{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:RE53LYEVTM4KVL7XP2TF2UFOEM","short_pith_number":"pith:RE53LYEV","schema_version":"1.0","canonical_sha256":"893bb5e0959b38aaaff77ea65d50ae23186ad63b076a6811d44a919a5c6f5596","source":{"kind":"arxiv","id":"2409.09613","version":1},"attestation_state":"computed","paper":{"title":"Rethinking KenLM: Good and Bad Model Ensembles for Efficient Text Quality Filtering in Large Web Corpora","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chanjun Park, Hyunsoo Ha, Jihoo Kim, Seonghoon Yang, Sukyung Lee, Yungi Kim","submitted_at":"2024-09-15T05:27:56Z","abstract_excerpt":"With the increasing demand for substantial amounts of high-quality data to train large language models (LLMs), efficiently filtering large web corpora has become a critical challenge. For this purpose, KenLM, a lightweight n-gram-based language model that operates on CPUs, is widely used. However, the traditional method of training KenLM utilizes only high-quality data and, consequently, does not explicitly learn the linguistic patterns of low-quality data. To address this issue, we propose an ensemble approach that leverages two contrasting KenLMs: (i) Good KenLM, trained on high-quality data"},"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":"2409.09613","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-09-15T05:27:56Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"00c909c6e8fd10133257f766d6ac6934b29213e443091343d6c4ab9b1e2308e3","abstract_canon_sha256":"b179fe23828c212d1b29e60c232988f1d3ce1a8644594db3b698f6d65521a595"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:07:12.084842Z","signature_b64":"6Ctjg2KekjaGK3Ku777rPmu/Su1ob7+NaYqXTSaqF9IA2CgWKhv0rpAx0w4vAD2LQaJ8KhL1qmL6Sr+F77BMAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"893bb5e0959b38aaaff77ea65d50ae23186ad63b076a6811d44a919a5c6f5596","last_reissued_at":"2026-07-05T09:07:12.084358Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:07:12.084358Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rethinking KenLM: Good and Bad Model Ensembles for Efficient Text Quality Filtering in Large Web Corpora","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chanjun Park, Hyunsoo Ha, Jihoo Kim, Seonghoon Yang, Sukyung Lee, Yungi Kim","submitted_at":"2024-09-15T05:27:56Z","abstract_excerpt":"With the increasing demand for substantial amounts of high-quality data to train large language models (LLMs), efficiently filtering large web corpora has become a critical challenge. For this purpose, KenLM, a lightweight n-gram-based language model that operates on CPUs, is widely used. However, the traditional method of training KenLM utilizes only high-quality data and, consequently, does not explicitly learn the linguistic patterns of low-quality data. To address this issue, we propose an ensemble approach that leverages two contrasting KenLMs: (i) Good KenLM, trained on high-quality data"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.09613","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/2409.09613/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":"2409.09613","created_at":"2026-07-05T09:07:12.084419+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.09613v1","created_at":"2026-07-05T09:07:12.084419+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.09613","created_at":"2026-07-05T09:07:12.084419+00:00"},{"alias_kind":"pith_short_12","alias_value":"RE53LYEVTM4K","created_at":"2026-07-05T09:07:12.084419+00:00"},{"alias_kind":"pith_short_16","alias_value":"RE53LYEVTM4KVL7X","created_at":"2026-07-05T09:07:12.084419+00:00"},{"alias_kind":"pith_short_8","alias_value":"RE53LYEV","created_at":"2026-07-05T09:07:12.084419+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.11289","citing_title":"LP Data Pipeline: Lightweight, Purpose-driven Data Pipeline for Large Language Models","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM","json":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM.json","graph_json":"https://pith.science/api/pith-number/RE53LYEVTM4KVL7XP2TF2UFOEM/graph.json","events_json":"https://pith.science/api/pith-number/RE53LYEVTM4KVL7XP2TF2UFOEM/events.json","paper":"https://pith.science/paper/RE53LYEV"},"agent_actions":{"view_html":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM","download_json":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM.json","view_paper":"https://pith.science/paper/RE53LYEV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.09613&json=true","fetch_graph":"https://pith.science/api/pith-number/RE53LYEVTM4KVL7XP2TF2UFOEM/graph.json","fetch_events":"https://pith.science/api/pith-number/RE53LYEVTM4KVL7XP2TF2UFOEM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM/action/storage_attestation","attest_author":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM/action/author_attestation","sign_citation":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM/action/citation_signature","submit_replication":"https://pith.science/pith/RE53LYEVTM4KVL7XP2TF2UFOEM/action/replication_record"}},"created_at":"2026-07-05T09:07:12.084419+00:00","updated_at":"2026-07-05T09:07:12.084419+00:00"}