are at gene"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.01951","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/2412.01951/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":"2412.01951","created_at":"2026-07-05T09:44:17.840905+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.01951v2","created_at":"2026-07-05T09:44:17.840905+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.01951","created_at":"2026-07-05T09:44:17.840905+00:00"},{"alias_kind":"pith_short_12","alias_value":"CKWX33GBOIO4","created_at":"2026-07-05T09:44:17.840905+00:00"},{"alias_kind":"pith_short_16","alias_value":"CKWX33GBOIO45WIX","created_at":"2026-07-05T09:44:17.840905+00:00"},{"alias_kind":"pith_short_8","alias_value":"CKWX33GB","created_at":"2026-07-05T09:44:17.840905+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.13125","citing_title":"Select and Improve: Understanding the Mechanics of Post-Training for Reasoning","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03962","citing_title":"Using Reward Uncertainty to Induce Diverse Behaviour in Reinforcement Learning","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01249","citing_title":"Trust Region On-Policy Distillation","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01075","citing_title":"On the Generalization Gap in Self-Evolving Language Model Reasoning","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2505.15134","citing_title":"The Unreasonable Effectiveness of Entropy Minimization in LLM Reasoning","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04855","citing_title":"The Role of Generator Access in Autoregressive Post-Training","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16259","citing_title":"Beyond Distribution Sharpening: The Importance of Task Rewards","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA","json":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA.json","graph_json":"https://pith.science/api/pith-number/CKWX33GBOIO45WIXQOIWAX2ZAA/graph.json","events_json":"https://pith.science/api/pith-number/CKWX33GBOIO45WIXQOIWAX2ZAA/events.json","paper":"https://pith.science/paper/CKWX33GB"},"agent_actions":{"view_html":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA","download_json":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA.json","view_paper":"https://pith.science/paper/CKWX33GB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.01951&json=true","fetch_graph":"https://pith.science/api/pith-number/CKWX33GBOIO45WIXQOIWAX2ZAA/graph.json","fetch_events":"https://pith.science/api/pith-number/CKWX33GBOIO45WIXQOIWAX2ZAA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA/action/storage_attestation","attest_author":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA/action/author_attestation","sign_citation":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA/action/citation_signature","submit_replication":"https://pith.science/pith/CKWX33GBOIO45WIXQOIWAX2ZAA/action/replication_record"}},"created_at":"2026-07-05T09:44:17.840905+00:00","updated_at":"2026-07-05T09:44:17.840905+00:00"}