{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:3B5LZOX3YMF3XYOH2SUG7SKR3K","short_pith_number":"pith:3B5LZOX3","canonical_record":{"source":{"id":"2402.01035","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-01T21:49:34Z","cross_cats_sorted":[],"title_canon_sha256":"50588440ad5c3cdb76b918a1bd41a8269875996a33f18cbc4ab97a2d208f4794","abstract_canon_sha256":"0ef641430f2671ebed077ae4d72430ed950cd5d95eb6f35ca9910924f0dfd452"},"schema_version":"1.0"},"canonical_sha256":"d87abcbafbc30bbbe1c7d4a86fc951dab6f6dab6bb51d5b95c408135988554a7","source":{"kind":"arxiv","id":"2402.01035","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.01035","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"arxiv_version","alias_value":"2402.01035v2","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.01035","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"pith_short_12","alias_value":"3B5LZOX3YMF3","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"pith_short_16","alias_value":"3B5LZOX3YMF3XYOH","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"pith_short_8","alias_value":"3B5LZOX3","created_at":"2026-07-05T07:42:23Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:3B5LZOX3YMF3XYOH2SUG7SKR3K","target":"record","payload":{"canonical_record":{"source":{"id":"2402.01035","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-01T21:49:34Z","cross_cats_sorted":[],"title_canon_sha256":"50588440ad5c3cdb76b918a1bd41a8269875996a33f18cbc4ab97a2d208f4794","abstract_canon_sha256":"0ef641430f2671ebed077ae4d72430ed950cd5d95eb6f35ca9910924f0dfd452"},"schema_version":"1.0"},"canonical_sha256":"d87abcbafbc30bbbe1c7d4a86fc951dab6f6dab6bb51d5b95c408135988554a7","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:42:23.657281Z","signature_b64":"6zTBn61RYGocK5L+P5V1gwrMmTMons/Msq2TPFB7WFDM0ODr0fHFyI8TddR9GNFLOva4i3tdlojme5pwhAwbDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d87abcbafbc30bbbe1c7d4a86fc951dab6f6dab6bb51d5b95c408135988554a7","last_reissued_at":"2026-07-05T07:42:23.656843Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:42:23.656843Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2402.01035","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:42:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"h0cL8Vu7OHEzyAu9hFXPl4JHrrqyG2abFqZLc932GFY1Y9+Ka9HurDJ0yAmg5VcM32dhEffMH1hQEe1PDVRfBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-11T03:13:37.327823Z"},"content_sha256":"261c29e9cf9fbc8fd1af28298256394df033642662f9ea37fca9a74bd58d0c7c","schema_version":"1.0","event_id":"sha256:261c29e9cf9fbc8fd1af28298256394df033642662f9ea37fca9a74bd58d0c7c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:3B5LZOX3YMF3XYOH2SUG7SKR3K","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Getting the most out of your tokenizer for pre-training and domain adaptation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Baptiste Rozi\\`ere, Gabriel Synnaeve, Gautier Dagan","submitted_at":"2024-02-01T21:49:34Z","abstract_excerpt":"Tokenization is an understudied and often neglected component of modern LLMs. Most published works use a single tokenizer for all experiments, often borrowed from another model, without performing ablations or analysis to optimize tokenization. Moreover, the tokenizer is generally kept unchanged when fine-tuning a base model. In this paper, we show that the size, pre-tokenization regular expression, and training data of a tokenizer can significantly impact the model's generation speed, effective context size, memory usage, and downstream performance. We train specialized Byte-Pair Encoding cod"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.01035","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/2402.01035/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:42:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"EW0UiFiDiqZNtCc4WVw639Wkbch7MOpC+HXVYkgJJ9y1BRBv8oyiYgXhv6ttlosCVgnc2F/h3t7X0OBqizxGDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-11T03:13:37.328403Z"},"content_sha256":"6c0c2e2e6f7844bd3579e5e2c3f0fae5e15631535eabc1018f58691acf0f9413","schema_version":"1.0","event_id":"sha256:6c0c2e2e6f7844bd3579e5e2c3f0fae5e15631535eabc1018f58691acf0f9413"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3B5LZOX3YMF3XYOH2SUG7SKR3K/bundle.json","state_url":"https://pith.science/pith/3B5LZOX3YMF3XYOH2SUG7SKR3K/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3B5LZOX3YMF3XYOH2SUG7SKR3K/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-11T03:13:37Z","links":{"resolver":"https://pith.science/pith/3B5LZOX3YMF3XYOH2SUG7SKR3K","bundle":"https://pith.science/pith/3B5LZOX3YMF3XYOH2SUG7SKR3K/bundle.json","state":"https://pith.science/pith/3B5LZOX3YMF3XYOH2SUG7SKR3K/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3B5LZOX3YMF3XYOH2SUG7SKR3K/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:3B5LZOX3YMF3XYOH2SUG7SKR3K","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"0ef641430f2671ebed077ae4d72430ed950cd5d95eb6f35ca9910924f0dfd452","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-01T21:49:34Z","title_canon_sha256":"50588440ad5c3cdb76b918a1bd41a8269875996a33f18cbc4ab97a2d208f4794"},"schema_version":"1.0","source":{"id":"2402.01035","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.01035","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"arxiv_version","alias_value":"2402.01035v2","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.01035","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"pith_short_12","alias_value":"3B5LZOX3YMF3","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"pith_short_16","alias_value":"3B5LZOX3YMF3XYOH","created_at":"2026-07-05T07:42:23Z"},{"alias_kind":"pith_short_8","alias_value":"3B5LZOX3","created_at":"2026-07-05T07:42:23Z"}],"graph_snapshots":[{"event_id":"sha256:6c0c2e2e6f7844bd3579e5e2c3f0fae5e15631535eabc1018f58691acf0f9413","target":"graph","created_at":"2026-07-05T07:42:23Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2402.01035/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Tokenization is an understudied and often neglected component of modern LLMs. Most published works use a single tokenizer for all experiments, often borrowed from another model, without performing ablations or analysis to optimize tokenization. Moreover, the tokenizer is generally kept unchanged when fine-tuning a base model. In this paper, we show that the size, pre-tokenization regular expression, and training data of a tokenizer can significantly impact the model's generation speed, effective context size, memory usage, and downstream performance. We train specialized Byte-Pair Encoding cod","authors_text":"Baptiste Rozi\\`ere, Gabriel Synnaeve, Gautier Dagan","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-01T21:49:34Z","title":"Getting the most out of your tokenizer for pre-training and domain adaptation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.01035","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:261c29e9cf9fbc8fd1af28298256394df033642662f9ea37fca9a74bd58d0c7c","target":"record","created_at":"2026-07-05T07:42:23Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"0ef641430f2671ebed077ae4d72430ed950cd5d95eb6f35ca9910924f0dfd452","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-01T21:49:34Z","title_canon_sha256":"50588440ad5c3cdb76b918a1bd41a8269875996a33f18cbc4ab97a2d208f4794"},"schema_version":"1.0","source":{"id":"2402.01035","kind":"arxiv","version":2}},"canonical_sha256":"d87abcbafbc30bbbe1c7d4a86fc951dab6f6dab6bb51d5b95c408135988554a7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d87abcbafbc30bbbe1c7d4a86fc951dab6f6dab6bb51d5b95c408135988554a7","first_computed_at":"2026-07-05T07:42:23.656843Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:42:23.656843Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6zTBn61RYGocK5L+P5V1gwrMmTMons/Msq2TPFB7WFDM0ODr0fHFyI8TddR9GNFLOva4i3tdlojme5pwhAwbDA==","signature_status":"signed_v1","signed_at":"2026-07-05T07:42:23.657281Z","signed_message":"canonical_sha256_bytes"},"source_id":"2402.01035","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:261c29e9cf9fbc8fd1af28298256394df033642662f9ea37fca9a74bd58d0c7c","sha256:6c0c2e2e6f7844bd3579e5e2c3f0fae5e15631535eabc1018f58691acf0f9413"],"state_sha256":"3d1095e34702bad7188d034e19d852e7e8fd104fd95abcb5761e047db6e58b38"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"JFGCHuKwVo/V2xTiuMdWb0hg1m5xtPMEk3SsNPf1HezNXeg/IAjW01IyXhq37CdZhzV57/6YK5hgXCCr0wtmCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-11T03:13:37.335139Z","bundle_sha256":"7f76af61178cc64be89a553a9cf2789140f062d7e36edf34e8af1fa3257a6e91"}}