Metraipolis › archive › dc_e9736b6bd73f
ufakzeka-1: Building and Evaluating a 151M-Parameter Turkish Language Model from Scratch — arXiv:2609.25081
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.25081
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T14:34:22.764Z
- SHA-256 of the archived text
5848f5d471da5f7962abbbc29b56027ab218690b2d235969463f6169c1ae7c6d- Stable URL of this page
https://metraipolis.com/documento/dc_e9736b6bd73f
How to cite
ufakzeka-1: Building and Evaluating a 151M-Parameter Turkish Language Model from Scratch — arXiv:2609.25081 — arXiv (en), https://arxiv.org/abs/2609.25081 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e9736b6bd73f (Metraipolis archive dc_e9736b6bd73f, sha256 5848f5d471da5f7962abbbc29b56027ab218690b2d235969463f6169c1ae7c6d)
BibTeX
@misc{dce9736b6bd73f,
title = {ufakzeka-1: Building and Evaluating a 151M-Parameter Turkish Language Model from Scratch — arXiv:2609.25081},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.25081},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e9736b6bd73f; sha256 5848f5d471da5f7962abbbc29b56027ab218690b2d235969463f6169c1ae7c6d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.25081. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We describe ufakzeka-1, a 151M-parameter (182M with embeddings) decoder-only Turkish language model pretrained from scratch on 13.5B tokens of openly licensed text and instruction-tuned for chat, at a total cost of about \$286 in cloud GPU, API and notebook time. The contribution is not the model's capability, which is what a model this size can be expected to have, but the record of building and measuring it: a Turkish byte-level tokenizer at 1.77 tokens per word, a three-stage pretraining schedule, a post-training mixture of openly licensed and generated data, and an evaluation battery of release gates, a rule-checked sweep of 5,508 conversations, judged conversations and hand tests, all with prompts held out from the training data, enforced by decontamination inside the data build and by a checked-in invariant script we run before each build. We report three findings that we believe transfer to other small-model efforts: a safety gate that had been "fixed" with training data written from its own questions read 64/64 while the honest figure was 34/64; training-seed variance was as large as the spread across every recipe we tried, so single-seed comparisons at this scale are uninformative; and data rounds repaired only what was absent from the data, while identity tracking over long context and multi-turn arithmetic did not move across any data change we tried, which we read as limits of the model size rather than gaps in the data, a reading the next, larger model will test. Weights, the data recipe, the evaluation code and the spend ledger are released under Apache-2.0. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.25081 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-08T13:46:15.666Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).