alkasaliss / democlassi Goto Github PK

Thanks for the awesome project! However, when I run the experiments_and_evaluation_utk_latest.ipynb file and trained the models myself, I had the following error:

RuntimeError Traceback (most recent call last)
in ()
9 cpu_or_gpu = "cuda" if torch.cuda.is_available() else "cpu"
10 sep_conv_model = SepConvModelMT()
---> 11 sep_conv_model.load_state_dict(torch.load(path_sep_conv, map_location=cpu_or_gpu))
12
13 resnet_model = PretrainedMT(model_name='resnet')

/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in load_state_dict(self, state_dict, strict)
828 if len(error_msgs) > 0:
829 raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
--> 830 self.class.name, "\n\t".join(error_msgs)))
831 return _IncompatibleKeys(missing_keys, unexpected_keys)
832

RuntimeError: Error(s) in loading state_dict for SepConvModelMT:
Missing key(s) in state_dict: "conv_base.conv1.depthwise.weight", "conv_base.conv1.depthwise.bias", "conv_base.conv1.pointwise.weight", "conv_base.conv1.pointwise.bias", "conv_base.batchnorm1.weight", "conv_base.batchnorm1.bias", "conv_base.batchnorm1.running_mean", "conv_base.batchnorm1.running_var", "conv_base.conv2.depthwise.weight", "conv_base.conv2.depthwise.bias", "conv_base.conv2.pointwise.weight", "conv_base.conv2.pointwise.bias", "conv_base.batchnorm2.weight", "conv_base.batchnorm2.bias", "conv_base.batchnorm2.running_mean", "conv_base.batchnorm2.running_var", "conv_base.conv3.depthwise.weight", "conv_base.conv3.depthwise.bias", "conv_base.conv3.pointwise.weight", "conv_base.conv3.pointwise.bias", "conv_base.batchnorm3.weight", "conv_base.batchnorm3.bias", "conv_base.batchnorm3.running_mean", "conv_base.batchnorm3.running_var", "conv_base.conv4.depthwise.weight", "conv_base.conv4.depthwise.bias", "conv_base.conv4.pointwise.weight", "conv_base.conv4.pointwise.bias", "conv_base.batchnorm4.weight", "conv_base.batchnorm4.bias", "conv_base.batchnorm4.running_mean", "conv_base.batchnorm4.running_var", "conv_base.conv5.depthwise.weight", "conv_base.conv5.depthwise.bias", "conv_base.conv5.pointwise.weight", "conv_base.conv5.pointwise.bias", "conv_base.batchnorm5.weight", "conv_base.batchnorm5.bias", "conv_base.batchnorm5.running_mean", "conv_base.batchnorm5.running_var", "conv_base.conv6.depthwise.weight", "conv_base.conv6.depthwise.bias", "conv_base.conv6.pointwise.weig...
Unexpected key(s) in state_dict: "optimizer", "model".

Hey, i am having while evaluating the model
The error occurs in this line
dataloader_fer_tl=get_fer_dataloader(256,path_fer,'PrivateTest',data_transforms=data_transform_fer)

when i run the second part: experiments_and_evaluation_utk_latest.ipynb
it has the error like:
Number of trainable parameters : 706,278
Number of non-trainable parameters : 0

RuntimeError Traceback (most recent call last)
in ()
4 resume_model=None, resume_optimizer=None, backup_step=5, backup_path=backup_path,
5 n_epochs_freeze=0, lr_after_freeze=None,
----> 6 loss_weights=[1/10, 1/0.16, 1/0.44])

7 frames
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in convert(t)
382
383 def convert(t):
--> 384 return t.to(device, dtype if t.is_floating_point() else None, non_blocking)
385
386 return self._apply(convert)

RuntimeError: CUDA out of memory. Tried to allocate 2.00 MiB (GPU 0; 14.73 GiB total capacity; 12.95 GiB already allocated; 4.94 MiB free; 91.58 MiB cached)

does it has any thing to do with the batch_size and i can't find the way to change it . does any body can tell me how to fix it.