meta-llama

/

Prompt-Guard-86M

like 189

Follow

Meta Llama 8,220

Hello, the performance of Meta's Prompt Guard is impressive.
I admire this model and would like to train other BERT-based models on datasets like injection and jailbreak data.

As you may know, the most important thing when training is the dataset.
Could you share the training dataset used for the Prompt Guard model? Perhaps of the open data sources?