metadata

license: apache-2.0
language:
  - iso
datasets:
  - allenai/MADLAD-400
library_name: transformers
pipeline_tag: text-generation
tags:
  - goldfish

iso_latn_full

Goldfish is a suite of monolingual language models trained for 350 languages. This model is the Isoko (Latin script) model trained on 8MB of data (all our data in the language), after accounting for an estimated byte premium of 1.48; content-matched text in Isoko takes on average 1.48x 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: iso_latn is an individual language code. It is not contained in any macrolanguage codes contained in Goldfish (for script latn).

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. Details for this model specifically:

Architecture: gpt2
Parameters: 109042176
Maximum sequence length: 512 tokens
Training text data (raw): 12.03MB
Training text data (byte premium scaled): 8.115MB
Training tokens: 2638336 (x10 epochs)
Vocabulary size: 29138
Compute cost: 1.3456800940032e+16 FLOPs or ~1.3 NVIDIA A6000 GPU hours

Training datasets (percentages prior to deduplication):

100.00000%: MADLAD-400 (CommonCrawl)

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},
}