Closed vmr013 closed 3 years ago
Your gpu memory not enough for training this model. Firstly, You can try to decrease your batch size.
I tried batch_size with values 8,4,2,1. For each there is this memory error.
This is model configuration from train.yaml
model:
learning_rate_E_G: 4e-4
learning_rate_D: 4e-4
beta1: 0
beta2: 0.999
batch_size: 16
num_workers: 16
grad_clip: 0.0
Do you see the issue? I don't see it
I got this error repeatedly before. Reason of this error is always that my gpu memory not enough. When start training, If you use Linux, you can check your gpu memory with run "watch nvidia-smi"
command in terminal.
Thanks
Hi, can someone tell me what can I do to fix this issues? This is what I get when running it on local machine or google colab.