Closed CloudTronUSA closed 1 year ago
If your model is training super slow, for example, 50seconds/batch, you might be running on your CPU add CUDA_VISIBLE_DEVICES=0 in front of your train command to force the trainer to use GPU.
If your model is training super slow, for example, 50seconds/batch, you might be running on your CPU add CUDA_VISIBLE_DEVICES=0 in front of your train command to force the trainer to use GPU.