TypeError: load_state_dict() got an unexpected keyword argument 'strict'

vision
1

When I try to finetune a model, I set the ‘strict’ argument to load the state_dict, however, an error appears:

“TypeError: load_state_dict() got an unexpected keyword argument ‘strict’”

My pytorch is 1.2.0, can anyone help me with it?

2

Anyone can help, thanks?

3

The strict argument was already added to 1.2.0 so could you post a code snippet to reproduce this error?

4

Thank you.

Here is the code:
train_model.load_state_dict(
#torch.load(os.path.join(final_output_path, config.model_prefix + str(curr_iter) + ‘.pth’)), resume=True) # this works fine
torch.load(os.path.join(final_output_path, config.model_prefix + str(curr_iter) + ‘.pth’)),strict=False) #this doesnot work and give the error I list above.

5

Your code is unfortunately not executable and I cannot reproduce the error in PyTorch 1.2.0:

>>> import torch
>>> torch.__version__
'1.2.0'
>>> import torch.nn as nn
>>> model = nn.Linear(1, 1)
>>> model.load_state_dict(nn.Linear(1, 1).state_dict(), strict=False)
<All keys matched successfully>
>>> model.load_state_dict(nn.Linear(1, 1).state_dict(), strict=True)
<All keys matched successfully>