How can I load a trained model in python with torch?

mravanelli / pytorch-kaldi

pytorch-kaldi is a project for developing state-of-the-art DNN/RNN hybrid speech recognition systems. The DNN part is managed by pytorch, while feature extraction, label computation, and decoding are performed with the kaldi toolkit.

2.36k stars 446 forks source link

How can I load a trained model in python with torch? #236

Open sun-peach opened 4 years ago

sun-peach commented 4 years ago

Hi, I have followed the librispeech cfg to train a model. Now I would like to load it in python code with torch. I will use it as a pretrained module within a complex system. Do you have any tutorial to do that? Thank you.

TParcollet commented 4 years ago

The .pkl that is saved is pretty standard ! You can load it with torch.load()

sun-peach commented 4 years ago

Thank you. Will it save the intermediate model somewhere? I make it run 15 epochs, while I don't see any .pkl files in the corresponding model directory (I set the out_folder in the config file) during training (currently, it has run 9 epochs). Should I just wait until it finish all the epochs?