From 21d875969a1455e4d5b26dcf32c843e6262d1f9c Mon Sep 17 00:00:00 2001 From: fra31 Date: Thu, 24 Feb 2022 17:51:02 +0000 Subject: [PATCH] update readme --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 58bd153..df2946c 100644 --- a/README.md +++ b/README.md @@ -48,8 +48,8 @@ The code is tested under Python 3.8.5 and PyTorch 1.8.0. It automatically downlo The following are examples of how to run the attacks in the different threat models. -### L0-bounded -In this case `k` represents the number of pixels to modify. For untargeted attacks +### L0-bounded (pixel and feature space) +In this case `k` represents the number of *pixels* to modify. For untargeted attacks ``` CUDA_VISIBLE_DEVICES=0 python eval.py --norm=L0 \ --model=[pt_vgg | pt_resnet] --n_queries=10000 --alpha_init=0.3 \