{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:LTY76VGEJ67O7M5BU2ZRCTPAGY","short_pith_number":"pith:LTY76VGE","schema_version":"1.0","canonical_sha256":"5cf1ff54c44fbeefb3a1a6b3114de03615ea4f4df1244c0cb12c17b8a638e79f","source":{"kind":"arxiv","id":"2408.06663","version":5},"attestation_state":"computed","paper":{"title":"Amuro and Char: Analyzing the Relationship between Pre-Training and Fine-Tuning of Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Kaiser Sun, Mark Dredze","submitted_at":"2024-08-13T06:28:43Z","abstract_excerpt":"The development of large language models leads to the formation of a pre-train-then-align paradigm, in which the model is typically pre-trained on a large text corpus and undergoes a tuning stage to align the model with human preference or downstream tasks. In this work, we investigate the relationship between pre-training and fine-tuning by fine-tuning multiple intermediate pre-trained model checkpoints. Our results on 18 datasets suggest that i) continual pre-training improves the model in a latent way that unveils after fine-tuning; ii) with extra fine-tuning, the datasets that the model do"},"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":"2408.06663","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-08-13T06:28:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"42be6d6acebfa9bdcb9335d9edf73c17b7016817c5988b28dee2baf98808677b","abstract_canon_sha256":"768b3771a7d75f4151064d0e17824082d24d0920c7cc79f898ce6aa3f088f117"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:33:36.167035Z","signature_b64":"Js5HDBi7L1OgzxdA7vh3tOxNL13VI6gb/QHcemVWHfuD28FYTZw+3B9WQKCEZss6WBzRSxxqg+Eov22UA/kBDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5cf1ff54c44fbeefb3a1a6b3114de03615ea4f4df1244c0cb12c17b8a638e79f","last_reissued_at":"2026-07-05T10:33:36.166548Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:33:36.166548Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Amuro and Char: Analyzing the Relationship between Pre-Training and Fine-Tuning of Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Kaiser Sun, Mark Dredze","submitted_at":"2024-08-13T06:28:43Z","abstract_excerpt":"The development of large language models leads to the formation of a pre-train-then-align paradigm, in which the model is typically pre-trained on a large text corpus and undergoes a tuning stage to align the model with human preference or downstream tasks. In this work, we investigate the relationship between pre-training and fine-tuning by fine-tuning multiple intermediate pre-trained model checkpoints. Our results on 18 datasets suggest that i) continual pre-training improves the model in a latent way that unveils after fine-tuning; ii) with extra fine-tuning, the datasets that the model do"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.06663","kind":"arxiv","version":5},"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/2408.06663/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":"2408.06663","created_at":"2026-07-05T10:33:36.166607+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.06663v5","created_at":"2026-07-05T10:33:36.166607+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.06663","created_at":"2026-07-05T10:33:36.166607+00:00"},{"alias_kind":"pith_short_12","alias_value":"LTY76VGEJ67O","created_at":"2026-07-05T10:33:36.166607+00:00"},{"alias_kind":"pith_short_16","alias_value":"LTY76VGEJ67O7M5B","created_at":"2026-07-05T10:33:36.166607+00:00"},{"alias_kind":"pith_short_8","alias_value":"LTY76VGE","created_at":"2026-07-05T10:33:36.166607+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11635","citing_title":"Are LLMs Bad at Moral Reasoning?","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY","json":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY.json","graph_json":"https://pith.science/api/pith-number/LTY76VGEJ67O7M5BU2ZRCTPAGY/graph.json","events_json":"https://pith.science/api/pith-number/LTY76VGEJ67O7M5BU2ZRCTPAGY/events.json","paper":"https://pith.science/paper/LTY76VGE"},"agent_actions":{"view_html":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY","download_json":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY.json","view_paper":"https://pith.science/paper/LTY76VGE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.06663&json=true","fetch_graph":"https://pith.science/api/pith-number/LTY76VGEJ67O7M5BU2ZRCTPAGY/graph.json","fetch_events":"https://pith.science/api/pith-number/LTY76VGEJ67O7M5BU2ZRCTPAGY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY/action/storage_attestation","attest_author":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY/action/author_attestation","sign_citation":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY/action/citation_signature","submit_replication":"https://pith.science/pith/LTY76VGEJ67O7M5BU2ZRCTPAGY/action/replication_record"}},"created_at":"2026-07-05T10:33:36.166607+00:00","updated_at":"2026-07-05T10:33:36.166607+00:00"}