{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IKDUZI2LHKELS5MJ3XDCACEL5H","short_pith_number":"pith:IKDUZI2L","schema_version":"1.0","canonical_sha256":"42874ca34b3a88b97589ddc620088be9d6d7920809784a1157858c378d645fa0","source":{"kind":"arxiv","id":"2412.04305","version":1},"attestation_state":"computed","paper":{"title":"ALMA: Alignment with Minimal Annotation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Andrew Cohen, Chunting Zhou, Jason Weston, Leonid Shamis, Luke Zettlemoyer, Marjan Ghazvininejad, Michihiro Yasunaga","submitted_at":"2024-12-05T16:26:31Z","abstract_excerpt":"Recent approaches to large language model (LLM) alignment typically require millions of human annotations or rely on external aligned models for synthetic data generation. This paper introduces ALMA: Alignment with Minimal Annotation, demonstrating that effective alignment can be achieved using only 9,000 labeled examples -- less than 1% of conventional approaches. ALMA generates large amounts of high-quality synthetic alignment data through new techniques: diverse prompt synthesis via few-shot learning, diverse response generation with multiple model checkpoints, and judge (reward model) enha"},"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":"2412.04305","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-05T16:26:31Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e216b33b6660bb9b5fe9b11a6b21c764e96b6d3b1569c86e03b5c20d53942783","abstract_canon_sha256":"ec3e6177bbc3add6f239ee993403add904c344910af29a95d2e00e7139b5d992"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:45:02.235956Z","signature_b64":"F6HSd5SRtUex6Nk/QsLVef6hb9PadvpYEul2QO4Dw6Uf9aBhofTnUO0IvvkjLpnIqD+mIqb7IS/0AL2d4yIVAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"42874ca34b3a88b97589ddc620088be9d6d7920809784a1157858c378d645fa0","last_reissued_at":"2026-07-05T09:45:02.235473Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:45:02.235473Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ALMA: Alignment with Minimal Annotation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Andrew Cohen, Chunting Zhou, Jason Weston, Leonid Shamis, Luke Zettlemoyer, Marjan Ghazvininejad, Michihiro Yasunaga","submitted_at":"2024-12-05T16:26:31Z","abstract_excerpt":"Recent approaches to large language model (LLM) alignment typically require millions of human annotations or rely on external aligned models for synthetic data generation. This paper introduces ALMA: Alignment with Minimal Annotation, demonstrating that effective alignment can be achieved using only 9,000 labeled examples -- less than 1% of conventional approaches. ALMA generates large amounts of high-quality synthetic alignment data through new techniques: diverse prompt synthesis via few-shot learning, diverse response generation with multiple model checkpoints, and judge (reward model) enha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.04305","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/2412.04305/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":"2412.04305","created_at":"2026-07-05T09:45:02.235526+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.04305v1","created_at":"2026-07-05T09:45:02.235526+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.04305","created_at":"2026-07-05T09:45:02.235526+00:00"},{"alias_kind":"pith_short_12","alias_value":"IKDUZI2LHKEL","created_at":"2026-07-05T09:45:02.235526+00:00"},{"alias_kind":"pith_short_16","alias_value":"IKDUZI2LHKELS5MJ","created_at":"2026-07-05T09:45:02.235526+00:00"},{"alias_kind":"pith_short_8","alias_value":"IKDUZI2L","created_at":"2026-07-05T09:45:02.235526+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.21342","citing_title":"PEDANTIC: A Dataset for the Automatic Examination of Definiteness in Patent Claims","ref_index":32,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H","json":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H.json","graph_json":"https://pith.science/api/pith-number/IKDUZI2LHKELS5MJ3XDCACEL5H/graph.json","events_json":"https://pith.science/api/pith-number/IKDUZI2LHKELS5MJ3XDCACEL5H/events.json","paper":"https://pith.science/paper/IKDUZI2L"},"agent_actions":{"view_html":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H","download_json":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H.json","view_paper":"https://pith.science/paper/IKDUZI2L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.04305&json=true","fetch_graph":"https://pith.science/api/pith-number/IKDUZI2LHKELS5MJ3XDCACEL5H/graph.json","fetch_events":"https://pith.science/api/pith-number/IKDUZI2LHKELS5MJ3XDCACEL5H/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H/action/storage_attestation","attest_author":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H/action/author_attestation","sign_citation":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H/action/citation_signature","submit_replication":"https://pith.science/pith/IKDUZI2LHKELS5MJ3XDCACEL5H/action/replication_record"}},"created_at":"2026-07-05T09:45:02.235526+00:00","updated_at":"2026-07-05T09:45:02.235526+00:00"}