|
|
| --- |
| license: apache-2.0 |
| language: |
| - pus |
| datasets: |
| - allenai/MADLAD-400 |
| - allenai/nllb |
| - oscar-corpus/OSCAR-2109 |
| - cis-lmu/Glot500 |
| - sil-ai/bloom-lm |
| library_name: transformers |
| pipeline_tag: text-generation |
| tags: |
| - goldfish |
| - arxiv:2408.10441 |
| --- |
| |
| # pus_arab_5mb |
|
|
| Goldfish is a suite of monolingual language models trained for 350 languages. |
| This model is the <b>Pushto</b> (Arabic script) model trained on 5MB of data, after accounting for an estimated byte premium of 1.59; content-matched text in Pushto takes on average 1.59x 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: pus_arab is a [macrolanguage](https://iso639-3.sil.org/code_tables/639/data) code. Individual language code pbt_arab (Southern Pashto) is included in Goldfish, although with less data. |
|
|
| 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: 35941376 |
| * Maximum sequence length: 512 tokens |
| * Training text data (raw): 7.93MB |
| * Training text data (byte premium scaled): 5.005MB |
| * Training tokens: 1271296 (x10 epochs) |
| * Vocabulary size: 44041 |
| * Compute cost: 961479133102080.0 FLOPs or ~0.1 NVIDIA A6000 GPU hours |
| |
| Training datasets (percentages prior to deduplication): |
| * 60.56138%: [MADLAD-400 (CommonCrawl)](https://huggingface.co/datasets/allenai/MADLAD-400) |
| * 16.95008%: [NLLB (CommonCrawl and ParaCrawl)](https://huggingface.co/datasets/allenai/nllb) |
| * 11.68420%: [OSCAR 2021/09](https://huggingface.co/datasets/oscar-corpus/OSCAR-2109) |
| * 6.73176%: [Glot500](https://huggingface.co/datasets/cis-lmu/Glot500), including [BLOOM](https://huggingface.co/datasets/sil-ai/bloom-lm), [CCNet](https://github.com/facebookresearch/cc_net), [Earthlings](https://publicdata.canterbury.ac.nz/Research/Geocorpus/CCGLU_v5.0/), [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/), [Tatoeba](https://tatoeba.org/en/), [TICO](https://tico-19.github.io/) |
| * 4.07251%: [Wikipedia 2023/08](https://dumps.wikimedia.org/) |
| * 0.00007%: [Tatoeba](https://tatoeba.org/en/) |
| |
| |
| ## 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}, |
| } |
| ``` |
| |
