Spaces:
Running
Running
Commit
·
8f8dd51
1
Parent(s):
2d50524
Update app.py
Browse files
app.py
CHANGED
|
@@ -47,44 +47,67 @@
|
|
| 47 |
|
| 48 |
# iface.launch()
|
| 49 |
|
| 50 |
-
from transformers import MarianMTModel, MarianTokenizer, pipeline
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
|
| 53 |
-
#
|
| 54 |
-
|
| 55 |
-
"
|
| 56 |
-
|
| 57 |
-
"Xhosa": "Helsinki-NLP/opus-mt-en-xh",
|
| 58 |
-
"Sesotho": "Helsinki-NLP/opus-mt-en-st",
|
| 59 |
-
"Setswana": "Helsinki-NLP/opus-mt-en-tn",
|
| 60 |
-
}
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
# Translation function
|
| 64 |
-
def translate(text, target_language):
|
| 65 |
-
model_name = language_models[target_language]
|
| 66 |
-
tokenizer = MarianTokenizer.from_pretrained(model_name)
|
| 67 |
-
model = MarianMTModel.from_pretrained(model_name)
|
| 68 |
-
|
| 69 |
-
# Setup pipeline
|
| 70 |
-
translation_pipeline = pipeline("translation", model=model, tokenizer=tokenizer)
|
| 71 |
-
|
| 72 |
-
# Translate
|
| 73 |
-
result = translation_pipeline(text)
|
| 74 |
-
return result[0]["translation_text"]
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
# Build Gradio interface
|
| 78 |
-
interface = gr.Interface(
|
| 79 |
-
fn=translate,
|
| 80 |
-
inputs=[
|
| 81 |
-
gr.Textbox(label="Enter English Text"),
|
| 82 |
-
gr.Dropdown(choices=list(language_models.keys()), label="Translate to"),
|
| 83 |
-
],
|
| 84 |
-
outputs="text",
|
| 85 |
-
title="African Language Translator",
|
| 86 |
-
description="Translate English text into Afrikaans, Zulu, Xhosa, Sesotho or Setswana",
|
| 87 |
-
)
|
| 88 |
-
|
| 89 |
-
# Launch the app
|
| 90 |
-
interface.launch()
|
|
|
|
| 47 |
|
| 48 |
# iface.launch()
|
| 49 |
|
| 50 |
+
# from transformers import MarianMTModel, MarianTokenizer, pipeline
|
| 51 |
+
# import gradio as gr
|
| 52 |
+
|
| 53 |
+
# # Define supported models for South African languages
|
| 54 |
+
# language_models = {
|
| 55 |
+
# "Afrikaans": "Helsinki-NLP/opus-mt-en-af",
|
| 56 |
+
# "Zulu": "Helsinki-NLP/opus-mt-en-zu",
|
| 57 |
+
# "Xhosa": "Helsinki-NLP/opus-mt-en-xh",
|
| 58 |
+
# "Sesotho": "Helsinki-NLP/opus-mt-en-st",
|
| 59 |
+
# "Setswana": "Helsinki-NLP/opus-mt-en-tn",
|
| 60 |
+
# }
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
# # Translation function
|
| 64 |
+
# def translate(text, target_language):
|
| 65 |
+
# model_name = language_models[target_language]
|
| 66 |
+
# tokenizer = MarianTokenizer.from_pretrained(model_name)
|
| 67 |
+
# model = MarianMTModel.from_pretrained(model_name)
|
| 68 |
+
|
| 69 |
+
# # Setup pipeline
|
| 70 |
+
# translation_pipeline = pipeline("translation", model=model, tokenizer=tokenizer)
|
| 71 |
+
|
| 72 |
+
# # Translate
|
| 73 |
+
# result = translation_pipeline(text)
|
| 74 |
+
# return result[0]["translation_text"]
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
# # Build Gradio interface
|
| 78 |
+
# interface = gr.Interface(
|
| 79 |
+
# fn=translate,
|
| 80 |
+
# inputs=[
|
| 81 |
+
# gr.Textbox(label="Enter English Text"),
|
| 82 |
+
# gr.Dropdown(choices=list(language_models.keys()), label="Translate to"),
|
| 83 |
+
# ],
|
| 84 |
+
# outputs="text",
|
| 85 |
+
# title="African Language Translator",
|
| 86 |
+
# description="Translate English text into Afrikaans, Zulu, Xhosa, Sesotho or Setswana",
|
| 87 |
+
# )
|
| 88 |
+
|
| 89 |
+
# # Launch the app
|
| 90 |
+
# interface.launch()
|
| 91 |
+
|
| 92 |
+
|
| 93 |
import gradio as gr
|
| 94 |
+
from transformers import MarianMTModel, MarianTokenizer
|
| 95 |
+
|
| 96 |
+
# Pick your language pair
|
| 97 |
+
model_name = "Helsinki-NLP/opus-mt-en-zu" # English to Zulu
|
| 98 |
+
|
| 99 |
+
# Load tokenizer and model locally
|
| 100 |
+
tokenizer = MarianTokenizer.from_pretrained(model_name)
|
| 101 |
+
model = MarianMTModel.from_pretrained(model_name)
|
| 102 |
+
|
| 103 |
+
|
| 104 |
+
def translate(text):
|
| 105 |
+
inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
|
| 106 |
+
translated = model.generate(**inputs)
|
| 107 |
+
return tokenizer.decode(translated[0], skip_special_tokens=True)
|
| 108 |
+
|
| 109 |
|
| 110 |
+
# Gradio interface
|
| 111 |
+
gr.Interface(
|
| 112 |
+
fn=translate, inputs="text", outputs="text", title="English to Zulu Translator"
|
| 113 |
+
).launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|