---
license: apache-2.0
language:
- grn
datasets:
- allenai/nllb
- allenai/MADLAD-400
- cis-lmu/Glot500
- statmt/cc100
- legacy-datasets/wikipedia
- oscar-corpus/OSCAR-2109
library_name: transformers
pipeline_tag: text-generation
tags:
- goldfish
- arxiv:2408.10441
---

# grn_latn_10mb

Goldfish is a suite of monolingual language models trained for 350 languages.
This model is the <b>Guarani</b> (Latin script) model trained on 10MB of data, after accounting for an estimated byte premium of 0.99; content-matched text in Guarani takes on average 0.99x 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: grn_latn is a [macrolanguage](https://iso639-3.sil.org/code_tables/639/data) code. None of its contained individual languages are included 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: 39087104
* Maximum sequence length: 512 tokens
* Training text data (raw): 9.92MB
* Training text data (byte premium scaled): 10.005MB
* Training tokens: 2616320 (x10 epochs)
* Vocabulary size: 50000
* Compute cost: 1978743348264960.0 FLOPs or ~0.2 NVIDIA A6000 GPU hours

Training datasets (percentages prior to deduplication):
* 49.98971%: [NLLB (CommonCrawl and ParaCrawl)](https://huggingface.co/datasets/allenai/nllb)
* 24.31205%: [MADLAD-400 (CommonCrawl)](https://huggingface.co/datasets/allenai/MADLAD-400)
* 15.49453%: [Glot500](https://huggingface.co/datasets/cis-lmu/Glot500), including [CC100](https://huggingface.co/datasets/statmt/cc100), [GiossaMedia](https://github.com/sgongora27/giossa-gongora-guarani-2021), [Wortschatz Leipzig Data](https://wortschatz.uni-leipzig.de/en/download), [NLLB_seed](https://github.com/facebookresearch/flores/blob/main/nllb_seed/README.md), [OSCAR](https://oscar-project.org/), [Wikipedia Hugging Face](https://huggingface.co/datasets/legacy-datasets/wikipedia)
* 6.12356%: [Wikipedia 2023/08](https://dumps.wikimedia.org/)
* 4.05310%: [AmericasNLP (excluding AmericasNLI)](https://turing.iimas.unam.mx/americasnlp/)
* 0.02706%: [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},
}
```