Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,61 +1,75 @@
|
|
1 |
import os
|
2 |
os.system("pip install torch transformers gradio matplotlib")
|
3 |
|
|
|
|
|
4 |
|
5 |
import torch
|
6 |
import gradio as gr
|
7 |
-
import numpy as np
|
8 |
import matplotlib.pyplot as plt
|
9 |
-
import
|
10 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
11 |
|
12 |
-
|
|
|
|
|
|
|
13 |
|
14 |
-
#
|
15 |
-
model_path = "HyperX-Sentience/RogueBERT-Toxicity-85K"
|
16 |
-
model = AutoModelForSequenceClassification.from_pretrained(model_path)
|
17 |
-
tokenizer = AutoTokenizer.from_pretrained(model_path)
|
18 |
-
|
19 |
-
# Move the model to CUDA if available
|
20 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
21 |
model.to(device)
|
22 |
|
23 |
-
#
|
24 |
labels = ["toxic", "severe_toxic", "obscene", "threat", "insult", "identity_hate"]
|
25 |
|
|
|
26 |
def predict_toxicity(comment):
|
27 |
-
|
28 |
-
inputs = tokenizer(comment, truncation=True, padding="max_length", max_length=128, return_tensors="pt")
|
29 |
inputs = {key: val.to(device) for key, val in inputs.items()}
|
30 |
|
31 |
with torch.no_grad():
|
32 |
outputs = model(**inputs)
|
33 |
-
|
|
|
34 |
|
35 |
-
|
|
|
36 |
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
|
43 |
-
# Gradio
|
44 |
demo = gr.Interface(
|
45 |
-
fn=
|
46 |
-
inputs=gr.Textbox(label="Enter a comment
|
47 |
-
outputs=gr.
|
48 |
-
|
49 |
-
|
50 |
-
y="Score",
|
51 |
-
title="Toxicity Analysis",
|
52 |
-
y_lim=[0, 1],
|
53 |
-
color="blue",
|
54 |
-
label="Toxicity Scores",
|
55 |
-
interactive=False
|
56 |
-
),
|
57 |
-
title="Toxicity Detection with RogueBERT",
|
58 |
-
description="Enter a comment to analyze its toxicity levels. The results will be displayed as a modern bar chart."
|
59 |
)
|
60 |
|
61 |
-
|
|
|
|
|
|
1 |
import os
|
2 |
os.system("pip install torch transformers gradio matplotlib")
|
3 |
|
4 |
+
# Install required packages
|
5 |
+
# !pip install torch transformers gradio matplotlib
|
6 |
|
7 |
import torch
|
8 |
import gradio as gr
|
|
|
9 |
import matplotlib.pyplot as plt
|
10 |
+
import numpy as np
|
11 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
12 |
|
13 |
+
# Load model and tokenizer from Hugging Face Hub
|
14 |
+
model_name = "HyperX-Sentience/RogueBERT-Toxicity-85K"
|
15 |
+
model = AutoModelForSequenceClassification.from_pretrained(model_name)
|
16 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
17 |
|
18 |
+
# Move model to CUDA if available
|
|
|
|
|
|
|
|
|
|
|
19 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
20 |
model.to(device)
|
21 |
|
22 |
+
# Toxicity category labels
|
23 |
labels = ["toxic", "severe_toxic", "obscene", "threat", "insult", "identity_hate"]
|
24 |
|
25 |
+
# Function to predict toxicity
|
26 |
def predict_toxicity(comment):
|
27 |
+
inputs = tokenizer([comment], truncation=True, padding="max_length", max_length=128, return_tensors="pt")
|
|
|
28 |
inputs = {key: val.to(device) for key, val in inputs.items()}
|
29 |
|
30 |
with torch.no_grad():
|
31 |
outputs = model(**inputs)
|
32 |
+
logits = outputs.logits
|
33 |
+
probabilities = torch.sigmoid(logits).cpu().numpy()[0]
|
34 |
|
35 |
+
toxicity_scores = {label: float(probabilities[i]) for i, label in enumerate(labels)}
|
36 |
+
return toxicity_scores
|
37 |
|
38 |
+
# Function to create a bar chart
|
39 |
+
def plot_toxicity(comment):
|
40 |
+
toxicity_scores = predict_toxicity(comment)
|
41 |
+
categories = list(toxicity_scores.keys())
|
42 |
+
scores = list(toxicity_scores.values())
|
43 |
+
|
44 |
+
plt.figure(figsize=(8, 5), facecolor='black')
|
45 |
+
ax = plt.gca()
|
46 |
+
ax.set_facecolor('black')
|
47 |
+
bars = plt.bar(categories, scores, color='#20B2AA', edgecolor='white') # Sea green
|
48 |
+
|
49 |
+
plt.xticks(color='white', fontsize=12)
|
50 |
+
plt.yticks(color='white', fontsize=12)
|
51 |
+
plt.title("Toxicity Score Analysis", color='white', fontsize=14)
|
52 |
+
plt.ylim(0, 1)
|
53 |
+
|
54 |
+
for bar in bars:
|
55 |
+
yval = bar.get_height()
|
56 |
+
plt.text(bar.get_x() + bar.get_width()/2, yval + 0.02, f'{yval:.2f}', ha='center', color='white', fontsize=10)
|
57 |
+
|
58 |
+
plt.tight_layout()
|
59 |
+
plt.savefig("toxicity_chart.png", facecolor='black')
|
60 |
+
plt.close()
|
61 |
+
|
62 |
+
return "toxicity_chart.png"
|
63 |
|
64 |
+
# Gradio UI
|
65 |
demo = gr.Interface(
|
66 |
+
fn=plot_toxicity,
|
67 |
+
inputs=gr.Textbox(label="Enter a comment"),
|
68 |
+
outputs=gr.Image(type="file", label="Toxicity Analysis"),
|
69 |
+
title="Toxicity Detector",
|
70 |
+
description="Enter a comment to analyze its toxicity scores across different categories.",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
)
|
72 |
|
73 |
+
# Launch the Gradio app
|
74 |
+
if __name__ == "__main__":
|
75 |
+
demo.launch()
|