{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:UICX466VDJVQDIKKVCDWUBS7CZ","short_pith_number":"pith:UICX466V","schema_version":"1.0","canonical_sha256":"a2057e7bd51a6b01a14aa8876a065f166a3628dfc1846c441b77b2c3875ed20f","source":{"kind":"arxiv","id":"2408.10914","version":1},"attestation_state":"computed","paper":{"title":"To Code, or Not To Code? Exploring Impact of Code in Pre-training","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Acyr Locatelli, Adrien Morisot, Ahmet \\\"Ust\\\"un, Ivan Zhang, Marzieh Fadaee, Raymond Ma, Sara Hooker, Viraat Aryabumi, Yixuan Su","submitted_at":"2024-08-20T14:58:13Z","abstract_excerpt":"Including code in the pre-training data mixture, even for models not specifically designed for code, has become a common practice in LLMs pre-training. While there has been anecdotal consensus among practitioners that code data plays a vital role in general LLMs' performance, there is only limited work analyzing the precise impact of code on non-code tasks. In this work, we systematically investigate the impact of code data on general performance. We ask \"what is the impact of code data used in pre-training on a large variety of downstream tasks beyond code generation\". We conduct extensive ab"},"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":"2408.10914","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-08-20T14:58:13Z","cross_cats_sorted":[],"title_canon_sha256":"0a50073349e04262d310ab81b8cae81e3376130d63cc21982124a38d125782e4","abstract_canon_sha256":"220ac55963eb1ce448f417274e5b7b57e93a03711c72dd2e15d149cc0595aafd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:57:19.619883Z","signature_b64":"W9UKLmnFATLpFX5+9zYFF/mGDOKPDgygbiVkZV1fqtjS+VAmyswA6iG8Rc7N273HbKXK2zvabWGCdjEZyN8lAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a2057e7bd51a6b01a14aa8876a065f166a3628dfc1846c441b77b2c3875ed20f","last_reissued_at":"2026-07-05T08:57:19.619489Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:57:19.619489Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"To Code, or Not To Code? Exploring Impact of Code in Pre-training","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Acyr Locatelli, Adrien Morisot, Ahmet \\\"Ust\\\"un, Ivan Zhang, Marzieh Fadaee, Raymond Ma, Sara Hooker, Viraat Aryabumi, Yixuan Su","submitted_at":"2024-08-20T14:58:13Z","abstract_excerpt":"Including code in the pre-training data mixture, even for models not specifically designed for code, has become a common practice in LLMs pre-training. While there has been anecdotal consensus among practitioners that code data plays a vital role in general LLMs' performance, there is only limited work analyzing the precise impact of code on non-code tasks. In this work, we systematically investigate the impact of code data on general performance. We ask \"what is the impact of code data used in pre-training on a large variety of downstream tasks beyond code generation\". We conduct extensive ab"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.10914","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/2408.10914/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":"2408.10914","created_at":"2026-07-05T08:57:19.619542+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.10914v1","created_at":"2026-07-05T08:57:19.619542+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.10914","created_at":"2026-07-05T08:57:19.619542+00:00"},{"alias_kind":"pith_short_12","alias_value":"UICX466VDJVQ","created_at":"2026-07-05T08:57:19.619542+00:00"},{"alias_kind":"pith_short_16","alias_value":"UICX466VDJVQDIKK","created_at":"2026-07-05T08:57:19.619542+00:00"},{"alias_kind":"pith_short_8","alias_value":"UICX466V","created_at":"2026-07-05T08:57:19.619542+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2502.02737","citing_title":"SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model","ref_index":146,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05267","citing_title":"Bridging Generation and Training: A Systematic Review of Quality Issues in LLMs for Code","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13107","citing_title":"Can Coding Agents Be General Agents?","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ","json":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ.json","graph_json":"https://pith.science/api/pith-number/UICX466VDJVQDIKKVCDWUBS7CZ/graph.json","events_json":"https://pith.science/api/pith-number/UICX466VDJVQDIKKVCDWUBS7CZ/events.json","paper":"https://pith.science/paper/UICX466V"},"agent_actions":{"view_html":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ","download_json":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ.json","view_paper":"https://pith.science/paper/UICX466V","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.10914&json=true","fetch_graph":"https://pith.science/api/pith-number/UICX466VDJVQDIKKVCDWUBS7CZ/graph.json","fetch_events":"https://pith.science/api/pith-number/UICX466VDJVQDIKKVCDWUBS7CZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ/action/storage_attestation","attest_author":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ/action/author_attestation","sign_citation":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ/action/citation_signature","submit_replication":"https://pith.science/pith/UICX466VDJVQDIKKVCDWUBS7CZ/action/replication_record"}},"created_at":"2026-07-05T08:57:19.619542+00:00","updated_at":"2026-07-05T08:57:19.619542+00:00"}