Llama 2
Meta
partial detail 12 method fields published
Evaluation
Safety RLHF ablation: long-tail safety robustness vs helpfulness
Impact of Safety RLHF: score two intermediate Llama 2-Chat checkpoints (one without adversarial prompts in RLHF, one with) with safety and helpfulness reward models; examine safety score distribution shifts and helpfulness preservation
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.