Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

KeyError: 'weight' when using resnet18 following mpc_cifar.py code

See original GitHub issue

I run into this error, when I try using CrypTen with resnet18 model for my code

KeyError                                  Traceback (most recent call last)
<ipython-input-11-8baf0b8ce013> in <module>
     16 
     17             input_size = get_input_size(val_loader, batch_size)
---> 18             private_model = construct_private_model(input_size, hosp_models[i])
     19             validate(loader=val_loader, model=private_model)
     20             print('\n')

<ipython-input-5-19f29188bd4f> in construct_private_model(input_size, model)
     15     else:
     16         model_upd = LeNet()
---> 17     private_model = crypten.nn.from_pytorch(model_upd, dummy_input).encrypt(src=0)
     18     return private_model
     19 

~/Downloads/mila project/udacity_DP_FL/CrypTen/crypten/nn/onnx_converter.py in from_pytorch(pytorch_model, dummy_input)
     35     # construct CrypTen model:
     36     f = _from_pytorch_to_bytes(pytorch_model, dummy_input)
---> 37     crypten_model = from_onnx(f)
     38     f.close()
     39 

~/Downloads/mila project/udacity_DP_FL/CrypTen/crypten/nn/onnx_converter.py in from_onnx(onnx_string_or_file)
    122     """Converts an onnx model to a CrypTen model"""
    123     converter = FromOnnx(onnx_string_or_file)
--> 124     crypten_model = converter.to_crypten()
    125     return crypten_model
    126 

~/Downloads/mila project/udacity_DP_FL/CrypTen/crypten/nn/onnx_converter.py in to_crypten(self)
    184             # add CrypTen module to graph
    185             crypten_module = crypten_class.from_onnx(
--> 186                 parameters=parameters, attributes=attributes
    187             )
    188             node_output_name = list(node.output)[0]

~/Downloads/mila project/udacity_DP_FL/CrypTen/crypten/nn/module.py in from_onnx(parameters, attributes)
   1584 
   1585         # initialize module:
-> 1586         in_channels = parameters["weight"].size(1)
   1587         out_channels = parameters["weight"].size(0)
   1588         module = Conv2d(

KeyError: 'weight'