Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import pipeline
|
| 3 |
+
|
| 4 |
+
# Load the model
|
| 5 |
+
tagger = pipeline(
|
| 6 |
+
"token-classification",
|
| 7 |
+
model="boffire/kabyle-pos",
|
| 8 |
+
aggregation_strategy="simple"
|
| 9 |
+
)
|
| 10 |
+
|
| 11 |
+
# Universal Dependencies POS tagset with descriptions
|
| 12 |
+
POS_DESCRIPTIONS = {
|
| 13 |
+
"ADJ": "Adjective",
|
| 14 |
+
"ADP": "Adposition",
|
| 15 |
+
"ADV": "Adverb",
|
| 16 |
+
"AUX": "Auxiliary verb",
|
| 17 |
+
"CCONJ": "Coordinating conjunction",
|
| 18 |
+
"DET": "Determiner",
|
| 19 |
+
"INTJ": "Interjection",
|
| 20 |
+
"NOUN": "Noun",
|
| 21 |
+
"NUM": "Numeral",
|
| 22 |
+
"PART": "Particle",
|
| 23 |
+
"PRON": "Pronoun",
|
| 24 |
+
"PROPN": "Proper noun",
|
| 25 |
+
"PUNCT": "Punctuation",
|
| 26 |
+
"SCONJ": "Subordinating conjunction",
|
| 27 |
+
"SYM": "Symbol",
|
| 28 |
+
"VERB": "Verb",
|
| 29 |
+
"X": "Other"
|
| 30 |
+
}
|
| 31 |
+
|
| 32 |
+
# Color mapping for POS tags
|
| 33 |
+
POS_COLORS = {
|
| 34 |
+
"NOUN": "#e3f2fd", "PROPN": "#bbdefb", "PRON": "#90caf9",
|
| 35 |
+
"VERB": "#e8f5e9", "AUX": "#c8e6c9",
|
| 36 |
+
"ADJ": "#fff3e0", "ADV": "#ffe0b2",
|
| 37 |
+
"ADP": "#f3e5f5", "PART": "#e1bee7",
|
| 38 |
+
"DET": "#fce4ec", "NUM": "#f8bbd0",
|
| 39 |
+
"CCONJ": "#e0f7fa", "SCONJ": "#b2ebf2",
|
| 40 |
+
"INTJ": "#fff9c4", "PUNCT": "#f5f5f5",
|
| 41 |
+
"SYM": "#eeeeee", "X": "#ffccbc"
|
| 42 |
+
}
|
| 43 |
+
|
| 44 |
+
|
| 45 |
+
def tag_text(text):
|
| 46 |
+
if not text or not text.strip():
|
| 47 |
+
return "", "Please enter some Kabyle text."
|
| 48 |
+
|
| 49 |
+
try:
|
| 50 |
+
results = tagger(text.strip())
|
| 51 |
+
except Exception as e:
|
| 52 |
+
return "", f"Error: {str(e)}"
|
| 53 |
+
|
| 54 |
+
if not results:
|
| 55 |
+
return "", "No tokens found."
|
| 56 |
+
|
| 57 |
+
# Build HTML visualization
|
| 58 |
+
html_parts = ['<div style="font-size: 1.15rem; line-height: 2.2; font-family: system-ui, sans-serif; padding: 10px;">']
|
| 59 |
+
|
| 60 |
+
for token in results:
|
| 61 |
+
word = token["word"]
|
| 62 |
+
label = token["entity_group"]
|
| 63 |
+
score = token["score"]
|
| 64 |
+
desc = POS_DESCRIPTIONS.get(label, label)
|
| 65 |
+
color = POS_COLORS.get(label, "#ffffff")
|
| 66 |
+
|
| 67 |
+
block = (
|
| 68 |
+
'<span style="display: inline-block; margin: 2px; vertical-align: top;">'
|
| 69 |
+
'<span style="background: ' + color + '; border: 1px solid #ccc; border-radius: 6px 6px 0 0; padding: 4px 8px; display: block; text-align: center; font-weight: 500;">'
|
| 70 |
+
+ word +
|
| 71 |
+
'</span>'
|
| 72 |
+
'<span style="background: #333; color: #fff; border-radius: 0 0 6px 6px; padding: 2px 8px; display: block; text-align: center; font-size: 0.75rem;">'
|
| 73 |
+
+ label + ' <span style="opacity: 0.7;">(' + f"{score:.2f}" + ')</span>'
|
| 74 |
+
'</span>'
|
| 75 |
+
'</span>'
|
| 76 |
+
)
|
| 77 |
+
html_parts.append(block)
|
| 78 |
+
|
| 79 |
+
html_parts.append('</div>')
|
| 80 |
+
|
| 81 |
+
# Build markdown table
|
| 82 |
+
table_lines = [
|
| 83 |
+
"| Token | POS Tag | Description | Confidence |",
|
| 84 |
+
"|-------|---------|-------------|------------|"
|
| 85 |
+
]
|
| 86 |
+
for token in results:
|
| 87 |
+
word = token["word"]
|
| 88 |
+
label = token["entity_group"]
|
| 89 |
+
desc = POS_DESCRIPTIONS.get(label, label)
|
| 90 |
+
score = token["score"]
|
| 91 |
+
table_lines.append(f"| {word} | `{label}` | {desc} | {score:.3f} |")
|
| 92 |
+
|
| 93 |
+
return "\n".join(html_parts), "\n".join(table_lines)
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
# Example sentences in Kabyle
|
| 97 |
+
examples = [
|
| 98 |
+
"Aṭas n medden i yessen.",
|
| 99 |
+
"Taqbaylit d tutlayt tamezwarut deg Lezzayer.",
|
| 100 |
+
"Yella wuccen ameqran deg wexxam.",
|
| 101 |
+
"Tameddakt-nneɣ teɣra adlis ameqran.",
|
| 102 |
+
"D nekkni i d-yusan d imezwura.",
|
| 103 |
+
]
|
| 104 |
+
|
| 105 |
+
demo = gr.Interface(
|
| 106 |
+
fn=tag_text,
|
| 107 |
+
inputs=gr.Textbox(
|
| 108 |
+
label="Kabyle Text",
|
| 109 |
+
placeholder="Enter a sentence in Kabyle (e.g., Aṭas n medden i yessen.)",
|
| 110 |
+
lines=2
|
| 111 |
+
),
|
| 112 |
+
outputs=[
|
| 113 |
+
gr.HTML(label="Tagged Visualization"),
|
| 114 |
+
gr.Markdown(label="Results Table")
|
| 115 |
+
],
|
| 116 |
+
title="Kabyle POS Tagger",
|
| 117 |
+
description="""
|
| 118 |
+
<div style="text-align: center;">
|
| 119 |
+
<h2>Kabyle Part-of-Speech Tagger</h2>
|
| 120 |
+
<p>Enter a sentence in <strong>Kabyle</strong> (Berber language) to see POS tags predicted by
|
| 121 |
+
<a href="https://huggingface.co/boffire/kabyle-pos" target="_blank">boffire/kabyle-pos</a>
|
| 122 |
+
(XLM-RoBERTa-base, F1: 87.5%).</p>
|
| 123 |
+
<p style="font-size: 0.9rem; color: #666;">
|
| 124 |
+
Tags follow the <a href="https://universaldependencies.org/u/pos/" target="_blank">Universal Dependencies</a> POS tagset.
|
| 125 |
+
</p>
|
| 126 |
+
</div>
|
| 127 |
+
""",
|
| 128 |
+
examples=examples,
|
| 129 |
+
allow_flagging="never",
|
| 130 |
+
cache_examples=False,
|
| 131 |
+
)
|
| 132 |
+
|
| 133 |
+
if __name__ == "__main__":
|
| 134 |
+
demo.launch()
|