goldfish-models / afb_arab_full

huggingface.co
Total runs: 26
24-hour runs: 0
7-day runs: 4
30-day runs: -13
Model's Last Updated: August 27 2024
text-generation

Introduction of afb_arab_full

Model Details of afb_arab_full

afb_arab_full

Goldfish is a suite of monolingual language models trained for 350 languages. This model is the Gulf Arabic (Arabic script) model trained on 14MB of data (all our data in the language), after accounting for an estimated byte premium of 1.37; content-matched text in Gulf Arabic 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: afb_arab is an individual language code. However, you may also want to consider the arb_arab (Standard Arabic) models which are trained on more data.

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: 124770816
  • Maximum sequence length: 512 tokens
  • Training text data (raw): 19.58MB
  • Training text data (byte premium scaled): 14.255MB
  • Training tokens: 3247616 (x10 epochs)
  • Vocabulary size: 50000
  • Compute cost: 1.656983420928e+16 FLOPs or ~1.6 NVIDIA A6000 GPU hours

Training datasets (percentages prior to deduplication):

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},
}

Runs of goldfish-models afb_arab_full on huggingface.co

26
Total runs
0
24-hour runs
0
3-day runs
4
7-day runs
-13
30-day runs

More Information About afb_arab_full huggingface.co Model

More afb_arab_full license Visit here:

https://choosealicense.com/licenses/apache-2.0

afb_arab_full huggingface.co

afb_arab_full huggingface.co is an AI model on huggingface.co that provides afb_arab_full's model effect (), which can be used instantly with this goldfish-models afb_arab_full model. huggingface.co supports a free trial of the afb_arab_full model, and also provides paid use of the afb_arab_full. Support call afb_arab_full model through api, including Node.js, Python, http.

goldfish-models afb_arab_full online free

afb_arab_full huggingface.co is an online trial and call api platform, which integrates afb_arab_full's modeling effects, including api services, and provides a free online trial of afb_arab_full, you can try afb_arab_full online for free by clicking the link below.

goldfish-models afb_arab_full online free url in huggingface.co:

https://huggingface.co/goldfish-models/afb_arab_full

afb_arab_full install

afb_arab_full is an open source model from GitHub that offers a free installation service, and any user can find afb_arab_full on GitHub to install. At the same time, huggingface.co provides the effect of afb_arab_full install, users can directly use afb_arab_full installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

afb_arab_full install url in huggingface.co:

https://huggingface.co/goldfish-models/afb_arab_full

Url of afb_arab_full

Provider of afb_arab_full huggingface.co

goldfish-models
ORGANIZATIONS

Other API from goldfish-models