locuslab/TCN

error

fengyuxie opened this issue · 3 comments

fengyuxie commented

Hi,
I am a student. When I run your code, I have a question to think about. Your MNIST pixel task is very good. The data set is very formal. When I have a self-data set, what changes should I make on your basis and how to do them?
Best Regards

fengyuxie commented

there is a error :CUDA out of memory. Tried to allocate 230.00 MiB (GPU 0; 6.00 GiB total capacity; 4.17 GiB already allocated; 158.63 MiB free; 16.12 MiB cached). but the batch_size=16 or 8 the error still happen
I changed the input_channel=1 seq_length = int(50176/ input_channels) 50176=224*224 n_classes = 22

jerrybai1995 commented

Hi @fengyuxie, it's hard to tell what the problem is. From the error you reported, your GPU only has 6GB of memory available. What machine/setting are you running on (how many GPUs, etc.)?

You could also use the NVIDIA APEX's half precision to resolve the memory issue, if needed.

fengyuxie commented

thanks for your suggestion,I will try to check out that again