Closed zhangfengyo closed 3 years ago
It seems that the channel dimension is wrong. What's the size of your training images?
160,will it be wrong?
------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2021年7月26日(星期一) 中午11:37 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [ecnuycxie/DG-Font] error occur in validation (#11)
It seems that the channel dimension is wrong. What's the size of your training images?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
START EPOCH[1]
SAVE CHECKPOINT[1] DONE
train
0% 0/1000 [00:00<?, ?it/s]/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:718: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /pytorch/c10/core/TensorImpl.h:1156.)
return torch.max_pool2d(input, kernel_size, stride, padding, dilation, ceil_mode)
10% 99/1000 [00:37<05:23, 2.78it/s]Epoch: [1/25] [100/1000] MODE[GAN] Avg Loss: D[5.74] G[0.91]
20% 199/1000 [01:13<04:48, 2.77it/s]Epoch: [1/25] [200/1000] MODE[GAN] Avg Loss: D[3.84] G[1.00]
30% 299/1000 [01:49<04:11, 2.78it/s]Epoch: [1/25] [300/1000] MODE[GAN] Avg Loss: D[3.22] G[0.87]
40% 399/1000 [02:25<03:35, 2.79it/s]Epoch: [1/25] [400/1000] MODE[GAN] Avg Loss: D[3.01] G[0.73]
50% 499/1000 [03:01<02:59, 2.79it/s]Epoch: [1/25] [500/1000] MODE[GAN] Avg Loss: D[2.82] G[0.67]
60% 599/1000 [03:37<02:23, 2.79it/s]Epoch: [1/25] [600/1000] MODE[GAN] Avg Loss: D[2.72] G[0.62]
70% 699/1000 [04:13<01:47, 2.79it/s]Epoch: [1/25] [700/1000] MODE[GAN] Avg Loss: D[2.68] G[0.57]
80% 799/1000 [04:49<01:12, 2.78it/s]Epoch: [1/25] [800/1000] MODE[GAN] Avg Loss: D[2.63] G[0.60]
90% 899/1000 [05:25<00:36, 2.79it/s]Epoch: [1/25] [900/1000] MODE[GAN] Avg Loss: D[2.62] G[0.55]
100% 999/1000 [06:01<00:00, 2.77it/s]Epoch: [1/25] [1000/1000] MODE[GAN] Avg Loss: D[2.56] G[0.52]
100% 1000/1000 [06:01<00:00, 2.77it/s]
validation
START EPOCH[2]
SAVE CHECKPOINT[2] DONE
train
[W pthreadpool-cpp.cc:90] Warning: Leaking Caffe2 thread-pool after fork. (function pthreadpool)
[W pthreadpool-cpp.cc:90] Warning: Leaking Caffe2 thread-pool after fork. (function pthreadpool)
[W pthreadpool-cpp.cc:90] Warning: Leaking Caffe2 thread-pool after fork. (function pthreadpool)
[W pthreadpool-cpp.cc:90] Warning: Leaking Caffe2 thread-pool after fork. (function pthreadpool)
10% 99/1000 [00:35<05:23, 2.78it/s]Epoch: [2/25] [100/1000] MODE[GAN_EMA] Avg Loss: D[2.27] G[-0.24]
20% 199/1000 [01:11<04:49, 2.77it/s]Epoch: [2/25] [200/1000] MODE[GAN_EMA] Avg Loss: D[2.16] G[0.04]
30% 299/1000 [01:48<04:13, 2.76it/s]Epoch: [2/25] [300/1000] MODE[GAN_EMA] Avg Loss: D[2.23] G[-0.13]
40% 399/1000 [02:23<03:36, 2.77it/s]Epoch: [2/25] [400/1000] MODE[GAN_EMA] Avg Loss: D[2.17] G[-0.01]
50% 499/1000 [03:00<03:00, 2.78it/s]Epoch: [2/25] [500/1000] MODE[GAN_EMA] Avg Loss: D[2.13] G[0.03]
60% 599/1000 [03:35<02:24, 2.78it/s]Epoch: [2/25] [600/1000] MODE[GAN_EMA] Avg Loss: D[2.15] G[0.11]
70% 699/1000 [04:11<01:47, 2.79it/s]Epoch: [2/25] [700/1000] MODE[GAN_EMA] Avg Loss: D[2.12] G[0.13]
80% 799/1000 [04:47<01:12, 2.79it/s]Epoch: [2/25] [800/1000] MODE[GAN_EMA] Avg Loss: D[2.10] G[0.17]
90% 899/1000 [05:23<00:36, 2.78it/s]Epoch: [2/25] [900/1000] MODE[GAN_EMA] Avg Loss: D[2.10] G[0.20]
100% 999/1000 [05:59<00:00, 2.79it/s]Epoch: [2/25] [1000/1000] MODE[GAN_EMA] Avg Loss: D[2.09] G[0.21]
100% 1000/1000 [06:00<00:00, 2.78it/s]
validation
Traceback (most recent call last):
File "main.py", line 400, in
it is hard to know the reason behind it
The original code is only for 80*80. If you want to utilize different resolutions of images, you'd better increase convolution layers in the content encoder and the mixer. Besides, Checking the initialization of each network and the size of each feature map will help you resolve your questions.
You should change args.att_to_use match with num_cls change main.py:192 to args.att_to_use = [i for i in range(args.num_cls)]
for me it worked
You should change args.att_to_use match with num_cls change main.py:192 to args.att_to_use = [i for i in range(args.num_cls)]
for me it worked
Yes, i have find it was the reason behind
for me it worked I modified args.att_to_use = [i for i in range(args.num_cls)], but after the second epoch the network fell into the effect of stopping motionless, and the graphics card resources are being occupied.
You should change args.att_to_use match with num_cls change main.py:192 to args.att_to_use = [i for i in range(args.num_cls)]
for me it worked
I modified args.att_to_use = [i for i in range(args.num_cls)], but after the second epoch the network fell into the effect of stopping motionless, and the graphics card resources are being occupied.
At this point, The code has entered the validation stage. During the validation process, it is required to save num_fonts * num_fonts * 2
images. So if there is no information being printed for a long time, it means that images are being continuously saved.
This can be seen here: https://github.com/ecnuycxie/DG-Font/blob/b9368fd2527a42714742786a47d3bff08b8a96c8/validation/validation.py#L65-L69
Traceback (most recent call last): File "main.py", line 400, in
main()
File "main.py", line 170, in main
main_worker(args.gpu, ngpus_per_node, args)
File "main.py", line 250, in main_worker
validationFunc(val_loader, networks, epoch, args, {'logger': logger})
File "/content/drive/My Drive/DG-Font-main/validation/validation.py", line 82, in validateUN
x_res_ematmp, = G_EMA.decode(c_src, s_ref, skip1, skip2)
File "/content/drive/My Drive/DG-Font-main/models/generator.py", line 51, in decode
out = self.decoder(cnt, skip1, skip2)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(*input, kwargs)
File "/content/drive/My Drive/DG-Font-main/models/generator.py", line 87, in forward
output = self.modeli
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(*input, *kwargs)
File "/content/drive/My Drive/DG-Font-main/models/blocks.py", line 15, in forward
return self.model(x)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(input, kwargs)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/container.py", line 139, in forward
input = module(input)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(*input, kwargs)
File "/content/drive/My Drive/DG-Font-main/models/blocks.py", line 32, in forward
residual = self.model(x)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(*input, *kwargs)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/container.py", line 139, in forward
input = module(input)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(input, kwargs)
File "/content/drive/My Drive/DG-Font-main/models/blocks.py", line 158, in forward
x = self.norm(x)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1051, in _call_impl
return forward_call(*input, **kwargs)
File "/content/drive/My Drive/DG-Font-main/models/blocks.py", line 208, in forward
True, self.momentum, self.eps)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py", line 2282, in batch_norm
input, weight, bias, running_mean, running_var, training, momentum, eps, torch.backends.cudnn.enabled
RuntimeError: weight should contain 512 elements not 2560