Hugging Face's logo

Spaces:
H-Layba
/
urdu-translator
Paused

App Files Community
urdu-translator
File size: 5,655 Bytes
3b2ad2b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch

MODEL_NAME = "Helsinki-NLP/opus-mt-ur-en"
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME)

EXAMPLES = [
    ["پاکستان ایک خوبصورت ملک ہے۔ اس کے پہاڑ اور دریا بہت مشہور ہیں۔"],
    ["اسلام آباد پاکستان کا دارالحکومت ہے۔ یہ ایک جدید اور صاف شہر ہے۔"],
    ["کرکٹ پاکستان کا سب سے مقبول کھیل ہے۔ پاکستانی عوام کرکٹ سے بہت محبت کرتے ہیں۔"],
    ["تعلیم ہر انسان کا بنیادی حق ہے۔ علم کی روشنی سے دنیا کو بہتر بنایا جا سکتا ہے۔"],
]

def translate(text, num_beams):
    if not text.strip():
        return "", 0

    inputs = tokenizer(
        text.strip(),
        return_tensors="pt",
        truncation=True,
        max_length=512
    )

    with torch.no_grad():
        outputs = model.generate(
            inputs["input_ids"],
            max_length=256,
            num_beams=int(num_beams),
            early_stopping=True
        )

    translation = tokenizer.decode(outputs[0], skip_special_tokens=True)
    word_count = len(translation.split())
    return translation, word_count


css = """
@import url('https://fonts.googleapis.com/css2?family=Noto+Nastaliq+Urdu&family=Inter:wght@400;500;600&display=swap');

body, .gradio-container { background: #0a0a0f !important; font-family: 'Inter', sans-serif; }

#title { text-align: center; padding: 2rem 1rem 1rem; }
#title h1 { font-size: 2rem; font-weight: 600; color: #e4e4f0; margin-bottom: 0.4rem; }
#title p { color: #6b6b8a; font-size: 0.9rem; }

#stats {
    display: flex; justify-content: center; gap: 2rem;
    padding: 0.5rem; margin-bottom: 1rem;
}
.stat-card {
    background: #12121e; border: 1px solid #2a2a3e;
    border-radius: 10px; padding: 0.6rem 1.2rem; text-align: center;
}
.stat-num { font-size: 1.2rem; font-weight: 600; color: #3b82f6; }
.stat-label { font-size: 0.7rem; color: #6b6b8a; margin-top: 2px; }

.urdu-box textarea {
    font-family: 'Noto Nastaliq Urdu', serif !important;
    font-size: 1.2rem !important; direction: rtl !important;
    text-align: right !important; line-height: 2.2 !important;
    background: #12121e !important; border: 1px solid #2a2a3e !important;
    color: #e4e4f0 !important; border-radius: 10px !important;
}
.urdu-box textarea:focus {
    border-color: #3b82f6 !important;
    box-shadow: 0 0 0 2px rgba(59, 130, 246, 0.15) !important;
}
.english-box textarea {
    font-family: 'Inter', sans-serif !important;
    font-size: 1.1rem !important;
    background: #12121e !important; border: 1px solid #2a2a3e !important;
    color: #60a5fa !important; border-radius: 10px !important;
    line-height: 1.8 !important;
}

button.primary {
    background: #3b82f6 !important; border: none !important;
    color: white !important; font-weight: 500 !important;
    border-radius: 8px !important;
}
button.primary:hover { background: #2563eb !important; }

footer { display: none !important; }
"""

with gr.Blocks(css=css, theme=gr.themes.Base()) as demo:

    gr.HTML("""
        <div id="title">
            <h1>🌐 Urdu → English Translator</h1>
            <p>اردو سے انگریزی ترجمہ — Helsinki-NLP opus-mt-ur-en trained on 2.7M sentence pairs</p>
        </div>
        <div id="stats">
            <div class="stat-card">
                <div class="stat-num">2.7M</div>
                <div class="stat-label">Training sentence pairs</div>
            </div>
            <div class="stat-card">
                <div class="stat-num">opus-mt</div>
                <div class="stat-label">Model architecture</div>
            </div>
            <div class="stat-card">
                <div class="stat-num">Marian</div>
                <div class="stat-label">Framework</div>
            </div>
        </div>
    """)

    with gr.Row():
        with gr.Column():
            input_text = gr.Textbox(
                label="اردو متن — Urdu Text",
                placeholder="یہاں اردو میں لکھیں یا پیسٹ کریں...",
                lines=7,
                elem_classes="urdu-box"
            )
            num_beams = gr.Slider(
                label="Beam search width (higher = better quality, slower)",
                minimum=1, maximum=8, value=4, step=1
            )
            translate_btn = gr.Button("ترجمہ کریں — Translate", variant="primary")

        with gr.Column():
            output_text = gr.Textbox(
                label="English Translation",
                lines=7,
                interactive=False,
                elem_classes="english-box"
            )
            word_count = gr.Number(label="Translation word count", interactive=False)

    gr.Examples(
        examples=EXAMPLES,
        inputs=input_text,
        label="مثالیں — Try these examples"
    )

    gr.HTML("""
        <div style="text-align:center; padding:1.5rem; color:#3a3a5a; font-size:0.8rem;">
            Built by <a href="https://huggingface.co/H-Layba" style="color:#3b82f6">H-Layba</a> ·
            Model: Helsinki-NLP/opus-mt-ur-en · OPUS parallel corpus
        </div>
    """)

    translate_btn.click(
        fn=translate,
        inputs=[input_text, num_beams],
        outputs=[output_text, word_count]
    )
    input_text.submit(
        fn=translate,
        inputs=[input_text, num_beams],
        outputs=[output_text, word_count]
    )

demo.launch()