Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,131 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import torch
|
| 3 |
+
from transformers import MBartForConditionalGeneration, MBart50TokenizerFast
|
| 4 |
+
|
| 5 |
+
# Load Facebook's mBART-50 model
|
| 6 |
+
model_name = "facebook/mbart-large-50-many-to-many-mmt"
|
| 7 |
+
tokenizer = MBart50TokenizerFast.from_pretrained(model_name)
|
| 8 |
+
model = MBartForConditionalGeneration.from_pretrained(model_name)
|
| 9 |
+
|
| 10 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 11 |
+
model.to(device)
|
| 12 |
+
|
| 13 |
+
def translate(text, source_lang, target_lang):
|
| 14 |
+
if not text.strip():
|
| 15 |
+
return "Please enter text to translate"
|
| 16 |
+
|
| 17 |
+
# Language codes for mBART-50
|
| 18 |
+
lang_map = {
|
| 19 |
+
"Norwegian": "nb_NO",
|
| 20 |
+
"English": "en_XX"
|
| 21 |
+
}
|
| 22 |
+
|
| 23 |
+
try:
|
| 24 |
+
# Set source language
|
| 25 |
+
tokenizer.src_lang = lang_map[source_lang]
|
| 26 |
+
|
| 27 |
+
# Tokenize
|
| 28 |
+
encoded = tokenizer(text, return_tensors="pt", truncation=True, max_length=512)
|
| 29 |
+
encoded = encoded.to(device)
|
| 30 |
+
|
| 31 |
+
# Generate translation
|
| 32 |
+
generated = model.generate(
|
| 33 |
+
**encoded,
|
| 34 |
+
forced_bos_token_id=tokenizer.lang_code_to_id[lang_map[target_lang]],
|
| 35 |
+
max_length=512,
|
| 36 |
+
num_beams=5,
|
| 37 |
+
early_stopping=True
|
| 38 |
+
)
|
| 39 |
+
|
| 40 |
+
# Decode result
|
| 41 |
+
result = tokenizer.batch_decode(generated, skip_special_tokens=True)[0]
|
| 42 |
+
return result
|
| 43 |
+
|
| 44 |
+
except Exception as e:
|
| 45 |
+
return f"Translation error: {str(e)}"
|
| 46 |
+
|
| 47 |
+
def swap_languages(source, target, text):
|
| 48 |
+
return target, source, ""
|
| 49 |
+
|
| 50 |
+
# Create Gradio interface
|
| 51 |
+
with gr.Blocks(
|
| 52 |
+
title="π³π΄ βοΈ π¬π§ Norwegian-English Translator",
|
| 53 |
+
theme=gr.themes.Soft()
|
| 54 |
+
) as demo:
|
| 55 |
+
|
| 56 |
+
gr.Markdown("""
|
| 57 |
+
# π Advanced Norwegian-English Translator
|
| 58 |
+
### Powered by Facebook's mBART-50 Model
|
| 59 |
+
""")
|
| 60 |
+
|
| 61 |
+
with gr.Row():
|
| 62 |
+
source_lang = gr.Dropdown(
|
| 63 |
+
choices=["Norwegian", "English"],
|
| 64 |
+
value="Norwegian",
|
| 65 |
+
label="Source Language"
|
| 66 |
+
)
|
| 67 |
+
|
| 68 |
+
swap_btn = gr.Button("π", size="sm")
|
| 69 |
+
|
| 70 |
+
target_lang = gr.Dropdown(
|
| 71 |
+
choices=["Norwegian", "English"],
|
| 72 |
+
value="English",
|
| 73 |
+
label="Target Language"
|
| 74 |
+
)
|
| 75 |
+
|
| 76 |
+
with gr.Row():
|
| 77 |
+
with gr.Column():
|
| 78 |
+
input_text = gr.Textbox(
|
| 79 |
+
lines=8,
|
| 80 |
+
placeholder="Enter text to translate...",
|
| 81 |
+
label="Input Text",
|
| 82 |
+
max_lines=15
|
| 83 |
+
)
|
| 84 |
+
|
| 85 |
+
with gr.Column():
|
| 86 |
+
output_text = gr.Textbox(
|
| 87 |
+
lines=8,
|
| 88 |
+
label="Translation",
|
| 89 |
+
interactive=False,
|
| 90 |
+
max_lines=15
|
| 91 |
+
)
|
| 92 |
+
|
| 93 |
+
translate_btn = gr.Button("π Translate", variant="primary", size="lg")
|
| 94 |
+
|
| 95 |
+
# Event handlers
|
| 96 |
+
translate_btn.click(
|
| 97 |
+
fn=translate,
|
| 98 |
+
inputs=[input_text, source_lang, target_lang],
|
| 99 |
+
outputs=output_text
|
| 100 |
+
)
|
| 101 |
+
|
| 102 |
+
swap_btn.click(
|
| 103 |
+
fn=swap_languages,
|
| 104 |
+
inputs=[source_lang, target_lang, input_text],
|
| 105 |
+
outputs=[source_lang, target_lang, input_text]
|
| 106 |
+
)
|
| 107 |
+
|
| 108 |
+
input_text.submit(
|
| 109 |
+
fn=translate,
|
| 110 |
+
inputs=[input_text, source_lang, target_lang],
|
| 111 |
+
outputs=output_text
|
| 112 |
+
)
|
| 113 |
+
|
| 114 |
+
gr.Markdown("""
|
| 115 |
+
### π Examples
|
| 116 |
+
**Norwegian to English:**
|
| 117 |
+
- "Hei, hvordan har du det?" β "Hello, how are you?"
|
| 118 |
+
- "Jeg kommer fra Norge" β "I come from Norway"
|
| 119 |
+
|
| 120 |
+
**English to Norwegian:**
|
| 121 |
+
- "Thank you very much" β "Tusen takk"
|
| 122 |
+
- "Good morning" β "God morgen"
|
| 123 |
+
""")
|
| 124 |
+
|
| 125 |
+
gr.Markdown("""
|
| 126 |
+
---
|
| 127 |
+
*Built with β€οΈ using Facebook's mBART-50 and Gradio*
|
| 128 |
+
""")
|
| 129 |
+
|
| 130 |
+
if __name__ == "__main__":
|
| 131 |
+
demo.launch(share=True)
|