search

GorkaAbad / NeuromorphicBackdoors

[CCS'22] Backdoor Attacks on Spiking NNs and Neuromorphic Datasets
8 stars 0 forks source link

program got KILLED when poisoning train data #1

Open Firewall12138 opened 1 year ago

Firewall12138 commented 1 year ago

i was testing the code on the remote server. list = [] for data in tqdm(new_data): list.append(np.array(data[0])) new_data = np.array(list) the codes running to new_data = np.array(list) while posioning the train data, the program got KILLED with no other error megs. But the problem did not happen when posioning test data. do you have any idea about handle this problem？Really appreciate it.

GorkaAbad commented 1 year ago

Hi! Probably it's because you are running out of memory. Since the test set is smaller it should work on that easier. You could reduce the training set if you cannot get more memory. Anyways, I will review the code and check if I can make it more efficient. I'm sorry. Best