Synapse-Guard-v1.2

Synapse-Guard-v1.2 is a lightweight Portuguese text classification model for content moderation.

This model does not generate harmful or adult content. It is a classifier designed to identify and categorize text.

Labels

• neutro – Normal and safe conversations.
• spam – Advertisements, scams, phishing and promotional messages.
• toxico – Offensive, abusive or insulting language.
• adulto – Adult or sexually explicit content.

Intended Uses

• AI assistants
• Chatbots
• Community moderation
• Forums
• Social networks
• Content filtering
• Spam detection
• Safety pipelines

Limitations

• Predictions may not always be correct.
• Performance depends on the domain.
• Should be used alongside other moderation systems.

Example

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="Comunidade-Synapse-BR/Synapse-Guard-v1.2"
)

print(classifier("Você é um idiota."))

License

Apache-2.0