/Pytorch-practice

Credits - https://github.com/pytorch/examples

Primary LanguagePython

Pytorch-practice

Credits - https://github.com/pytorch/examples

Primary objective for using this repo for practice:

  • To get familiar with pytorch syntax.
  • To get to know the best practices while coding a DNN.
  • The repo uses argparse library to create CLI style code.

Secondary objectives for using this repo:

  • To learn to debug pytorch via VSCode IDE debugger tool and not via notebook.
  • To learn to setup Windows Subsystem for Linux in the new laptop and point VSCode to that disk image.
  • To setup python ecosystem on WSL.

Difficulties faced while setting up the workspace:

  • Python linter was not able to detect few methods.
  • Dimensions of the layers were passed wrongly and were throwing an error.

Solutions to the above:

  • Add the respective entries in the pylint settings (Linter that was used for this).
  • Dimension issue was detected when running the debugger tool and it was easy to detect since it was happening during training.
  • The stacktrace interpretation pointed it to the line where the matrix multiplication happened and the line was compared with the one in the repo and helped in fixing the issue.