Andreas Jansson
Andreas Jansson
Hi @fishbotics, good question! The documentation could probably be more clear around this. Keepsake doesn't store diffs like Git, for each experiment and checkpoint it stores the whole files and...
@kvthr GPU metrics would be fantastic, I've found memory and per-GPU utilization to be very helpful when debugging bottlenecks. FLOPS / MACs would be really good too. Thanks for that...
test failures look legit
One potential design for this could be something along the lines of: ``` experiment = keepsake.init(path=".", params=..., logs_path="./tensorboard-logs/") ``` where `logs_path` is stored in the experiment and synced _to the...
@aaronleesmith Currently we don't validate the `` component of the filter string. For example, if `foo` is a non-existing parameter or metric for a set of experiments, we return results...
I’ve started but it’s a little more complicated so realistically it’ll be hard to get done by Tuesday. Rachel, is it important for the presentation? I could skip some of...
Okay, I will work on that tomorrow.
PR: https://github.com/mir-dataset-loaders/mirdata/pull/149
@expectopatronum @rabitt I should be able to get this wrapped up before the end of this week, sorry it's taken so long!
Hi @expectopatronum, this work was blocked by a refactoring piece. But that's done now, so I should hopefully have time to finish this within the next week or so. Thanks...