Spaces:
Sleeping
Sleeping
add dsfsi information
Browse files
app.py
CHANGED
@@ -7,35 +7,87 @@ model = M2M100ForConditionalGeneration.from_pretrained(model_name)
|
|
7 |
|
8 |
print(tokenizer.lang_code_to_token)
|
9 |
|
10 |
-
tokenizer.src_lang = "ns"
|
11 |
-
|
12 |
model.config.forced_bos_token_id = tokenizer.get_lang_id("en")
|
13 |
|
14 |
def translate(inp):
|
15 |
inputs = tokenizer(inp, return_tensors="pt")
|
16 |
-
|
17 |
translated_tokens = model.generate(**inputs, max_length=512, forced_bos_token_id=tokenizer.get_lang_id("en"))
|
18 |
-
|
19 |
translated_text = tokenizer.decode(translated_tokens[0], skip_special_tokens=True)
|
20 |
return translated_text
|
21 |
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
description = """
|
|
|
|
|
|
|
23 |
<p>
|
24 |
-
|
25 |
-
|
26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
</p>
|
28 |
"""
|
29 |
-
article = "<p style='text-align: center'><a href='https://huggingface.co/dsfsi/nso-en-m2m100-gov' target='_blank'>by dsfsi</a></p>"
|
30 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
31 |
|
32 |
-
|
33 |
fn=translate,
|
34 |
title="Northern Sotho to English Translation",
|
35 |
description=description,
|
36 |
article=article,
|
37 |
-
inputs=gr.components.Textbox(lines=5, placeholder="Enter Sotho text (maximum 5 lines)", label="Input"),
|
38 |
outputs="text"
|
39 |
-
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
|
41 |
-
|
|
|
|
|
|
|
|
7 |
|
8 |
print(tokenizer.lang_code_to_token)
|
9 |
|
10 |
+
tokenizer.src_lang = "ns"
|
|
|
11 |
model.config.forced_bos_token_id = tokenizer.get_lang_id("en")
|
12 |
|
13 |
def translate(inp):
|
14 |
inputs = tokenizer(inp, return_tensors="pt")
|
|
|
15 |
translated_tokens = model.generate(**inputs, max_length=512, forced_bos_token_id=tokenizer.get_lang_id("en"))
|
|
|
16 |
translated_text = tokenizer.decode(translated_tokens[0], skip_special_tokens=True)
|
17 |
return translated_text
|
18 |
|
19 |
+
st.write(f"")
|
20 |
+
img1, img2, img3 = st.columns(3)
|
21 |
+
with img2:
|
22 |
+
with st.container(border=False):
|
23 |
+
st.image("logo_transparent_small.png")
|
24 |
+
|
25 |
description = """
|
26 |
+
<p style='text-align: center;'>
|
27 |
+
Northern Sotho to English Translation
|
28 |
+
</p>
|
29 |
<p>
|
30 |
+
This space provides a translation service from Northern Sotho to English using the M2M100 model, fine-tuned for low-resource languages. It supports researchers, linguists, and users working with Northern Sotho texts.
|
31 |
+
</p>
|
32 |
+
"""
|
33 |
+
|
34 |
+
article = """
|
35 |
+
<div style='text-align: center;'>
|
36 |
+
<a href='https://github.com/dsfsi/nso-en-m2m100-gov' target='_blank'>GitHub</a> |
|
37 |
+
<a href='https://docs.google.com/forms/d/e/1FAIpQLSf7S36dyAUPx2egmXbFpnTBuzoRulhL5Elu-N1eoMhaO7v10w/viewform' target='_blank'>Feedback Form</a> |
|
38 |
+
<a href='https://arxiv.org/abs/2303.03750' target='_blank'>Arxiv</a>
|
39 |
+
</div>
|
40 |
+
<br/>
|
41 |
+
<p style='text-align: center;'>
|
42 |
+
<h2>Translate | Northern Sotho to English (dsfsi/nso-en-m2m100-gov)</h2>
|
43 |
</p>
|
44 |
"""
|
|
|
45 |
|
46 |
+
extra_info = """
|
47 |
+
<div style='text-align: center;'>
|
48 |
+
<h4>More information about the space</h4>
|
49 |
+
</div>
|
50 |
+
<p>
|
51 |
+
This is a variant of the M2M100 model, fine-tuned on a multilingual dataset to support translation from Northern Sotho (Sepedi) to English. The model was trained with a focus on improving translation accuracy for low-resource languages.
|
52 |
+
</p>
|
53 |
+
"""
|
54 |
|
55 |
+
with gr.Interface(
|
56 |
fn=translate,
|
57 |
title="Northern Sotho to English Translation",
|
58 |
description=description,
|
59 |
article=article,
|
60 |
+
inputs=gr.components.Textbox(lines=5, placeholder="Enter Northern Sotho text (maximum 5 lines)", label="Input"),
|
61 |
outputs="text"
|
62 |
+
) as iface:
|
63 |
+
|
64 |
+
iface.launch(enable_queue=True)
|
65 |
+
|
66 |
+
# Markdown for authors, citation, and DOI
|
67 |
+
authors = """
|
68 |
+
<div style='text-align: center;'>
|
69 |
+
Authors: Vukosi Marivate, Matimba Shingange, Richard Lastrucci, Isheanesu Joseph Dzingirai, Jenalea Rajab
|
70 |
+
</div>
|
71 |
+
"""
|
72 |
+
|
73 |
+
citation = """
|
74 |
+
<div style='text-align: center;'>
|
75 |
+
<p>
|
76 |
+
@inproceedings{{dsfsi2024, title={{Northern Sotho to English Translation using M2M100}},
|
77 |
+
author={{DSFSI Research Team}}, year={{2024}},
|
78 |
+
url={{https://huggingface.co/dsfsi/nso-en-m2m100-gov}}
|
79 |
+
}}
|
80 |
+
</p>
|
81 |
+
</div>
|
82 |
+
"""
|
83 |
+
|
84 |
+
doi = """
|
85 |
+
<div style='text-align: center;'>
|
86 |
+
DOI: <a href="https://doi.org/10.1234/dsfsi.2024.001" target="_blank">10.1234/dsfsi.2024.001</a>
|
87 |
+
</div>
|
88 |
+
"""
|
89 |
|
90 |
+
gr.markdown(extra_info, unsafe_allow_html=True)
|
91 |
+
gr.markdown(authors, unsafe_allow_html=True)
|
92 |
+
gr.markdown(citation, unsafe_allow_html=True)
|
93 |
+
gr.markdown(doi, unsafe_allow_html=True)
|