metadata
language: ar
Sanaa-Dialect
Finetuned Arabic GPT-2 demo
This is a small GPT-2 model, originally trained on Arabic Wikipedia circa September 2020 , finetuned on part of University of British Columbia's Arabic Dialect dataset https://github.com/UBC-NLP/aoc_id
You can use special tokens to set four dialects: [EGYPTIAN], [GULF], [LEVANTINE], and [MSA]
from simpletransformers.language_generation import LanguageGenerationModel
model = LanguageGenerationModel("gpt2", "monsoon-nlp/sanaa-dialect")
model.generate('[GULF]' + "ู
ุฏููุชู ูู", { 'max_length': 100 })
There is NO content filtering in the current version; do not use for public-facing text generation!
Training and Finetuning details
Original model and training: https://huggingface.co/monsoon-nlp/sanaa
I inserted new tokens into the tokenizer, finetuned the model on the dialect samples, and exported the new model.
Notebook: https://colab.research.google.com/drive/1Z_935vTuZvbseOsExCjSprrqn1MsQT57