I am using LSTM to make Language Model and I found out freezing embedding weight makes Volatile GPU-Util go up. (30% -> 80%). I don’t get it why it happens. Would you give me some advice about the issue(?)?
Volatile GPU-Util
weight_vec = torch.load('./data/pretrained_embedding.pt') model.emb.weight.data.copy_(weight_vec) model.emb.weight.requires_grad = False
Hi,
This might be because the autograd is not running, so the the CPU has less work to do and can send work to the GPU faster?
I see. I asked it just out of curiosity So, CPU is supposed to work on autograd even in cuda(gpu) mode? I didn’t know that. Many thanks.