Llama 2
Meta
rich detail 17 method fields published
Evaluation
How Far Can Model-Based Evaluation Go? (reward-model calibration)
Robustness/calibration of Llama 2's helpfulness and safety reward models against human preference annotations (3 annotators, 7-point Likert); check for Goodhart divergence using a more general reward trained on open-source reward modeling datasets
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.