Classifiers are policy enforced by the process serving the model.
Input classifiers get applied before it reaches the model so somebody hacking an open-weight model would skip this. Streaming classifiers get polled during decoding; hackers delete this check in the sampling loop.
But both are always applied in closed weight models.
Set Llama Guard to 1.0 and nothing is ever unsafe.
Input classifiers get applied before it reaches the model so somebody hacking an open-weight model would skip this. Streaming classifiers get polled during decoding; hackers delete this check in the sampling loop.
But both are always applied in closed weight models.
Set Llama Guard to 1.0 and nothing is ever unsafe.