Spaces:
Sleeping
Sleeping
Add application file and requirements
Browse files- app.py +82 -0
- requirements.txt +3 -0
app.py
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import pipeline
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
model_name = "ituvtu/distilbert-ag-news"
|
| 6 |
+
|
| 7 |
+
id2label = {0: "World", 1: "Sports", 2: "Business", 3: "Sci/Tech"}
|
| 8 |
+
|
| 9 |
+
try:
|
| 10 |
+
pipe = pipeline("text-classification", model=model_name)
|
| 11 |
+
print(f"Model {model_name} has been successfully downloaded.")
|
| 12 |
+
except Exception as e:
|
| 13 |
+
print(f"Model loading error: {e}")
|
| 14 |
+
|
| 15 |
+
pipe = lambda text, **kwargs: [{"label": f"ERROR: {e}", "score": 0.0}]
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
def predict_news_category(text):
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
predictions = pipe(text, return_all_scores=True)
|
| 22 |
+
|
| 23 |
+
formatted_output = {}
|
| 24 |
+
|
| 25 |
+
try:
|
| 26 |
+
# {"World": 0.1, "Sports": 0.7, ...}
|
| 27 |
+
if predictions and predictions[0]:
|
| 28 |
+
for pred in predictions[0]:
|
| 29 |
+
label_num_str = pred['label']
|
| 30 |
+
score = pred['score']
|
| 31 |
+
|
| 32 |
+
label_index = int(label_num_str.split('_')[-1])
|
| 33 |
+
readable_label = id2label.get(label_index, "Unknown")
|
| 34 |
+
|
| 35 |
+
formatted_output[readable_label] = score
|
| 36 |
+
|
| 37 |
+
except Exception as e:
|
| 38 |
+
return {f"Помилка обробки: {e}": 0.0}
|
| 39 |
+
|
| 40 |
+
return formatted_output
|
| 41 |
+
|
| 42 |
+
|
| 43 |
+
examples = [
|
| 44 |
+
# 0: World
|
| 45 |
+
["Global leaders meet for climate change summit."],
|
| 46 |
+
["Tensions rise in Middle East after new treaty."],
|
| 47 |
+
["Brexit trade deal negotiations stall over fishing rights."],
|
| 48 |
+
|
| 49 |
+
# 1: Sports
|
| 50 |
+
["Olympic athlete breaks world record in 100m sprint."],
|
| 51 |
+
["Local football team wins championship."],
|
| 52 |
+
["Manchester United announces record-breaking sponsorship deal."],
|
| 53 |
+
|
| 54 |
+
# 2: Business
|
| 55 |
+
["Federal Reserve to announce interest rate decision."],
|
| 56 |
+
["Stock market hits all-time high amid tech boom."],
|
| 57 |
+
["Tesla unveils new battery technology, stock soars."],
|
| 58 |
+
|
| 59 |
+
# 3: Sci/Tech
|
| 60 |
+
["New discovery on Mars could change everything."],
|
| 61 |
+
["Researchers develop new AI capable of writing code."],
|
| 62 |
+
["Massive cybersecurity breach exposes government data."]
|
| 63 |
+
]
|
| 64 |
+
|
| 65 |
+
# Interface
|
| 66 |
+
title = "News classifier (AG News)"
|
| 67 |
+
description = "Enter a news headline, and the DistilBERT model (trained on AG News) will determine its category. Try one of the 12 examples below to test the model!"
|
| 68 |
+
|
| 69 |
+
iface = gr.Interface(
|
| 70 |
+
fn=predict_news_category,
|
| 71 |
+
|
| 72 |
+
inputs=gr.Textbox(lines=3, placeholder="Enter the headline of the news here..."),
|
| 73 |
+
|
| 74 |
+
outputs=gr.Label(),
|
| 75 |
+
|
| 76 |
+
title=title,
|
| 77 |
+
description=description,
|
| 78 |
+
examples=examples
|
| 79 |
+
)
|
| 80 |
+
|
| 81 |
+
# Launch
|
| 82 |
+
iface.launch()
|
requirements.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
gradio
|
| 2 |
+
transformers
|
| 3 |
+
torch
|