Loading model take 30+ minutes

NVIDIA / tacotron2

Tacotron 2 - PyTorch implementation with faster-than-realtime inference

BSD 3-Clause "New" or "Revised" License

5.12k stars 1.39k forks source link

Loading model take 30+ minutes #572

Open Wemmons831 opened 2 years ago

Wemmons831 commented 2 years ago

Hello I am unaware if this is normal behavior but whenever I train Tacotron the line model = load_model(hparams) in train.py takes upwards of 30 minutes to execute. has anyone else encountered this behavior.

Wemmons831 commented 2 years ago

loading dummy data to cuda such as torch.randn(2).to(device) seems to lower down the loading times slightly