Klaus Greff

Yeah, that one is clearly a problem. It started as a feature because in the beginning, I thought it was convenient to be able to use `a.d=3` and `a =...

Hi. I'm sorry for not being able to fix this ATM. And while I'm happy to hear that you consider tackling this yourself, I think I should warn/discourage you: This...

This is an interesting usecase that I haven't considered yet. Thanks for bringing it up. Unfortunately I don't have time to dig into it further now, and I will be...

Wow, that is a strange issue! It seems that list-comprehensions (and all other comprehensions actually) somehow fail to access the custom locals dictionary. Thanks for reporting this! I'll have to...

This is a pretty strange issue indeed. Thanks @absolutelyNoWarranty for digging in a bit further. Based on your comment I have run a few tests, and it seems like comprehensions...

No clean solution unfortunately, but you do hav a few choices: 1. You can move temp outside the config function 2. You can delete `temp` later: ``` python cv =...

You can access the run object that contains all the information from any captured function using the `_run` parameter: ```python @ex.capture def get_info(_run): print(_run._id) print(_run.experiment_info["name"]) ```

I liked the idea of sacred geometry and have experimented with it for a while. Here is the result: 

Hi, thank you for the nice words! And your usecase makes a lot of sense and should definitely be supported in Sacred. Unfortunately there is currently no way of telling...

Yeah I agree. I wasn't aware of this behavior of tensorflow. It should at the very least be documented and probably an info-logging statement should also mention it. Alternatively we...