kpv_cyrl_full

Goldfish is a suite of monolingual language models trained for 350 languages. This model is the Komi-Zyrian (Cyrillic script) model trained on 5MB of data (all our data in the language), after accounting for an estimated byte premium of 1.67; content-matched text in Komi-Zyrian takes on average 1.67x as many UTF-8 bytes to encode as English. The Goldfish models are trained primarily for comparability across languages and for low-resource languages; Goldfish performance for high-resource languages is not designed to be comparable with modern large language models (LLMs).

Note: kpv_cyrl is an individual language code. Macrolanguage code kom_cyrl (Komi) is included in Goldfish. Consider using that model depending on your use case.

All training and hyperparameter details are in our paper, Goldfish: Monolingual Language Models for 350 Languages (Chang et al., 2024).

Training code and sample usage: https://github.com/tylerachang/goldfish

Sample usage also in this Google Colab: link

Model details:

To access all Goldfish model details programmatically, see https://github.com/tylerachang/goldfish/blob/main/model_details.json. All models are trained with a [CLS] (same as [BOS]) token prepended, and a [SEP] (same as [EOS]) token separating sequences. For best results, make sure that [CLS] is prepended to your input sequence (see sample usage linked above)! Details for this model specifically:

Architecture: gpt2
Parameters: 124770816
Maximum sequence length: 512 tokens
Training text data (raw): 8.67MB
Training text data (byte premium scaled): 5.195MB
Training tokens: 1355776 (x10 epochs)
Vocabulary size: 50000
Compute cost: 6919013007360000.0 FLOPs or ~0.7 NVIDIA A6000 GPU hours

Training datasets (percentages prior to deduplication):

99.99709%: Languages of Russia
0.00291%: Tatoeba

Citation

If you use this model, please cite:

@article{chang-etal-2024-goldfish,
  title={Goldfish: Monolingual Language Models for 350 Languages},
  author={Chang, Tyler A. and Arnett, Catherine and Tu, Zhuowen and Bergen, Benjamin K.},
  journal={Preprint},
  year={2024},
  url={https://www.arxiv.org/abs/2408.10441},
}