save and load model

[thuliyt12]

Hi, i am trying to use your model to do some speaker verification experiment. After training, the model parameters are stored as below:

torch.save(Model.state_dict(),PATH)

loaded as below:

Model.load_state_dict(torch.load(PATH))

her comes the error:

AttributeError: 'Variable' object has no attribute 'copy_'

this is because state saves tdnn.context, which is a 'variable', into state_dict. When load the state_dict, variable has no attribute 'copy_'

How do you solve this problem?

[SiddGururani]

Hi! Thanks for pointing that out.

It's possible that changing the line #21:
self.register_buffer('context',Variable(torch.LongTensor(context)))
to
self.register_buffer('context',torch.LongTensor(context))

might work. I haven't tested it out but it seems like self.context doesn't need to be a Variable.

[SiddGururani]

I believe the issue should be fixed now so I'm gonna go ahead and close it. Please feel free to reopen if you find another..... issue 😆