Max Pumperla

@PhilipMay that was precisely the question I had for you before merging that PR. It's a little unfortunate, but not that big of a deal. If you can please provide...

@PhilipMay there's no way to silence the progress bar by default? I don't have time to look into this right now, but I'd rather fix this feature than revert it....

@easyas314159 @PhilipMay I'd love to get a discussion started around this and plan/design this with you. Callbacks and early stopping have been on my bucket list for ages, your help...

@genixpro @ninnghazad thanks for checking in and your info. I'd be happy to see ATPE integrated into hyperopt core, no reason not to extend the search algo zoo. A contribution...

@genixpro wow, thanks for the PR. Well, the next step is to review. After we merge, I can do a release.

@JonnoFTW would your instructions mentioned by @ivan-marroquin be helpful for others here as well?

hey @aldanor, thanks for all your suggestions. 0. About the license, I think only @jaberg can decide this. 1. I agree with you that fading out python2 makes sense, but...

@PhilipMay we now have "proper" documentation here http://hyperopt.github.io/hyperopt/. you can contribute to it by adding `md` files to `templates` in the `docs` folder (see https://github.com/hyperopt/hyperopt/tree/master/docs). Afterwards simply add the page...

I think the loss-based stopping is more useful in practice, but why not give users options? We could think about a simple `stopping_criterion` argument to fmin that is None by...

@PhilipMay just a rough idea, may not even work as suggested. but if you come up with a good idea, I'm happy to discuss, as you know.