-
Notifications
You must be signed in to change notification settings - Fork 3
/
mnist_runs.py
31 lines (26 loc) · 1.11 KB
/
mnist_runs.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
import subprocess
datadir = './data/'
datasets = ['mnist']#['large_linsep_4', 'clf_2', 'linsep', 'linsep_4', 'gauss_2']#['sklearn-digits', 'dna', 'satimage', 'svmguide1', 'pendigits', 'usps']
fracs = [ 0.4, 0.5]#, 0.6, 0.7, 0.8, 0.9]
num_epochs = 100
select_every = [20]#, 35, 50]
# select_every = [20, 60, 100]
warm_method = [0] # 0 = online, 1 = onestep warmstart
num_runs = 10
for dset in datasets:
for sel in select_every:
for f in fracs:
for isOneStepWarm in warm_method:
args = ['python3']
args.append('dss_deep.py')
#args.append('new_run_onestep_selection_minibatch.py') # selection every few!
#args.append('run_knnsubmod_selection_fullbatch.py') # selection using KNNsubmod indices
args.append(datadir + dset)
args.append(dset)
args.append(str(f))
args.append(str(num_epochs))
args.append(str(sel))
args.append(str(isOneStepWarm))
args.append(str(num_runs))
print(args)
subprocess.run(args)