Llama 3
Meta
partial detail 11 method fields published
Evaluation
Model robustness to benchmark design choices (MMLU perturbations)
Robustness of pretrained Llama 3 on MMLU to (1) few-shot label bias, (2) label variants, (3) answer order, (4) prompt format
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.