{"as_of":"2026-08-10T03:15:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:88db7b077b674b43c33f283491855f1296720469df77fb5f5f7a968e55927b4d","coverage":[{"denominator":58,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":58,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-01T21:22:42.436125Z","state":"measured"},{"denominator":58,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":58,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-09T06:31:02.800959+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2607.16117/citation-record","integrity":"/paper/2607.16117/integrity","json":"/paper/2607.16117/citation-record.json","paper":"/paper/2607.16117"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:36.763206Z","title":"Alabi, Yanke Mao, Haonan Gao, and En-Shiun Annie Lee","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":1,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:36.763206Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:82be34f75033b001302ce32cb3afd77a63866482f6cefa6b6b7aa9e4a1236d77","observation_id":"4afdfa54-f88a-46e0-9a15-bcbed05c5422","resolution":{"observed_at":"2026-08-01T21:22:36.763206Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:36.812325Z","title":"Do all languages cost the same? tokenization in the era of commercial language models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":2,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:36.812325Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:29d7c4c925f40a110d7de821e722518cfac7fd91955c545f082f1716e3e28fc6","observation_id":"7ebeb8e9-7f88-4a7f-b891-d2444363f441","resolution":{"observed_at":"2026-08-01T21:22:36.812325Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:36.902178Z","title":"Massively multilingual sentence embeddings for zero-shot cross-lingual transfer and beyond","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":3,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:36.902178Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:0e11d5e5b50023ad04484b48b1a4659e1418fdb02b74b916c920e257816f7ab3","observation_id":"690b540f-d575-476a-8e7d-c6a90a6cc88c","resolution":{"observed_at":"2026-08-01T21:22:36.902178Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2502.00894","last_updated":"2025-02-02T20:06:39Z","snapshot_observed_at":"2026-08-09T17:18:47.016847Z","submitted_at":"2025-02-02T20:06:39Z","title":"MorphBPE: A Morpho-Aware Tokenizer Bridging Linguistic Complexity for Efficient LLM Training Across Morphologies","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2502.00894","snapshot_observed_at":"2026-08-01T21:22:36.983051Z","title":"Morphbpe: A morpho-aware tokenizer bridging linguistic complexity for efficient llm training across morphologies, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":4,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:36.983051Z"},"links":{"cited_paper":"/paper/2502.00894","citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:108261480abe944f7c469d4fb3561184497416cd2151fa88c060fd50b3884787","observation_id":"89d56c74-8406-4bfb-a1c7-5fb00e869092","resolution":{"observed_at":"2026-08-01T21:22:36.983051Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2004.05150","last_updated":"2020-12-02T17:52:35Z","snapshot_observed_at":"2026-07-31T17:17:17.205582Z","submitted_at":"2020-04-10T17:54:09Z","title":"Longformer: The Long-Document Transformer","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2004.05150","snapshot_observed_at":"2026-08-01T21:22:37.066132Z","title":"Peters, and Arman Cohan","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.066132Z"},"links":{"cited_paper":"/paper/2004.05150","citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:5466bc001046e244a171370c62a1d1e9adffaa262b37849cb5347b907739fc68","observation_id":"d9a50f12-f94e-496d-a3b0-c902f0c676c2","resolution":{"observed_at":"2026-08-01T21:22:37.066132Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":"10.1007/s10579-014-9287-y","metadata_source":"openalex","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"A massively parallel corpus: the bible in 100 languages","venue":"Language Resources and Evaluation","work_id":"97d93c16-b1d1-4078-a263-656ef38923d7","year":2015},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":6,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.209113Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:60667e7888b1644c55cb31ce2ca70b3440ebcef4e90ca055eb11aa3aa96efec9","observation_id":"1db98461-7eb8-4595-bc6f-c35a2a4aa943","resolution":{"observed_at":"2026-08-01T21:23:49.821748Z","resolver_source":"doi","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-09T06:31:02.800959+00:00","source":"crossref"},{"observed_at":"2026-08-09T06:30:57.326959+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:37.281928Z","title":"Clark, Dan Garrette, Iulia Turc, and John Wieting","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":7,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.281928Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:be2b710fc84826291792689d97bd32c73cbddd116324dac2a704f6364062dc5a","observation_id":"56cfdd58-8a93-49ce-9d91-b3cb70c42225","resolution":{"observed_at":"2026-08-01T21:22:37.281928Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:37.384422Z","title":"BERT : Pre-training of deep bidirectional transformers for language understanding","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":8,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.384422Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:dbd718cfa5d7326f9db90a1d75829b7b465b7f968e715a1c7f24d36f3f0d8623","observation_id":"1c1471e1-2942-487b-a3f8-870c9090db12","resolution":{"observed_at":"2026-08-01T21:22:37.384422Z","resolver_source":null,"status":"malformed_identifier"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:37.479630Z","title":"Multiscale byte language models -- a hierarchical architecture for causal million-length sequence modeling","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":9,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.479630Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:055a2b33633019f00a592a82d06ad5b821343ba461c6a857c481045428e7e82e","observation_id":"f84267cb-b37d-44e6-a32e-f03f760090f7","resolution":{"observed_at":"2026-08-01T21:22:37.479630Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:37.572373Z","title":"O mer Veysel C a g atan, Akash Kundu, Martin Bernstorff, Shitao Xiao, Akshita Sukhlecha, Bhavish Pahwa, Rafa Po \\'s wiata, Kranthi Kiran GV, Shawon Ashraf, Daniel Auras, Bj \\","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":10,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.572373Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:fef24ca825c335445a0f27840f85c51068f46d710e76ed51072e3b4a8e438c62","observation_id":"af455e76-ba7b-4c17-a00b-03e5f6650175","resolution":{"observed_at":"2026-08-01T21:22:37.572373Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2602.15763","last_updated":"2026-02-24T10:44:44Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2026-02-17T17:50:56Z","title":"GLM-5: from Vibe Coding to Agentic Engineering","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2602.15763","snapshot_observed_at":"2026-08-01T21:22:37.680542Z","title":"Glm-5: from vibe coding to agentic engineering, 2026","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":11,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.680542Z"},"links":{"cited_paper":"/paper/2602.15763","citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:a89f8006fa788be8753d6aca3d1fbfbeb53d018b16c83d4fbfcf2404183b4cb9","observation_id":"9b48849d-5b01-4662-8309-4b87b91a1977","resolution":{"observed_at":"2026-08-01T21:22:37.680542Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2407.21783","last_updated":"2024-11-23T23:27:33Z","snapshot_observed_at":"2026-07-06T18:55:11.576666Z","submitted_at":"2024-07-31T17:54:27Z","title":"The Llama 3 Herd of Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2407.21783","snapshot_observed_at":"2026-08-01T21:22:37.776856Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":12,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.776856Z"},"links":{"cited_paper":"/paper/2407.21783","citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:1d5715b3251b0f5fd15f8d535ee57eecf7182e4fc0d236bcb6385db98fcb1850","observation_id":"3c9eeacf-439a-4216-9443-06f56ccc44ce","resolution":{"observed_at":"2026-08-01T21:22:37.776856Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:37.864150Z","title":"Bitext mining using distilled sentence representations for low-resource languages","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":13,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.864150Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:e499764eedac112194385ee328301dd7b2c454f1a9e3aad9765e48175157fd38","observation_id":"57a89f98-0c42-4bb8-affa-5c4801d5e21e","resolution":{"observed_at":"2026-08-01T21:22:37.864150Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:37.939070Z","title":"HTML Bible Index -- Ukrainian","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":14,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:37.939070Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:0309d2d730b3f7bb88d4beaa74347c858afae56c88ac21b3731b15ab121d8110","observation_id":"66cc1f93-437d-4249-a7c2-cc3e3cfc178e","resolution":{"observed_at":"2026-08-01T21:22:37.939070Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.019718Z","title":"Perceiver: General perception with iterative attention","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":15,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.019718Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:644bf2992e78a505dee6f69b2e792dc4a1c26d344e54e0fce9285da4db915ff8","observation_id":"4e885ffe-ebd8-4dfa-812c-11fbb4ae657e","resolution":{"observed_at":"2026-08-01T21:22:38.019718Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.087203Z","title":"Ticls: Tightly coupled language text spotter","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":16,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.087203Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:4080c41796e9003b08925f11fa2e98ad83d2a204c87916338091066d63a7f6aa","observation_id":"51b4d924-55d4-48cf-a363-b90deaada197","resolution":{"observed_at":"2026-08-01T21:22:38.087203Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.175841Z","title":"Lotz, Ingo Ziegler, Phillip Rust, and Desmond Elliott","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":17,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.175841Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:4a3e1dadf34798b9395efce2d21096b4d6e7bebc5bac1bdcb55787631cd939c4","observation_id":"91cc27fb-f243-4255-b5ae-5eb7e8441060","resolution":{"observed_at":"2026-08-01T21:22:38.175841Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.269022Z","title":"Supervised contrastive learning","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":18,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.269022Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:12fe6b4071c6647cd63cc4a88c5c84eb31cd747ad8a586aaae8f7a51692cd047","observation_id":"471917f3-ca0e-4399-8062-c12b98fd00ba","resolution":{"observed_at":"2026-08-01T21:22:38.269022Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.363007Z","title":"Ocr-free document understanding transformer","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":19,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.363007Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:3d822d16b12504519108ba21e96ef6dc10087968b7b0454a3eb3efb59db871d4","observation_id":"b4e2f324-9862-4c83-bdeb-5478d9520292","resolution":{"observed_at":"2026-08-01T21:22:38.363007Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.463697Z","title":"Adam: A method for stochastic optimization","venue":null,"work_id":null,"year":2015},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":20,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.463697Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:673fd97eb811ff992fa2963d8ea74a87a0c49319ca01c63b69a407f93f746351","observation_id":"ab8c803d-d344-4aa5-a89c-107a7c3ec915","resolution":{"observed_at":"2026-08-01T21:22:38.463697Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.628112Z","title":"E uroparl: A parallel corpus for statistical machine translation","venue":null,"work_id":null,"year":2005},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":21,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.628112Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:a86f877837bbb302182f6fedb368deb888749992dd367c22cd138c4ec6f1ab19","observation_id":"c89ecd3e-30f7-4298-baeb-c123b5584fa9","resolution":{"observed_at":"2026-08-01T21:22:38.628112Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.779807Z","title":"Subword regularization: Improving neural network translation models with multiple subword candidates","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":22,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.779807Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:a3a4c1e742219b6455edd6c2e3d868948b69f81e1c055bf67657c0924a0756de","observation_id":"b48456d1-f3bd-4221-b18a-c26cb4e44827","resolution":{"observed_at":"2026-08-01T21:22:38.779807Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.914733Z","title":"S entence P iece: A simple and language independent subword tokenizer and detokenizer for neural text processing","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":23,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.914733Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:9361d460c335c149cbc2c78b44a55df9bd688215a43821d0a435890a39b928f7","observation_id":"7ba85ef0-8e0b-4cd5-bbe4-627f9a986fef","resolution":{"observed_at":"2026-08-01T21:22:38.914733Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:38.983778Z","title":"P ix2 S truct: Screenshot parsing as pretraining for visual language understanding","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":24,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:38.983778Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:b510d6420ea6262d278b041d335e56e3fffc8620b66ffc809cd7aec6c9820d9e","observation_id":"493f913f-00d2-4540-b3bc-2b1c2c5420a3","resolution":{"observed_at":"2026-08-01T21:22:38.983778Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:39.135595Z","title":"Trocr: transformer-based optical character recognition with pre-trained models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":25,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:39.135595Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:075bf5025e6e0272a5a38ed8f63d62b927d8fe6cd53bd7afdab988ef4686ceeb","observation_id":"dedbb092-5bd6-4977-9bf3-ec1232f7e99f","resolution":{"observed_at":"2026-08-01T21:22:39.135595Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:39.275565Z","title":"Tokenization impacts multilingual language modeling: Assessing vocabulary allocation and overlap across languages","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":26,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:39.275565Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:dd2580c86255bc7692b3440170c366e5d3e554493173b166fd01e727b95509e1","observation_id":"de52aea0-b594-4c14-b241-c165b657feee","resolution":{"observed_at":"2026-08-01T21:22:39.275565Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":"10.18653/v1/2023.acl-long.900","metadata_source":"openalex","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Character-aware models improve visual text rendering","venue":null,"work_id":"4d4b03d3-9527-4401-a02f-dae4c459bc8f","year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":27,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:39.449884Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:a1d96a582b48b7b9ddfc94ab26e8fe24055d5a89d696294ff72267a73d1fb0d1","observation_id":"ac2178cf-71ee-42c3-82be-3dd35c4887e3","resolution":{"observed_at":"2026-08-01T21:23:49.478668Z","resolver_source":"doi","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-09T06:31:02.800959+00:00","source":"crossref"},{"observed_at":"2026-08-09T06:30:57.326959+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":"10.1007/978-3-031-73226-3_21","metadata_source":"openalex","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Glyph-byt5: A customized text encoder for accurate visual text rendering","venue":"Lecture notes in computer science","work_id":"82536e16-a5e4-4f84-a2ec-da7e134c9517","year":2024},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":28,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:39.605288Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:f9a6dfc7e3d4e3458aedfa6a4df9d07f14e4ce4ff57ebbc36ac2ee4b8690858e","observation_id":"1de230ae-c90c-4808-848b-05ba9dfb49e1","resolution":{"observed_at":"2026-08-01T21:23:49.345965Z","resolver_source":"doi","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-09T06:31:02.800959+00:00","source":"crossref"},{"observed_at":"2026-08-09T06:30:57.326959+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:39.790370Z","title":"Decoupled weight decay regularization","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":29,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:39.790370Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:4ac6ddcc21cc7acd1e79145833b09ca3de94dedfc4309d1ef4edb9c3b9dbd2a8","observation_id":"cae33c4a-309e-4b18-8237-aa5ce9036c2f","resolution":{"observed_at":"2026-08-01T21:22:39.790370Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:39.929199Z","title":"Beyond fertility: STRR as a metric for multilingual tokenization evaluation","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":30,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:39.929199Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:75d6306919efce5f936a72fef0330b3966c04abc84e6f8a298a44a95e059fa24","observation_id":"7152fbb3-9129-4a9b-971d-25084ce08c86","resolution":{"observed_at":"2026-08-01T21:22:39.929199Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.053631Z","title":"Hierarchical autoregressive transformers: Combining byte- and word-level processing for robust, adaptable language models","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":31,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.053631Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:7e0e17aac7966bdd69192f924c6588f5c57909a8a8da28a14f7ea620be4be821","observation_id":"4fac4be3-b216-46be-81f3-7fa56dff9d7a","resolution":{"observed_at":"2026-08-01T21:22:40.053631Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2207.04672","last_updated":"2022-08-25T17:10:53Z","snapshot_observed_at":"2026-07-06T13:29:47.927628Z","submitted_at":"2022-07-11T07:33:36Z","title":"No Language Left Behind: Scaling Human-Centered Machine Translation","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2207.04672","snapshot_observed_at":"2026-08-01T21:22:40.189860Z","title":null,"venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":32,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.189860Z"},"links":{"cited_paper":"/paper/2207.04672","citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:c08b35b72d52c74d6e85b838efe79593581d4076dd231932168af4db085cdfaa","observation_id":"664c8bad-32ff-47e9-ab49-1e76f9ff9703","resolution":{"observed_at":"2026-08-01T21:22:40.189860Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.322457Z","title":"Byte latent transformer: Patches scale better than tokens","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":33,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.322457Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:d81a6a9ddfa700882909fd24044f90db4412a50971dbe2ab551a275f413d4ad1","observation_id":"aefc44b6-41f4-4a9f-b909-64291cf33feb","resolution":{"observed_at":"2026-08-01T21:22:40.322457Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.485573Z","title":"Language model tokenizers introduce unfairness between languages","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":34,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.485573Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:1f2e5e2b537a5aae6b95a8e0b3873fab6446e9151d0f56e8feabfd45192edc80","observation_id":"46aa5a51-013f-4128-b092-16f4a6a767c0","resolution":{"observed_at":"2026-08-01T21:22:40.485573Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.573849Z","title":"Language Models are Unsupervised Multitask Learners , 2019","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":35,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.573849Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:6eb527cb3cb92743c85b0b44efc56386313ba7e4e3c355b24dff51016f65af30","observation_id":"1990d8bf-9dc6-4202-9278-839befd4f4fe","resolution":{"observed_at":"2026-08-01T21:22:40.573849Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.652257Z","title":"Learning transferable visual models from natural language supervision","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":36,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.652257Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:3fd4f3711b20919b7d403cc7da6b7ecf2c15293a4120e922edbf129454f7780a","observation_id":"6e76818f-677f-4f88-aa77-f5dbdd3cfe6b","resolution":{"observed_at":"2026-08-01T21:22:40.652257Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.764452Z","title":"When every token counts: Optimal segmentation for low-resource language models","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":37,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.764452Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:5bea1baff0ffabfd11240e48baf62af5c5d90bbfb69aaa6c94aa48df9e62262c","observation_id":"4122f4c8-1c4b-47c0-93ea-8717ca75631b","resolution":{"observed_at":"2026-08-01T21:22:40.764452Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.866372Z","title":"Making monolingual sentence embeddings multilingual using knowledge distillation","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":38,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.866372Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:69cafdbad5177c4bb4a2056a3c668ad1c65aeab0df5c6b70ae31eb5836baa632","observation_id":"fbd82c04-164b-4c27-a563-ae7bf0763a07","resolution":{"observed_at":"2026-08-01T21:22:40.866372Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:40.965157Z","title":"How good is your tokenizer? on the monolingual performance of multilingual language models","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":39,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:40.965157Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:81fa38834e5509cc7476d78a481699af3977efe641636a9a6164e6f909b423fc","observation_id":"708f2e87-c438-477a-a744-7abc754436c6","resolution":{"observed_at":"2026-08-01T21:22:40.965157Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.052936Z","title":"Lotz, Emanuele Bugliarello, Elizabeth Salesky, Miryam de Lhoneux, and Desmond Elliott","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":40,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.052936Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:e09be47e27c9b7116b14ea103534e57944c07a08e31ac0147a4d83aff6822470","observation_id":"e93f06ce-3a4d-43f8-893d-0a810a01ee8b","resolution":{"observed_at":"2026-08-01T21:22:41.052936Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.122188Z","title":"Robust open-vocabulary translation from visual text representations","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":41,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.122188Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:a9fc6f0578a3c5c473b36cee21af0c306633578d1c081914f97e6ee1b816a166","observation_id":"e818e859-dcd0-4289-8205-6ec2795292d9","resolution":{"observed_at":"2026-08-01T21:22:41.122188Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.193745Z","title":"Neural machine translation of rare words with subword units","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":42,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.193745Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:3f4c801a598ab64d591500c9c61234334b0e07a8b01262652d2769e751c7920b","observation_id":"1365610e-2bd3-485b-a54e-c061e205b892","resolution":{"observed_at":"2026-08-01T21:22:41.193745Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.268227Z","title":"Dropout: A simple way to prevent neural networks from overfitting","venue":null,"work_id":null,"year":1929},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":43,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.268227Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:8a7eb14226f52855b191409e7c04e68d16173ef2f17c02dce360834750a865ca","observation_id":"3e0a7752-7c68-436e-908b-176d6a0ff37c","resolution":{"observed_at":"2026-08-01T21:22:41.268227Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.340888Z","title":"Roformer: Enhanced transformer with rotary position embedding","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":44,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.340888Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:dfd9c4015a95eeb5a3c58c71b644885c58c31ce99f2876df3592bf7750ead66a","observation_id":"c9fffd0b-2f2a-4b62-9e69-cd84c0bacdff","resolution":{"observed_at":"2026-08-01T21:22:41.340888Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.408834Z","title":"Tran, Sebastian Ruder, Jai Gupta, Hyung Won Chung, Dara Bahri, Zhen Qin, Simon Baumgartner, Cong Yu, and Donald Metzler","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":45,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.408834Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:cd32780b26c9cb194de245c398abfa0c1d1f2624b8010c1604487121d94fcd10","observation_id":"acfe7422-013c-4cc1-9c27-d9272676e2dc","resolution":{"observed_at":"2026-08-01T21:22:41.408834Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1807.03748","last_updated":"2019-01-22T18:47:12Z","snapshot_observed_at":"2026-07-06T06:49:24.960992Z","submitted_at":"2018-07-10T16:52:11Z","title":"Representation Learning with Contrastive Predictive Coding","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1807.03748","snapshot_observed_at":"2026-08-01T21:22:41.479804Z","title":"Representation learning with contrastive predictive coding, 2019","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":46,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.479804Z"},"links":{"cited_paper":"/paper/1807.03748","citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:7b7d8a8bfd56b37c6026590f8910b56ceb447416caf4beb08fd9c0dc20cdb5a4","observation_id":"29f73f0c-4deb-4a60-9613-bf83de9ab48c","resolution":{"observed_at":"2026-08-01T21:22:41.479804Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.567529Z","title":"Attention is all you need","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":47,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.567529Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:d2ad0975b38be2135041658b47b35ce1388940bb67d54eeb0567e60bcbc3fab4","observation_id":"617299bd-5135-4124-9555-716636fe1476","resolution":{"observed_at":"2026-08-01T21:22:41.567529Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.668392Z","title":"Error patterns in historical ocr: A comparative analysis of trocr and a vision-language model, 2026","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":48,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.668392Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:a46f1b85e2091c2157401dfdb49aebba1156f940c742b4aa49ea1eec1059f4c5","observation_id":"11952a5d-fd7e-4996-987c-c84521da349f","resolution":{"observed_at":"2026-08-01T21:22:41.668392Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.767793Z","title":null,"venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":49,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.767793Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:ac4d38f244a10a7a792cef0d087737a4767c8c0465e094f06ff2104ba39a3e52","observation_id":"2b1bc23a-7e06-44ad-ad71-17b0b215c266","resolution":{"observed_at":"2026-08-01T21:22:41.767793Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.836533Z","title":"The Holy Bible international: Text and audio bibles","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":50,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.836533Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:77fc8eabf015443c471f26b1c6522ebdfb88892854f4541f0c0fc1327e856a7f","observation_id":"0f8bf97d-6fe7-460e-892e-9b1b9dbaa99f","resolution":{"observed_at":"2026-08-01T21:22:41.836533Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.913579Z","title":"On layer normalization in the transformer architecture","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":51,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.913579Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:290363743137c2d88f240d2d5bc1949d1289f5bb402fe05290a8f5e471483cb3","observation_id":"708009d9-68af-40ca-b455-d605ac09cc3d","resolution":{"observed_at":"2026-08-01T21:22:41.913579Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:41.984457Z","title":"B y T 5: Towards a token-free future with pre-trained byte-to-byte models","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":52,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:41.984457Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:12d2d0ed6c5a88583e840805481bd748891e5b9d2ff59cf9a9a580e383656b16","observation_id":"478fbc81-65c8-4675-8d82-87983efe1de8","resolution":{"observed_at":"2026-08-01T21:22:41.984457Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.09388","last_updated":"2025-05-14T13:41:34Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2025-05-14T13:41:34Z","title":"Qwen3 Technical Report","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.09388","snapshot_observed_at":"2026-08-01T21:22:42.054784Z","title":"Qwen3 technical report, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":53,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:42.054784Z"},"links":{"cited_paper":"/paper/2505.09388","citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:8bea20609f37db8089c1fc9c30b7838dbe04c335c89a8f65f9c57fba45cda62a","observation_id":"298aecb2-20e4-4f5e-b979-36255ca1d114","resolution":{"observed_at":"2026-08-01T21:22:42.054784Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:42.124243Z","title":"Megabyte: Predicting million-byte sequences with multiscale transformers","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":54,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:42.124243Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:5670a5f2ea0e0be8ccd91fd3f1ba2394cbfbe8ee4bfd9725faf83a8798651e1e","observation_id":"53396375-1549-4551-9fa7-5bac8d3560ea","resolution":{"observed_at":"2026-08-01T21:22:42.124243Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:42.199436Z","title":"Big bird: Transformers for longer sequences","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":55,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:42.199436Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:6487319cca9316fe9ba70e8bd15ff0f046374e4a2bfcf6285298c5988132b8fd","observation_id":"2c442f75-bb28-49ac-a844-8fb21ebba32d","resolution":{"observed_at":"2026-08-01T21:22:42.199436Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:42.285016Z","title":"Root mean square layer normalization","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":56,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:42.285016Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:d24ef3c0e1b59acffaa6fc855771b903edfcbe8466aab8574e42a3809d5a0396","observation_id":"072e5444-8eeb-43ff-90a2-3b30afa727cb","resolution":{"observed_at":"2026-08-01T21:22:42.285016Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:42.357323Z","title":"MIRACL : A multilingual retrieval dataset covering 18 diverse languages","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":57,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:42.357323Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:109d55526c56463f61c092368f688d2edc749e609200f8f626d803e585efe3da","observation_id":"5b5aa7f9-d62b-4d8d-8579-5c62997a6ba5","resolution":{"observed_at":"2026-08-01T21:22:42.357323Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-01T21:22:42.436125Z","title":"The U nited N ations parallel corpus v1.0","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content","version":1},"reference_index":58,"source":"arxiv_source","source_observed_at":"2026-08-01T21:22:42.436125Z"},"links":{"citing_paper":"/paper/2607.16117"},"observation_digest":"sha256:c50965c3dc284c6e50835f56b04230ba2cc2dfb3ad55ef3cb8f91eb7288f89f2","observation_id":"d1798f42-01cd-43cb-938c-cf700bd193e7","resolution":{"observed_at":"2026-08-01T21:22:42.436125Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2607.16117","last_updated":"2026-07-17T16:55:27Z","latest_version":1,"primary_category":"cs.CL","snapshot_observed_at":"2026-08-09T20:20:16.087466Z","submitted_at":"2026-07-17T16:55:27Z","title":"Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content"},"reference_resolution":{"displayed":58,"state_counts":{"malformed_identifier":1,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":54,"verified_exact":3,"verified_fuzzy":0},"total_outbound_references":58},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-09T06:31:02.800959+00:00","source":"crossref"},{"observed_at":"2026-08-09T06:30:57.326959+00:00","source":"retraction_watch"}],"thesis":"As of 10 August 2026, this Paper Citation Record lists 58 of 58 outbound references and 0 inbound Pith citation observations for arXiv:2607.16117."}