'freeze_modules_based_on_opts()' is freezing module parameters twice

[hub-bla]

Example

opts = argparse.Namespace(**{"model.freeze_modules": "conv1"})


model = nn.Sequential(
     OrderedDict([
          ('conv1', nn.Conv2d(20,64,5)),
          ('conv2', nn.Conv2d(20,64,5))
        ])
)


print(freeze_modules_based_on_opts(opts, model))

[KRITSABIG]

Example

opts = argparse.Namespace(**{"model.freeze_modules": "conv1"})


model = nn.Sequential(
     OrderedDict([
          ('conv1', nn.Conv2d(20,64,5)),
          ('conv2', nn.Conv2d(20,64,5))
        ])
)


print(freeze_modules_based_on_opts(opts, model))

[mustafahadbe]

Example

opts = argparse.Namespace(**{"model.freeze_modules": "conv1"})


model = nn.Sequential(
     OrderedDict([
          ('conv1', nn.Conv2d(20,64,5)),
          ('conv2', nn.Conv2d(20,64,5))
        ])
)


print(freeze_modules_based_on_opts(opts, model))

``mu

[Mandeep0001]

GOT IT

[mohammad7t]

Freezing modules (i.e. disabling .train()) is different with freezing parameters (i.e. setting requires_grad=False). As this issue is closely related to #10 , I'm closing this one.