|
import os |
|
import yaml |
|
import gdown |
|
import gradio as gr |
|
from predict import PredictTri |
|
from gradio import blocks |
|
|
|
output_path = "tashkeela-d2.pt" |
|
gdrive_templ = "https://drive.google.com/file/d/{}/view?usp=sharing" |
|
if not os.path.exists(output_path): |
|
model_gdrive_id = "1FGelqImFkESbTyRsx_elkKIOZ9VbhRuo" |
|
gdown.download(gdrive_templ.format(model_gdrive_id), output=output_path, quiet=False) |
|
|
|
output_path = "vocab.vec" |
|
if not os.path.exists(output_path): |
|
vocab_gdrive_id = "1-0muGvcSYEf8RAVRcwXay4MRex6kmCii" |
|
gdown.download(gdrive_templ.format(vocab_gdrive_id), output=output_path, quiet=False) |
|
|
|
with open("config.yaml", 'r', encoding="utf-8") as file: |
|
config = yaml.load(file, Loader=yaml.FullLoader) |
|
|
|
config["train"]["max-sent-len"] = config["predictor"]["window"] |
|
config["train"]["max-token-count"] = config["predictor"]["window"] * 3 |
|
|
|
def diacritze(text, do_partial): |
|
predictor = PredictTri(config, text) |
|
diacritized_lines = predictor.predict_partial(do_partial=do_partial) |
|
return diacritized_lines |
|
|
|
with gr.Blocks() as demo: |
|
gr.Markdown( |
|
""" |
|
# Partial Diacritization: A Context-Contrastive Inference Approach |
|
## Authors: Muhammad ElNokrashy, Badr AlKhamissi |
|
""") |
|
|
|
with gr.Row(): |
|
check_box = gr.Checkbox(label="Partial", info="Apply Partial Diacritics or Full Diacritics") |
|
threshold_txt = gr.Textbox("") |
|
|
|
input_txt = gr.Textbox( |
|
placeholder="اكتب هنا", |
|
lines=5, |
|
label="Input", |
|
type='text', |
|
rtl=True, |
|
text_align='right', |
|
) |
|
|
|
output_txt = gr.Textbox( |
|
lines=5, |
|
label="Output", |
|
type='text', |
|
rtl=True, |
|
text_align='right', |
|
) |
|
|
|
btn = gr.Button(value="Shakkel") |
|
btn.click(diacritze, inputs=[input_txt, check_box], outputs=[output_txt]) |
|
|
|
if __name__ == "__main__": |
|
demo.queue().launch( |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
) |
|
|
