{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5RGIQLQLZE7GR4HSQX7DY53FAY","short_pith_number":"pith:5RGIQLQL","schema_version":"1.0","canonical_sha256":"ec4c882e0bc93e68f0f285fe3c7765060db46a1b5bed553312f31246cd3e3236","source":{"kind":"arxiv","id":"2407.06542","version":2},"attestation_state":"computed","paper":{"title":"LIONs: An Empirically Optimized Approach to Align Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Qingyang Wu, Xiao Yu, Yu Li, Zhou Yu","submitted_at":"2024-07-09T04:34:39Z","abstract_excerpt":"Alignment is a crucial step to enhance the instruction-following and conversational abilities of language models. Despite many recent work proposing new algorithms, datasets, and training pipelines, there is a lack of comprehensive studies measuring the impact of various design choices throughout the whole training process. We first conduct a rigorous analysis over a three-stage training pipeline consisting of supervised fine-tuning, offline preference learning, and online preference learning. We have found that using techniques like sequence packing, loss masking in SFT, increasing the prefer"},"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":"2407.06542","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T04:34:39Z","cross_cats_sorted":[],"title_canon_sha256":"0c59010e21c1094cbcbd174241208f4afe6557a9ab4fb2c2bd13acc5acd73518","abstract_canon_sha256":"02d8c422788bde74175ca644a1d35a4e862749ec46c353b26f8051e24e004d17"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:14:54.850818Z","signature_b64":"KsKCUTuBOjk4nY7K1OqDVG7yYsbGHCUtuXbsDwLrzSYpW/J8LRY7POJfS/Rho3h8QUUKg+kr0W6StzO2cKoRDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ec4c882e0bc93e68f0f285fe3c7765060db46a1b5bed553312f31246cd3e3236","last_reissued_at":"2026-07-05T09:14:54.850408Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:14:54.850408Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LIONs: An Empirically Optimized Approach to Align Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Qingyang Wu, Xiao Yu, Yu Li, Zhou Yu","submitted_at":"2024-07-09T04:34:39Z","abstract_excerpt":"Alignment is a crucial step to enhance the instruction-following and conversational abilities of language models. Despite many recent work proposing new algorithms, datasets, and training pipelines, there is a lack of comprehensive studies measuring the impact of various design choices throughout the whole training process. We first conduct a rigorous analysis over a three-stage training pipeline consisting of supervised fine-tuning, offline preference learning, and online preference learning. We have found that using techniques like sequence packing, loss masking in SFT, increasing the prefer"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.06542","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/2407.06542/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":"2407.06542","created_at":"2026-07-05T09:14:54.850463+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.06542v2","created_at":"2026-07-05T09:14:54.850463+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.06542","created_at":"2026-07-05T09:14:54.850463+00:00"},{"alias_kind":"pith_short_12","alias_value":"5RGIQLQLZE7G","created_at":"2026-07-05T09:14:54.850463+00:00"},{"alias_kind":"pith_short_16","alias_value":"5RGIQLQLZE7GR4HS","created_at":"2026-07-05T09:14:54.850463+00:00"},{"alias_kind":"pith_short_8","alias_value":"5RGIQLQL","created_at":"2026-07-05T09:14:54.850463+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.21610","citing_title":"SCOPE and SCION: A Benchmark and an Auditable Reference Pipeline for Schema Induction and Fusion from Text","ref_index":15,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY","json":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY.json","graph_json":"https://pith.science/api/pith-number/5RGIQLQLZE7GR4HSQX7DY53FAY/graph.json","events_json":"https://pith.science/api/pith-number/5RGIQLQLZE7GR4HSQX7DY53FAY/events.json","paper":"https://pith.science/paper/5RGIQLQL"},"agent_actions":{"view_html":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY","download_json":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY.json","view_paper":"https://pith.science/paper/5RGIQLQL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.06542&json=true","fetch_graph":"https://pith.science/api/pith-number/5RGIQLQLZE7GR4HSQX7DY53FAY/graph.json","fetch_events":"https://pith.science/api/pith-number/5RGIQLQLZE7GR4HSQX7DY53FAY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY/action/storage_attestation","attest_author":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY/action/author_attestation","sign_citation":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY/action/citation_signature","submit_replication":"https://pith.science/pith/5RGIQLQLZE7GR4HSQX7DY53FAY/action/replication_record"}},"created_at":"2026-07-05T09:14:54.850463+00:00","updated_at":"2026-07-05T09:14:54.850463+00:00"}