Hi Jame,

Now I can use your code and my own data to train model😁But it seems the
gpu can't be used on my computer,I have three cards,like this

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.48 Driver Version: 410.48
|
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr.
ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute
M. |
|===============================+======================+======================|
| 0 TITAN Xp Off | 00000000:02:00.0 On |
N/A |
| 41% 67C P2 86W / 250W | 2094MiB / 12194MiB | 0%
Default |
+-------------------------------+----------------------+----------------------+
| 1 GeForce GTX 108... Off | 00000000:03:00.0 Off |
N/A |
| 31% 45C P8 16W / 250W | 12MiB / 11178MiB | 0%
Default |
+-------------------------------+----------------------+----------------------+
| 2 GeForce GTX 108... Off | 00000000:81:00.0 Off |
N/A |
| 29% 33C P8 15W / 250W | 12MiB / 11178MiB | 0%
Default |
+-------------------------------+----------------------+----------------------+

And my environment should contain the cuda version needed. I think your
code is a cuda version,because I can see these in the main.py:

main.py:net = NET().cuda()
main.py:criterion = nn.CrossEntropyLoss().cuda()
main.py: images = [Variable(image).cuda() for image in images]
main.py: Net = NET().cuda()
main.py: inputs, labels = [Variable(input).cuda() for input in
inputs], labels
main.py: inputs, labels = [Variable(input).cuda() for input in
true_inputs], Variable(true_labels).cuda()

so I wonder if you can use your code under GPU environment smoothly

jamesaud [email protected] 于2019年3月31日周日 下午6:03写道：

These variables:

NUM_EVENTS = 1000
NUM_NOISE_EVENTS = 1000
MAX_RADIUS = 4
DURATION = args.duration
MIN_EVENTS = 300
MIN_MAGNITUDE = None
NOISE_TIMESPAN = Day(60)
ATTEMPTS = 30

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
jamesaud/seismic-analysis-toolbox#1 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AYC8hqt36KBZPRUq3hBy4cnIKwegbjA6ks5vcIf4gaJpZM4b-u66
.

Originally posted by @mingzhaochina in jamesaud/seismic-analysis-toolbox#1 (comment)

Hello！I don't understand why there is:

assert epoch > 0, "Epoch should start at 1"

it will stop data_validate.py

Thank you for your reply!I find two problems when running the code and so far I havn't find a solution on how to fix it.

1.main.py
When I try to use "named" loader in config_crossvalidation.json,the error is:
File "main.py", line 111, in
train_sampler = utils.make_weighted_sampler(dataset_train, NUM_CLASSES, weigh_classes=WEIGH_CLASSES) if WEIGH_CLASSES else None
File "/home/zhaoming/earthquake-detector-9000_branch/earthquake-detector-9000_newest/earthquake-detector-9000/utils.py", line 57, in make_weighted_sampler
weights = make_weights_for_classes(dataset, num_classes, weigh_classes)
File "/home/zhaoming/earthquake-detector-9000_branch/earthquake-detector-9000_newest/earthquake-detector-9000/utils.py", line 46, in make_weights_for_classes
weight_per_class[i] = N/count[i]
ZeroDivisionError: division by zero

2. When run data_validator.py
   the code will always stops at
   reload(main)
   and with the error message"failed to run neural net"