search

albanie / pytorch-mcn

Convert models from MatConvNet to PyTorch
MIT License
93 stars 13 forks source link

I suggest change python/ipython to python3/ipython3 for using Python3.5 obviously in importer.sh #2

Closed TaihuLight closed 6 years ago

TaihuLight commented 6 years ago

Steps for Convert models from MatConvNet to PyTorch with pytorch-mcn (1)run pytorch-mcn/compare/ensure_dags.m in matlab to convert models to dagnn... (2)edit and run shell script importer.sh to Convert models from MatConvNet to PyTorch with pytorch-mcn (3)the results Saving imported model definition to ./models/resnext_50_32x4d_pt_mcn.py Saving imported weights to ./models/resnext_50_32x4d_pt_mcn.pth

Bugs? (1)Due to the default Python is 2.7 on the most Linux OS, and Python3 means Python3.5 on the the most Linux OS. Thus, I suggest change the python/ipython to python3/ipython3 for using Python3.5 obviously in importer.sh. (2) The waning "skipping conv bias term" skipping conv bias term skipping conv bias term skipping conv bias term skipping conv bias term skipping conv bias term (3) Errors

/home/chengzi/Downloads/pytorch-mcn/python/ptmcn_utils.py(66)conv2d_mod() 65 if not valid_conv_groups: import ipdb ; ipdb.set_trace() ---> 66 assert valid_conv_groups, msg 67 num_groups = int(in_ch / fsize[2])

albanie commented 6 years ago

(1) Changing the python command to python3 is a good suggestion. (2) The skipping conv bias term is not a bug, it's simply a notification to the user that there are no bias parameters associated with the corresponding convolutional weights. However, this could be restricted to a verbose mode. (3) This error is a safety check which fails if the tool is not able to determine the convolutional weight shape.

albanie commented 6 years ago

Am closing, but feel free to re-open if needed.