-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdeepnet_supervisor_analysis.py
42 lines (35 loc) · 1.63 KB
/
deepnet_supervisor_analysis.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
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
"""
Script to manage schedule / experiments. Controls repetitive script calls
and particular parameter settings.
@authors: Leon Sütfeld, Flemming Brieger
"""
import time
import os
import sys
import subprocess
"""
################################################################################
### RULES TO CONSTRUCT "AF_SET" ################################################
################################################################################
- must start with the number of AFs, then "_", then the set name
- must contain "swish" if swish AF is used at all
- must contain "scaled" if the scaled version of the respective AF(s) is to be used
- set name is the AFs name (in lowercase) or the name of the AF blend set, e.g., "blendSF7" or "blend9_swish"
- "elu" is called "jelu" ("just elu") to differentiate it from relu and selu
- "lu" (linear unit) is called "linu" for the same reason
- "swish" alone is called "jswish" ("just swish")
- [EXAMPLE] af_set='1_relu'
- [EXAMPLE] af_set='1_jswish'
- [EXAMPLE] af_set='1_jelu_scaled'
- [EXAMPLE] af_set='9_blend9_siwsh_scaled'
################################################################################
################################################################################
################################################################################
"""
os.system("nvidia-smi")
command = "python3 " + 'deepnet_main.py' + \
" -path_relative " + './' + \
" -experiment_name=" + 'debugging' + \
" -mode=" + 'analysis' + \
" -task=" + 'cifar100'
subprocess.run(command, shell=True)