inconsistent tensor size

[simmoncn]

Hi, do you know the following errors. thanks

/root/torch/install/bin/luajit: /root/torch/install/share/lua/5.1/threads/threads.lua:183: [thread 1 callback] .../Photo-Realistic-Super-Resoluton-master/data/dataset.lua:339: inconsistent tensor size at /root/torch/pkg/torch/lib/TH/generic/THTensorCopy.c:7
stack traceback:
[C]: in function 'copy'
.../Photo-Realistic-Super-Resoluton-master/data/dataset.lua:339: in function 'tableToOutput'
.../Photo-Realistic-Super-Resoluton-master/data/dataset.lua:359: in function <.../Photo-Realistic-Super-Resoluton-master/data/dataset.lua:347>
[C]: in function 'xpcall'
/root/torch/install/share/lua/5.1/threads/threads.lua:234: in function 'callback'
/root/torch/install/share/lua/5.1/threads/queue.lua:65: in function </root/torch/install/share/lua/5.1/threads/queue.lua:41>
[C]: in function 'pcall'
/root/torch/install/share/lua/5.1/threads/queue.lua:40: in function 'dojob'
[string " local Queue = require 'threads.queue'..."]:15: in main chunk

[yulunzhang]

Hi, I also have the same problem. In /data/dataset.lua line 339, I checked the size of data[i] and dataTable[i] and found that they really had different sizes. The size of data[i] is 1x96x96, while dataTable[i] has four different sizes (e.g., 1x128x96, 1x96x128, 1x143x96, 1x96x144) corresponding to 'nThreads = 4'. As a result, the error 'inconsistent tensor size' occurred.
The error information are as follows:
lunge@lunge:/media/lunge/Users/projects/SRGAN/Photo-Realistic-Super-Resoluton$ th run_sr.lua
{
ntrain : inf
beta1 : 0.9
name : "super_resolution"
dataset : "folder"
niter : 250
lr : 0.001
model_folder : "model"
gpu : 1
nThreads : 1
t_folder : "/media/lunge/Data/IMAGENET2015/CLS_LOC_dataset/train"
batchSize : 32
loadSize : 96
}
Starting donkey with id: 1 seed: 1
table: 0x41e4bfe0
Loading train metadata from cache
Dataset: folder Size: 1281167
cunn used
/home/lunge/torch/install/bin/luajit: /home/lunge/torch/install/share/lua/5.1/threads/threads.lua:183: [thread 1 callback] ...s/SRGAN/Photo-Realistic-Super-Resoluton/data/dataset.lua:339: inconsistent tensor size at /home/lunge/torch/pkg/torch/lib/TH/generic/THTensorCopy.c:7
stack traceback:
[C]: in function 'copy'
...s/SRGAN/Photo-Realistic-Super-Resoluton/data/dataset.lua:339: in function 'tableToOutput'
...s/SRGAN/Photo-Realistic-Super-Resoluton/data/dataset.lua:359: in function <...s/SRGAN/Photo-Realistic-Super-Resoluton/data/dataset.lua:347>
[C]: in function 'xpcall'
/home/lunge/torch/install/share/lua/5.1/threads/threads.lua:234: in function 'callback'
/home/lunge/torch/install/share/lua/5.1/threads/queue.lua:65: in function </home/lunge/torch/install/share/lua/5.1/threads/queue.lua:41>
[C]: in function 'pcall'
/home/lunge/torch/install/share/lua/5.1/threads/queue.lua:40: in function 'dojob'
[string " local Queue = require 'threads.queue'..."]:15: in main chunk
stack traceback:
[C]: in function 'error'
/home/lunge/torch/install/share/lua/5.1/threads/threads.lua:183: in function 'dojob'
...ects/SRGAN/Photo-Realistic-Super-Resoluton/data/data.lua:76: in function 'getBatch'
run_sr.lua:85: in function 'opfunc'
/home/lunge/torch/install/share/lua/5.1/optim/adam.lua:37: in function 'adam'
run_sr.lua:133: in main chunk
[C]: in function 'dofile'
...unge/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:145: in main chunk
[C]: at 0x00405d50
lunge@lunge:/media/lunge/Users/projects/SRGAN/Photo-Realistic-Super-Resoluton$
The size of dataTable[i] should be 1x96x96, I think. Would you please help us solve this problem? Thank you! @leehomyc

[leehomyc]

Hi I don't have this problem while training.

Yes data[i] has size 96x96. However dataTable[i] uses sampleHookTrain which relates to the function local trainHook=function(self,path) in donkey_folder_supres.lua. The function returns two outputs (input_y, and input_y2) where the first one is of size 96x96 and the second one has size 24x24. I do not know where other sizes come from.

If you know what caused the problem, please let me know and I can update the repository. Thanks!

[zhangqianhui]

@leehomyc

I have also got this problem.

[zhangqianhui]

Can you tell us your torch and other libs version?

[leehomyc]

Hello!

Sorry I fixed the issue which is caused by image size. When the input image's height and width are different there will be such error.

[zhangqianhui]

@leehomyc thanks.