khk_cyrl_100mb / README.md
goldfish-models's picture
Upload README.md with huggingface_hub
5619ba1 verified
|
raw
history blame
2.57 kB
---
license: apache-2.0
language:
- mon
- khk
datasets:
- allenai/nllb
library_name: transformers
pipeline_tag: text-generation
tags:
- goldfish
- arxiv:2408.10441
---
# khk_cyrl_100mb
Goldfish is a suite of monolingual language models trained for 350 languages.
This model is the <b>Halh Mongolian</b> (Cyrillic script) model trained on 100MB of data, after accounting for an estimated byte premium of 1.80; content-matched text in Halh Mongolian takes on average 1.80x 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: khk_cyrl is an [individual language](https://iso639-3.sil.org/code_tables/639/data) code. Macrolanguage code mon_cyrl (Mongolian) 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)](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): 180.18MB
* Training text data (byte premium scaled): 100.005MB
* Training tokens: 21806592 (x10 epochs)
* Vocabulary size: 50000
* Compute cost: 1.11278789296128e+17 FLOPs or ~10.5 NVIDIA A6000 GPU hours
Training datasets (percentages prior to deduplication):
* 100.00000%: [NLLB (CommonCrawl and ParaCrawl)](https://huggingface.co/datasets/allenai/nllb)
## 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},
}
```