Hugging Face's logo

goldfish-models
/
krc_cyrl_full

Text Generation
Transformers
PyTorch
Karachay-Balkar
gpt2
goldfish
text-generation-inference
Inference Endpoints
Model card Files Community
krc_cyrl_full / README.md
goldfish-models's picture
goldfish-models
Upload README.md with huggingface_hub
5a75fd3 verified
preview code
|
raw
history blame contribute delete
3.23 kB
metadata
license: apache-2.0
language:
  - krc
datasets:
  - allenai/MADLAD-400
  - cis-lmu/Glot500
  - legacy-datasets/wikipedia
  - oscar-corpus/OSCAR-2109
library_name: transformers
pipeline_tag: text-generation
tags:
  - goldfish
  - arxiv:2408.10441

krc_cyrl_full

Goldfish is a suite of monolingual language models trained for 350 languages. This model is the Karachay-Balkar (Cyrillic script) model trained on 21MB of data (all our data in the language), after accounting for an estimated byte premium of 1.87; content-matched text in Karachay-Balkar takes on average 1.87x 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: krc_cyrl is an individual language code. It is not contained in any macrolanguage codes contained in Goldfish (for script cyrl).

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): 39.22MB
  • Training text data (byte premium scaled): 21.025MB
  • Training tokens: 4627456 (x10 epochs)
  • Vocabulary size: 50000
  • Compute cost: 2.3609825427456e+16 FLOPs or ~2.2 NVIDIA A6000 GPU hours

Training datasets (percentages prior to deduplication):

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},
}