{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:JRANJ5GWYB7XNPMQLE3SLXQPPQ","short_pith_number":"pith:JRANJ5GW","schema_version":"1.0","canonical_sha256":"4c40d4f4d6c07f76bd90593725de0f7c0e88df8a47fa198615e77c467b0146de","source":{"kind":"arxiv","id":"2409.04599","version":1},"attestation_state":"computed","paper":{"title":"BPE Gets Picky: Efficient Vocabulary Refinement During Tokenizer Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Catherine Arnett, Elizaveta Korotkova, Ivan P. Yamshchikov, Pavel Chizhov","submitted_at":"2024-09-06T20:12:34Z","abstract_excerpt":"Language models can largely benefit from efficient tokenization. However, they still mostly utilize the classical BPE algorithm, a simple and reliable method. This has been shown to cause such issues as under-trained tokens and sub-optimal compression that may affect the downstream performance. We introduce Picky BPE, a modified BPE algorithm that carries out vocabulary refinement during tokenizer training. Our method improves vocabulary efficiency, eliminates under-trained tokens, and does not compromise text compression. Our experiments show that our method does not reduce the downstream per"},"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":"2409.04599","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-09-06T20:12:34Z","cross_cats_sorted":[],"title_canon_sha256":"1aaf8f9475b9f94caaba630567e6bdab4b07d57d807bef0945d73d64a34db1f7","abstract_canon_sha256":"ccaf65be5eef0205d8c1f05f1ff5c1cd324209ad675a555b4470308eb6c3b59e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:04:15.010873Z","signature_b64":"367+/n3Ovjtgo12jmLl1YS3StHavoQeEWFXTCmRv9yZ3I/TJu5R2jO66gOKxk8FWlz9d8P1PA3EhanSrvEGCCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4c40d4f4d6c07f76bd90593725de0f7c0e88df8a47fa198615e77c467b0146de","last_reissued_at":"2026-07-05T09:04:15.010466Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:04:15.010466Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BPE Gets Picky: Efficient Vocabulary Refinement During Tokenizer Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Catherine Arnett, Elizaveta Korotkova, Ivan P. Yamshchikov, Pavel Chizhov","submitted_at":"2024-09-06T20:12:34Z","abstract_excerpt":"Language models can largely benefit from efficient tokenization. However, they still mostly utilize the classical BPE algorithm, a simple and reliable method. This has been shown to cause such issues as under-trained tokens and sub-optimal compression that may affect the downstream performance. We introduce Picky BPE, a modified BPE algorithm that carries out vocabulary refinement during tokenizer training. Our method improves vocabulary efficiency, eliminates under-trained tokens, and does not compromise text compression. Our experiments show that our method does not reduce the downstream per"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.04599","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/2409.04599/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":"2409.04599","created_at":"2026-07-05T09:04:15.010521+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.04599v1","created_at":"2026-07-05T09:04:15.010521+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.04599","created_at":"2026-07-05T09:04:15.010521+00:00"},{"alias_kind":"pith_short_12","alias_value":"JRANJ5GWYB7X","created_at":"2026-07-05T09:04:15.010521+00:00"},{"alias_kind":"pith_short_16","alias_value":"JRANJ5GWYB7XNPMQ","created_at":"2026-07-05T09:04:15.010521+00:00"},{"alias_kind":"pith_short_8","alias_value":"JRANJ5GW","created_at":"2026-07-05T09:04:15.010521+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2506.14123","citing_title":"Sampling from Your Language Model One Byte at a Time","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ","json":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ.json","graph_json":"https://pith.science/api/pith-number/JRANJ5GWYB7XNPMQLE3SLXQPPQ/graph.json","events_json":"https://pith.science/api/pith-number/JRANJ5GWYB7XNPMQLE3SLXQPPQ/events.json","paper":"https://pith.science/paper/JRANJ5GW"},"agent_actions":{"view_html":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ","download_json":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ.json","view_paper":"https://pith.science/paper/JRANJ5GW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.04599&json=true","fetch_graph":"https://pith.science/api/pith-number/JRANJ5GWYB7XNPMQLE3SLXQPPQ/graph.json","fetch_events":"https://pith.science/api/pith-number/JRANJ5GWYB7XNPMQLE3SLXQPPQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ/action/storage_attestation","attest_author":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ/action/author_attestation","sign_citation":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ/action/citation_signature","submit_replication":"https://pith.science/pith/JRANJ5GWYB7XNPMQLE3SLXQPPQ/action/replication_record"}},"created_at":"2026-07-05T09:04:15.010521+00:00","updated_at":"2026-07-05T09:04:15.010521+00:00"}