{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:KA5534IKF7GIBQXVXX2ZGYPMNU","short_pith_number":"pith:KA5534IK","schema_version":"1.0","canonical_sha256":"503bddf10a2fcc80c2f5bdf59361ec6d117c5b54a2fec028c68ed75558469aa4","source":{"kind":"arxiv","id":"2402.17759","version":2},"attestation_state":"computed","paper":{"title":"Towards Optimal Learning of Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Furu Wei, Li Dong, Minlie Huang, Qingxiu Dong, Yaru Hao, Yuxian Gu","submitted_at":"2024-02-27T18:52:19Z","abstract_excerpt":"This work studies the general principles of improving the learning of language models (LMs), which aims at reducing the necessary training steps for achieving superior performance. Specifically, we present a theory for the optimal learning of LMs. We first propose an objective that optimizes LM learning by maximizing the data compression ratio in an \"LM-training-as-lossless-compression\" view. Then, we derive a theorem, named Learning Law, to reveal the properties of the dynamics in the optimal learning process under our objective. The theorem is then validated by experiments on a linear classi"},"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":"2402.17759","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-27T18:52:19Z","cross_cats_sorted":[],"title_canon_sha256":"b3c1718af27f0745169270d2aab5a243abf76e132ecd9d83615080e2290d54bc","abstract_canon_sha256":"64e430a80813164801533edb1cdab0fde3ae628308338a9b4e51c93962794095"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:51:40.261176Z","signature_b64":"tyVpacrK6/VGusHdPURqvPQchFJRVuIowaSVDy36Ql+8mIZyUdL3HnBqYMb+DQZosjrpDinGRGQevPx4FH1wAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"503bddf10a2fcc80c2f5bdf59361ec6d117c5b54a2fec028c68ed75558469aa4","last_reissued_at":"2026-07-05T07:51:40.260737Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:51:40.260737Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Optimal Learning of Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Furu Wei, Li Dong, Minlie Huang, Qingxiu Dong, Yaru Hao, Yuxian Gu","submitted_at":"2024-02-27T18:52:19Z","abstract_excerpt":"This work studies the general principles of improving the learning of language models (LMs), which aims at reducing the necessary training steps for achieving superior performance. Specifically, we present a theory for the optimal learning of LMs. We first propose an objective that optimizes LM learning by maximizing the data compression ratio in an \"LM-training-as-lossless-compression\" view. Then, we derive a theorem, named Learning Law, to reveal the properties of the dynamics in the optimal learning process under our objective. The theorem is then validated by experiments on a linear classi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.17759","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/2402.17759/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":"2402.17759","created_at":"2026-07-05T07:51:40.260794+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.17759v2","created_at":"2026-07-05T07:51:40.260794+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.17759","created_at":"2026-07-05T07:51:40.260794+00:00"},{"alias_kind":"pith_short_12","alias_value":"KA5534IKF7GI","created_at":"2026-07-05T07:51:40.260794+00:00"},{"alias_kind":"pith_short_16","alias_value":"KA5534IKF7GIBQXV","created_at":"2026-07-05T07:51:40.260794+00:00"},{"alias_kind":"pith_short_8","alias_value":"KA5534IK","created_at":"2026-07-05T07:51:40.260794+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2406.11354","citing_title":"Preserving Knowledge in Large Language Model with Model-Agnostic Self-Decompression","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU","json":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU.json","graph_json":"https://pith.science/api/pith-number/KA5534IKF7GIBQXVXX2ZGYPMNU/graph.json","events_json":"https://pith.science/api/pith-number/KA5534IKF7GIBQXVXX2ZGYPMNU/events.json","paper":"https://pith.science/paper/KA5534IK"},"agent_actions":{"view_html":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU","download_json":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU.json","view_paper":"https://pith.science/paper/KA5534IK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.17759&json=true","fetch_graph":"https://pith.science/api/pith-number/KA5534IKF7GIBQXVXX2ZGYPMNU/graph.json","fetch_events":"https://pith.science/api/pith-number/KA5534IKF7GIBQXVXX2ZGYPMNU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU/action/storage_attestation","attest_author":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU/action/author_attestation","sign_citation":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU/action/citation_signature","submit_replication":"https://pith.science/pith/KA5534IKF7GIBQXVXX2ZGYPMNU/action/replication_record"}},"created_at":"2026-07-05T07:51:40.260794+00:00","updated_at":"2026-07-05T07:51:40.260794+00:00"}