ParsGPT 🦁

How to use

You can use this model directly with a pipeline for text generation:

Installing requirements

pip install -U transformers
pip install -U hazm

How to generate using pipeline

from transformers import pipeline
import hazm


normalizer = hazm.Normalizer(persian_numbers=False)

def normalize_input(text):
    text = normalizer.normalize(text)
    return text

def sents_as_output(text, num_sents=1):
    sents = hazm.sent_tokenize(text)
    if num_sents > 0:
        return " ".join(sents[:num_sents])
    return " ".join(sents[0])


generator = pipeline('text-generation', "HooshvareLab/gpt2-fa")
text = "در یک اتفاق شگفت انگیز، پژوهشگران"
text = normalize_input(text)

outputs = generator(text)
for output in outputs:
    generated = output["generated_text"]
    print(sents_as_output(generated))

در یک اتفاق شگفت انگیز، پژوهشگران قصد دارند با استفاده از داده‌های حاصل از چندین تلسکوپ، عکس‌هایی با وضوح مختلف از سیاره‌ی مشتری و زحل تهیه کنند.

GPT2 Visualization

This visualization is powered by Ecco (an interactive language modeling visualization).

Notebook
Ecco Visualization

How to fine-tune

Notebook
Simple Version

Examples - Models

If you fine-tuned gpt2-fa on your dataset, share it with us. All things you need to do, create a pull request and share yours with us. We are looking forward to it.

Model	Description	How to Use
HooshvareLab/gpt2-fa-poetry	The model was fine-tuned on ChronologicalPersianPoetryDataset.
HooshvareLab/gpt2-fa-comment	The model can generate comments based on your aspects, and the model was fine-tuned on persiannlp/parsinlu. Currently, the model only supports aspects in the food and movie scope.

Citation

Please cite in publications as the following:

@misc{ParsGPT2,
  author = {Hooshvare Team},
  title = {ParsGPT2, a Persian version of GPT2},
  year = {2021},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/hooshvare/parsgpt}},
}

Questions?

Post a Github issue on the ParsGPT2 Issues repo.