Closed GeorgeBohw closed 6 years ago
This may be caused by tensor[i]
since this project didn't implement index selection.
@longcw thanks ,i check my code now
@longcw Right! tensor[i] cause the problem,but now i meet with another problem "KeyError: 'CudaTransferBackward'.Do you know the reason?
I am not sure what is CudaTransferBackward
represent for. But you can try to run the model on CPU for converting. (GPU operations are not supported).
@longcw Thanks very much,i try
@longcw Very Sorry to trouble you! I have a question,If the inputs are two variables,How should i feed the inputs in to function "pytorch2caffe(input_var, output_var, protofile, caffemodel)"
@longcw Right! tensor[i] cause the problem,but now i meet with another problem "KeyError: 'CudaTransferBackward'.Do you know the reason?
How to solve the problem about keyError:'IndexBackward' ?
When i convert my model from pytorch to caffe,it meet with the following error: keyError:'IndexBackward' What's wrong with it ,anybody can tell me?which layer's problem?