README.md

---
license: apache-2.0
language:
- yor
datasets:
- allenai/nllb
- allenai/MADLAD-400
- cis-lmu/Glot500
- castorini/afriberta-corpus
- allenai/c4
- legacy-datasets/wikipedia
- csebuetnlp/xlsum
- oscar-corpus/OSCAR-2109
library_name: transformers
pipeline_tag: text-generation
tags:
- goldfish
- arxiv:2408.10441
---

# yor_latn_100mb

Goldfish is a suite of monolingual language models trained for 350 languages.
This model is the <b>Yoruba</b> (Latin script) model trained on 100MB of data, after accounting for an estimated byte premium of 1.37; content-matched text in Yoruba takes on average 1.37x 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: yor_latn is an [individual language](https://iso639-3.sil.org/code_tables/639/data) 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)](https://www.arxiv.org/abs/2408.10441).

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

Sample usage also in this Google Colab: [link](https://colab.research.google.com/drive/1rHFpnQsyXJ32ONwCosWZ7frjOYjbGCXG?usp=sharing)

## 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): 137.50MB
* Training text data (byte premium scaled): 100.005MB
* Training tokens: 31010816 (x10 epochs)
* Vocabulary size: 50000
* Compute cost: 1.58317887356928e+17 FLOPs or ~15.0 NVIDIA A6000 GPU hours

Training datasets (percentages prior to deduplication):
* 46.33023%: [NLLB (CommonCrawl and ParaCrawl)](https://huggingface.co/datasets/allenai/nllb)
* 25.88587%: [MADLAD-400 (CommonCrawl)](https://huggingface.co/datasets/allenai/MADLAD-400)
* 23.42418%: [Glot500](https://huggingface.co/datasets/cis-lmu/Glot500), including [AfriBERTa](https://huggingface.co/datasets/castorini/afriberta-corpus), [AfroMAFT](https://zenodo.org/record/6990611#.Y0-yU-xBw-Q), [Earthlings](https://publicdata.canterbury.ac.nz/Research/Geocorpus/CCGLU_v5.0/), [Wortschatz Leipzig Data](https://wortschatz.uni-leipzig.de/en/download), [MC4](https://huggingface.co/datasets/allenai/c4), [Menyo20K](https://github.com/uds-lsv/menyo-20k_MT), [OSCAR](https://oscar-project.org/), [TeDDi](https://github.com/MorphDiv/TeDDi_sample), [W2C](https://lindat.mff.cuni.cz/repository/xmlui/handle/11858/00-097C-0000-0022-6133-9), [Wikipedia Hugging Face](https://huggingface.co/datasets/legacy-datasets/wikipedia), [XLSum](https://huggingface.co/datasets/csebuetnlp/xlsum)
* 2.89553%: [AfriBERTa](https://huggingface.co/datasets/castorini/afriberta-corpus)
* 0.95515%: [Wikipedia 2023/08](https://dumps.wikimedia.org/)
* 0.49782%: [eBible](https://ebible.org/find/)
* 0.01122%: [OSCAR 2021/09](https://huggingface.co/datasets/oscar-corpus/OSCAR-2109)


## 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},
}
```