juefeix / lbcnn.torch Goto Github PK

Creating model from file: models/resnet-binary-felix.lua
.....
It took a long time to sentence that is " model = require('models/' .. opt.netType)(opt) ", why?

@juefeix Hi, nice work and thank you for sharing. In LBCNN, the number of trainable parameters do decrease a lot, but the calculation procedure seems to grow. So I guess it will take longer time for inference than regular CNNs. Is that true?

Since your Top1 and Top5 is acc rate not err rate, the condition of testTop1 < bestTop1 should be testTop1 > bestTop1 isn't is right?

1. There is no guide for running lbcnn version resnet with the dataset Imagenet even there r requirements to download the Imagenet data set! What happened? Actually, when I run yr code with Imagenet dataset, there is an error about out of memory even the batch size is small enough. Is it sure that there is no problems for running lbcnn version resnet with the dataset Imagenet?
   There is no problem for running fb.resnet with Imagenet.

2. Would you share lbcnn version AlexNet source code for training the dataset of Imagenet?

where can I change the number of epochs?

hi,
I'm trying to run the code on one of the datasets but it gives me this:
=> Creating model from file: models/resnet-binary-felix.lua
| combining dataset into a single file
/home/hadeel/torch/install/bin/luajit: cannot open <~/LBCNN/cifar-10-batches-t7/cifar-10-batches-t7/data_batch_1.t7> in mode r at /home/hadeel/torch/pkg/torch/lib/TH/THDiskFile.c:673
stack traceback:
[C]: at 0x7f079279b200
[C]: in function 'DiskFile'
/home/hadeel/torch/install/share/lua/5.1/torch/File.lua:405: in function 'load'
/home/hadeel/project/lbcnn.torch/datasets/cifar10-gen.lua:18: in function 'convertToTensor'
/home/hadeel/project/lbcnn.torch/datasets/cifar10-gen.lua:41: in function 'exec'
./datasets/init.lua:28: in function 'create'
./dataloader.lua:15: in function 'create'
main.lua:31: in main chunk
[C]: in function 'dofile'
...deel/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
[C]: at 0x561e5eadb570

please can you help me with this

This is a novel paper! We are interesting in it.