{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:KSR3P6I7VZCADRLLTSYLO5TDEV","short_pith_number":"pith:KSR3P6I7","schema_version":"1.0","canonical_sha256":"54a3b7f91fae4401c56b9cb0b77663256dfe4cba7f915eb9f5381c34dde0fa71","source":{"kind":"arxiv","id":"2410.04571","version":3},"attestation_state":"computed","paper":{"title":"EnsemW2S: Enhancing Weak-to-Strong Generalization with Large Language Model Ensembles","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aakriti Agrawal, Anirudh Satheesh, Bang An, Bayan Bruss, Chenghao Deng, Furong Huang, John Langford, Mucong Ding, Zora Che","submitted_at":"2024-10-06T18:06:42Z","abstract_excerpt":"With Large Language Models (LLMs) rapidly approaching and potentially surpassing human-level performance, it has become imperative to develop approaches capable of effectively supervising and enhancing these powerful models using smaller, human-level models exposed to only human-level data. We address this critical weak-to-strong (W2S) generalization challenge by proposing a novel method aimed at improving weak experts, by training on the same limited human-level data, enabling them to generalize to complex, super-human-level tasks. Our approach, called **EnsemW2S**, employs a token-level ense"},"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":"2410.04571","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-06T18:06:42Z","cross_cats_sorted":[],"title_canon_sha256":"2f6cdf60d954d6ca79275d90be7da07f3cd83935bd59144cfbf3f76e7560accb","abstract_canon_sha256":"68fc88c2adb4a6698b80d606998bff46ba4e9e43b12b9c36576124767cdafddc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:41:35.830856Z","signature_b64":"IVk1PPhiG5nGolIHEFXzM6IMnyef1TY+F4rFQaUPC3SRNOMnxQ6BCsULRin5Ai21Xi+xiiTTqAXHGW0wzvrVCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"54a3b7f91fae4401c56b9cb0b77663256dfe4cba7f915eb9f5381c34dde0fa71","last_reissued_at":"2026-07-05T11:41:35.830360Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:41:35.830360Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"EnsemW2S: Enhancing Weak-to-Strong Generalization with Large Language Model Ensembles","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aakriti Agrawal, Anirudh Satheesh, Bang An, Bayan Bruss, Chenghao Deng, Furong Huang, John Langford, Mucong Ding, Zora Che","submitted_at":"2024-10-06T18:06:42Z","abstract_excerpt":"With Large Language Models (LLMs) rapidly approaching and potentially surpassing human-level performance, it has become imperative to develop approaches capable of effectively supervising and enhancing these powerful models using smaller, human-level models exposed to only human-level data. We address this critical weak-to-strong (W2S) generalization challenge by proposing a novel method aimed at improving weak experts, by training on the same limited human-level data, enabling them to generalize to complex, super-human-level tasks. Our approach, called **EnsemW2S**, employs a token-level ense"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.04571","kind":"arxiv","version":3},"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/2410.04571/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":"2410.04571","created_at":"2026-07-05T11:41:35.830415+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.04571v3","created_at":"2026-07-05T11:41:35.830415+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.04571","created_at":"2026-07-05T11:41:35.830415+00:00"},{"alias_kind":"pith_short_12","alias_value":"KSR3P6I7VZCA","created_at":"2026-07-05T11:41:35.830415+00:00"},{"alias_kind":"pith_short_16","alias_value":"KSR3P6I7VZCADRLL","created_at":"2026-07-05T11:41:35.830415+00:00"},{"alias_kind":"pith_short_8","alias_value":"KSR3P6I7","created_at":"2026-07-05T11:41:35.830415+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.05710","citing_title":"On the Blessing of Pre-training in Weak-to-Strong Generalization","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV","json":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV.json","graph_json":"https://pith.science/api/pith-number/KSR3P6I7VZCADRLLTSYLO5TDEV/graph.json","events_json":"https://pith.science/api/pith-number/KSR3P6I7VZCADRLLTSYLO5TDEV/events.json","paper":"https://pith.science/paper/KSR3P6I7"},"agent_actions":{"view_html":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV","download_json":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV.json","view_paper":"https://pith.science/paper/KSR3P6I7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.04571&json=true","fetch_graph":"https://pith.science/api/pith-number/KSR3P6I7VZCADRLLTSYLO5TDEV/graph.json","fetch_events":"https://pith.science/api/pith-number/KSR3P6I7VZCADRLLTSYLO5TDEV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV/action/storage_attestation","attest_author":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV/action/author_attestation","sign_citation":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV/action/citation_signature","submit_replication":"https://pith.science/pith/KSR3P6I7VZCADRLLTSYLO5TDEV/action/replication_record"}},"created_at":"2026-07-05T11:41:35.830415+00:00","updated_at":"2026-07-05T11:41:35.830415+00:00"}