{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:5XM6AAKJY3FS7G5IXB3E7VIDW4","short_pith_number":"pith:5XM6AAKJ","schema_version":"1.0","canonical_sha256":"edd9e00149c6cb2f9ba8b8764fd503b732356bd0cbf95f2ffbc1214725c7c597","source":{"kind":"arxiv","id":"2305.13230","version":2},"attestation_state":"computed","paper":{"title":"To Repeat or Not To Repeat: Insights from Scaling LLM under Token-Crisis","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Fuzhao Xue, Wangchunshu Zhou, Yang You, Yao Fu, Zangwei Zheng","submitted_at":"2023-05-22T17:02:15Z","abstract_excerpt":"Recent research has highlighted the importance of dataset size in scaling language models. However, large language models (LLMs) are notoriously token-hungry during pre-training, and high-quality text data on the web is approaching its scaling limit for LLMs. To further enhance LLMs, a straightforward approach is to repeat the pre-training data for additional epochs. In this study, we empirically investigate three key aspects under this approach. First, we explore the consequences of repeating pre-training data, revealing that the model is susceptible to overfitting, leading to multi-epoch deg"},"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":"2305.13230","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-22T17:02:15Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"2a1df183179e3ddb1b8a9a8b1eed80bc3516d0b223fb5f4a86dc630650619463","abstract_canon_sha256":"2a1e3a842b2084ad4c248f112d29e97d4e1f834dd091064fa715ad4248733175"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:58:40.230982Z","signature_b64":"cHilGne5LwrllGXn8Y454I2pD0B1a5uMYYOiDWofAeqoyKpdFLLFSF06F4hV5Sg8olSWy+dqHO1tG4MEovMoAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"edd9e00149c6cb2f9ba8b8764fd503b732356bd0cbf95f2ffbc1214725c7c597","last_reissued_at":"2026-07-05T06:58:40.230496Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:58:40.230496Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"To Repeat or Not To Repeat: Insights from Scaling LLM under Token-Crisis","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Fuzhao Xue, Wangchunshu Zhou, Yang You, Yao Fu, Zangwei Zheng","submitted_at":"2023-05-22T17:02:15Z","abstract_excerpt":"Recent research has highlighted the importance of dataset size in scaling language models. However, large language models (LLMs) are notoriously token-hungry during pre-training, and high-quality text data on the web is approaching its scaling limit for LLMs. To further enhance LLMs, a straightforward approach is to repeat the pre-training data for additional epochs. In this study, we empirically investigate three key aspects under this approach. First, we explore the consequences of repeating pre-training data, revealing that the model is susceptible to overfitting, leading to multi-epoch deg"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.13230","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/2305.13230/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":"2305.13230","created_at":"2026-07-05T06:58:40.230558+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.13230v2","created_at":"2026-07-05T06:58:40.230558+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.13230","created_at":"2026-07-05T06:58:40.230558+00:00"},{"alias_kind":"pith_short_12","alias_value":"5XM6AAKJY3FS","created_at":"2026-07-05T06:58:40.230558+00:00"},{"alias_kind":"pith_short_16","alias_value":"5XM6AAKJY3FS7G5I","created_at":"2026-07-05T06:58:40.230558+00:00"},{"alias_kind":"pith_short_8","alias_value":"5XM6AAKJ","created_at":"2026-07-05T06:58:40.230558+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2506.12119","citing_title":"Mixture-of-Experts Can Surpass Dense LLMs Under Strictly Equal Resource","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2311.16867","citing_title":"The Falcon Series of Open Language Models","ref_index":138,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4","json":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4.json","graph_json":"https://pith.science/api/pith-number/5XM6AAKJY3FS7G5IXB3E7VIDW4/graph.json","events_json":"https://pith.science/api/pith-number/5XM6AAKJY3FS7G5IXB3E7VIDW4/events.json","paper":"https://pith.science/paper/5XM6AAKJ"},"agent_actions":{"view_html":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4","download_json":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4.json","view_paper":"https://pith.science/paper/5XM6AAKJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.13230&json=true","fetch_graph":"https://pith.science/api/pith-number/5XM6AAKJY3FS7G5IXB3E7VIDW4/graph.json","fetch_events":"https://pith.science/api/pith-number/5XM6AAKJY3FS7G5IXB3E7VIDW4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4/action/storage_attestation","attest_author":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4/action/author_attestation","sign_citation":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4/action/citation_signature","submit_replication":"https://pith.science/pith/5XM6AAKJY3FS7G5IXB3E7VIDW4/action/replication_record"}},"created_at":"2026-07-05T06:58:40.230558+00:00","updated_at":"2026-07-05T06:58:40.230558+00:00"}