Why is Edward, imho, the greatest idea since sliced bread, so poorly supported?


#1

I see a lot of people that have excellent questions in this forum, that receive no replies or guidance. It’s Very disheartening. By comparison, the PyMC3 people answered one of my questions with full code in 5 minutes flat (thanks to the brilliant junpenglao!) and they don’t have Google’s or TensorFlow’s muscle behind them. So what is the problem?


#3

Thanks for your feedback and kind words about Edward. Right now I’m the sole maintainer, and it was written while I’ve been finishing my PhD and other research.

I’ve now joined Google full-time and I’m working with a team to design the new and improved Edward 2.0. The aim is for it to be officially supported and adopted by Google. Stay tuned!


#4

I’m glad to hear it. I honestly think that the Edward bayesian network/ hierarchical models portion of TensorFlow will soon become hugely beneficial to society and lucrative to Google. I hope Google stops sleeping at the wheel and devotes many more hands to it ASAP!


#5

Is Google hiring for this team in particular? I would definitely be interested.


#6

I think you apply as a software engineer or research scientist position. Then you would select our team for matching. Maybe @derifatives can say more?


#7

Thats great news about Edward and congrats on the new position. Its really putting advanced modeling within reach of medical scientists, i.e. non-ML researchers and developers, like myself. Thanks for all your work on this project


#8

Hi Dustin, do you have any rough time schedule for Edward 2.0 relase?