File size: 1,841 Bytes
0834331 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 |
import gradio as gr
from transformers import pipeline
prefix = "<2id> "
madlad = pipeline("translation", model="google/madlad400-3b-mt")
lulu = pipeline("translation", model="tirtohadi/lulu7")
def translate(text):
# Split input text into paragraphs
paragraphs = text.split("\n\n") # Assuming paragraphs are separated by two newline characters
# Translate each paragraph
translated_paragraphs_lulu = []
translated_paragraphs_madlad = []
for paragraph in paragraphs:
# Call your custom model here to translate each paragraph
translated_paragraph_madlad = madlad(prefix + paragraph, max_length=400)[0]["translation_text"]
translated_paragraphs_madlad.append(translated_paragraph_madlad)
translated_paragraph_lulu = lulu(paragraph, max_length=400)[0]["translation_text"]
translated_paragraphs_lulu.append(translated_paragraph_lulu)
# Join translated paragraphs back into text
translated_text_lulu = "\n\n".join(translated_paragraphs_lulu)
translated_text_madlad = "\n\n".join(translated_paragraphs_madlad)
return translated_text_lulu,translated_text_madlad
with gr.Blocks() as demo:
gr.HTML("<h2>Lulu - Google Comparison</h2>")
gr.Markdown("This app compares translations between Lulu (Christian domain specific trained) and Google (Madlad-400-3B-MT)")
with gr.Row():
input_text1 = gr.Textbox(label="English Text",lines=5)
output_lulu = gr.Textbox(label="Indonesian - Lulu",lines=5)
output_madlad = gr.Textbox(label="Indonesian - Google",lines=5)
with gr.Row():
with gr.Column(scale=2):
btn = gr.Button("Translate")
btn.click(fn=translate, inputs=input_text1, outputs=[output_lulu,output_madlad], api_name="translate")
demo.launch() |