Goldfish is a suite of monolingual language models trained for 350 languages.
This model is the
Malagasy
(Latin script) model trained on 720MB of data (all our data in the language), after accounting for an estimated byte premium of 1.27; content-matched text in Malagasy takes on average 1.27x 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: mlg_latn is a
macrolanguage
code. Individual language codes plt_latn (Plateau Malagasy) and tdx_latn (Tandroy-Mahafaly Malagasy) are included in Goldfish, although with less data.
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): 913.08MB
Training text data (byte premium scaled): 720.805MB
Training tokens: 210497024 (x10 epochs)
Vocabulary size: 50000
Compute cost: 1.074300621815808e+18 FLOPs or ~101.6 NVIDIA A6000 GPU hours
Training datasets (percentages prior to deduplication):
@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 mlg_latn_full on huggingface.co
30
Total runs
0
24-hour runs
0
3-day runs
15
7-day runs
24
30-day runs
More Information About mlg_latn_full huggingface.co Model
mlg_latn_full huggingface.co is an AI model on huggingface.co that provides mlg_latn_full's model effect (), which can be used instantly with this goldfish-models mlg_latn_full model. huggingface.co supports a free trial of the mlg_latn_full model, and also provides paid use of the mlg_latn_full. Support call mlg_latn_full model through api, including Node.js, Python, http.
mlg_latn_full huggingface.co is an online trial and call api platform, which integrates mlg_latn_full's modeling effects, including api services, and provides a free online trial of mlg_latn_full, you can try mlg_latn_full online for free by clicking the link below.
goldfish-models mlg_latn_full online free url in huggingface.co:
mlg_latn_full is an open source model from GitHub that offers a free installation service, and any user can find mlg_latn_full on GitHub to install. At the same time, huggingface.co provides the effect of mlg_latn_full install, users can directly use mlg_latn_full installed effect in huggingface.co for debugging and trial. It also supports api for free installation.