search

lmjohns3 / theanets

Neural network toolkit for Python
http://theanets.rtfd.org
MIT License
328 stars 73 forks source link

Changed cascaded trainers to print out arguments, and accept special tra... #3

Closed kastnerkyle closed 11 years ago

kastnerkyle commented 11 years ago

...iner_specific_args. Also changed layerwise trainer to work with updated properties etc.

kastnerkyle commented 11 years ago

I am going to take a crack at updating this soon - I found a few things that seem out of place. Specifically, I would like to make the trainer_args parameter optional, the way I have it now it is required for a cascade trainer... I will incorporate your comments into the changes I make - does the optional trainer_args for cascade trainers sound OK by you @lmjohns3 ?

kastnerkyle commented 11 years ago

Found out I had merged my files wrong - I was doing a try... except then always doing the thing in the try! Not good... but should be fixed now.

@lmjohns3 I had to update the way cascaded trainers were called from main.py as well. I used this gist to test it.

lmjohns3 commented 11 years ago

I think I'm going to close this since it's a close match for the existing issue about having multiple trainers. Please re-open if you need to.