{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:PLRMKJUKLJVBJUWMKPUCMW7Z6U","short_pith_number":"pith:PLRMKJUK","schema_version":"1.0","canonical_sha256":"7ae2c5268a5a6a14d2cc53e8265bf9f51d5ea9e47e6356b69b0b3508882ad51e","source":{"kind":"arxiv","id":"2405.16043","version":1},"attestation_state":"computed","paper":{"title":"Theoretical Analysis of Weak-to-Strong Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aravindan Vijayaraghavan, David Sontag, Hunter Lang","submitted_at":"2024-05-25T03:48:12Z","abstract_excerpt":"Strong student models can learn from weaker teachers: when trained on the predictions of a weaker model, a strong pretrained student can learn to correct the weak model's errors and generalize to examples where the teacher is not confident, even when these examples are excluded from training. This enables learning from cheap, incomplete, and possibly incorrect label information, such as coarse logical rules or the generations of a language model. We show that existing weak supervision theory fails to account for both of these effects, which we call pseudolabel correction and coverage expansion"},"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":"2405.16043","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-25T03:48:12Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"d92f56cec5911c26d5352e609b37b1039bf77999e9f9dbe2547d3e6ecbd34bd5","abstract_canon_sha256":"e6522dcf77be803fc1b7ae698f941da6e0526a7a61e4de9a35a4374b423e6344"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:23:24.804596Z","signature_b64":"He6/COZprDSsloXO16QYaP4ToY8X/N6sHCnvWZwplAK45KUCId8GO7wSCfwxHa7PbsCGeXHztNI40G03WSuQCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7ae2c5268a5a6a14d2cc53e8265bf9f51d5ea9e47e6356b69b0b3508882ad51e","last_reissued_at":"2026-07-05T08:23:24.803970Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:23:24.803970Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Theoretical Analysis of Weak-to-Strong Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aravindan Vijayaraghavan, David Sontag, Hunter Lang","submitted_at":"2024-05-25T03:48:12Z","abstract_excerpt":"Strong student models can learn from weaker teachers: when trained on the predictions of a weaker model, a strong pretrained student can learn to correct the weak model's errors and generalize to examples where the teacher is not confident, even when these examples are excluded from training. This enables learning from cheap, incomplete, and possibly incorrect label information, such as coarse logical rules or the generations of a language model. We show that existing weak supervision theory fails to account for both of these effects, which we call pseudolabel correction and coverage expansion"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.16043","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/2405.16043/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":"2405.16043","created_at":"2026-07-05T08:23:24.804044+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.16043v1","created_at":"2026-07-05T08:23:24.804044+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.16043","created_at":"2026-07-05T08:23:24.804044+00:00"},{"alias_kind":"pith_short_12","alias_value":"PLRMKJUKLJVB","created_at":"2026-07-05T08:23:24.804044+00:00"},{"alias_kind":"pith_short_16","alias_value":"PLRMKJUKLJVBJUWM","created_at":"2026-07-05T08:23:24.804044+00:00"},{"alias_kind":"pith_short_8","alias_value":"PLRMKJUK","created_at":"2026-07-05T08:23:24.804044+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2502.05075","citing_title":"Discrepancies are Virtue: Weak-to-Strong Generalization through Lens of Intrinsic Dimension","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U","json":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U.json","graph_json":"https://pith.science/api/pith-number/PLRMKJUKLJVBJUWMKPUCMW7Z6U/graph.json","events_json":"https://pith.science/api/pith-number/PLRMKJUKLJVBJUWMKPUCMW7Z6U/events.json","paper":"https://pith.science/paper/PLRMKJUK"},"agent_actions":{"view_html":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U","download_json":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U.json","view_paper":"https://pith.science/paper/PLRMKJUK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.16043&json=true","fetch_graph":"https://pith.science/api/pith-number/PLRMKJUKLJVBJUWMKPUCMW7Z6U/graph.json","fetch_events":"https://pith.science/api/pith-number/PLRMKJUKLJVBJUWMKPUCMW7Z6U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U/action/storage_attestation","attest_author":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U/action/author_attestation","sign_citation":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U/action/citation_signature","submit_replication":"https://pith.science/pith/PLRMKJUKLJVBJUWMKPUCMW7Z6U/action/replication_record"}},"created_at":"2026-07-05T08:23:24.804044+00:00","updated_at":"2026-07-05T08:23:24.804044+00:00"}