danijar/layered

Issues

• Color training costs based on label value

  #27 opened 9 years ago by danijar
  4

• Question about Softmax delta function

  #30 opened 8 years ago by howardyclo
  0

• Fix spelling of gradient descent

  #29 opened 9 years ago by danijar
  0

• syntax error in tutorial

  #26 opened 9 years ago by ebin123456
  0

• is there an example use simple data like iris to run this code

  #28 opened 9 years ago by nlp-study
  4

• Matplotlib is super slow on Mac

  #24 opened 9 years ago by danijar
  2

• Set up Pylint on Travis

  #10 opened 9 years ago by danijar
  0

• Add descriptions to activation and cost function tables in readme

  #21 opened 9 years ago by danijar
  0

• Test installation from PyPi

  #12 opened 9 years ago by danijar
  0

• Rename squared cost to SquaredError

  #23 opened 9 years ago by danijar
  0

• Rename linear activation function to identity

  #22 opened 9 years ago by danijar
  0

• Keep window open after training

  #25 opened 9 years ago by danijar
  0

• Plot window is not responding on Mac

  #19 opened 9 years ago by danijar
  0

• Plot crashes on exit

  #9 opened 9 years ago by danijar
  1

• Plot not showing up inside virtualenv on Linux

  #20 opened 9 years ago by danijar
  1

• Test activation and cost function deltas against their numeric gradients

  #18 opened 9 years ago by danijar
  1

• Update readme to include quickstart explaining features

  #17 opened 9 years ago by danijar
  0

• Set up Sphinx documentation

  #13 opened 9 years ago by danijar
  0

• Extract logic from main file into a class

  #16 opened 9 years ago by danijar
  0

• Store and load models and learned weights

  #11 opened 9 years ago by danijar
  1

• Dump evaluation results

  #15 opened 9 years ago by danijar
  0

• Plot evaluation results

  #7 opened 9 years ago by danijar
  1

• Rename rounds on training set to epochs

  #8 opened 9 years ago by danijar
  0

• Update readme since downloading datasets is no longer needed

  #14 opened 9 years ago by danijar
  0

• Average or just add up gradients in batch backpropagation?

  #1 opened 9 years ago by danijar
  1

• Split user and testing dependencies for faster Travis builds

  #5 opened 9 years ago by danijar
  0

• Set up Travis CI integration

  #3 opened 9 years ago by danijar
  0

• Automatically download and cache datasets

  #2 opened 9 years ago by danijar
  0

• Implement batch backpropagation as matrix times matrix multiplication

  #6 opened 9 years ago by danijar
  1

• Add tests to validate activation and cost functions against their numerical gradient

  #4 opened 9 years ago by danijar
  0