🛡️ ChaosGuard AI Dashboard
One-page demo: XLM-R multi-label toxicity + uncertainty (MC Dropout) + XAI (Integrated Gradients).
No data stored
Bangla/Banglish/English
Input
Text
আপনি একটা বাজে মানুষ
Threshold
Max length
MC samples
Analyze
Predictions
Probability, uncertainty (std), entropy, and toxic flag.
Visuals
Charts update after analysis.
Explain label
{% for l in labels %}
{{l}}
{% endfor %}
IG steps
Explain
Legend:
Red
supports label,
Blue
opposes label.
(hover tokens to see score)
Disclaimer: This is a research prototype and may produce incorrect predictions. Do not use as the sole basis for moderation decisions.