-
Notifications
You must be signed in to change notification settings - Fork 12
/
Copy pathtrain_itrPCRNet.sh
executable file
·35 lines (26 loc) · 1.04 KB
/
train_itrPCRNet.sh
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
#!/bin/bash
cd $PWD/utils/pc_distance/
make -f makefile_10.0 clean
make -f makefile_10.0 # Change name to makefile_8.0 if you have CUDA-8.0 and Ubuntu-14.04
cd $PWD/../..
PY="python3"
# Prefer python3
# For python2.7, follow steps mentioned below:
# Open train_iPCRNet.py, test_iPCRNet.py, statistical_analysis.py
# a. Remove "from numpy import matlib as npm"
# b. Replace "npm" with "np.matlib"
LOG_DIR="log_itrPCRNet" # Folder name to store log.4
MODE="train" # Either train or test.
RESULTS="best_model" # Keep saving network model after each epoch.
NOISE=False # To train network with noise in source data. (False/True)
# Train iterative PCRNet
$PY iterative_PCRNet.py -log $LOG_DIR -mode $MODE -results $RESULTS -noise $NOISE
WEIGHTS=$PWD/$LOG_DIR/$RESULTS".ckpt"
LOG_RESULTS="results_itrPCRNet"
# Test iterative PCRNet
$PY results_itrPCRNet.py -weights $WEIGHTS -log $LOG_RESULTS -noise $NOISE
# Visualize the results for various templates
for i in 0 700 1900 3000 4000 5000
do
$PY visualize_results.py -weights $WEIGHTS -idx $i
done