Llama 3.1
Meta
partial detail 9 method fields published
Evaluation
Child safety risk assessment
Child safety risk assessments by a team of experts across multiple attack vectors (including Llama 3 languages), with content specialists; findings inform fine-tuning mitigations and expand evaluation benchmark coverage
How it was run
Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.
Llama 3.2
Meta
partial detail 9 method fields published
See how this test was run
Llama 3.3
Meta
partial detail 9 method fields published
See how this test was run
Llama 4
Meta
partial detail 9 method fields published