Resume experiments

Open bfirsh opened this issue 5 years ago • 1 comments

Why?

Sometimes the training process dies, and you want to resume from a checkpoint, keeping your partially trained model weights and metric history.

How?

Editing an existing experiment might be problematic (see #177). Perhaps we could create a new experiment with all the data from the previous experiment but with a new ID somehow.

Nov 07 '20 23:11 bfirsh

Commenting to say that I love the library so far. Being able to resume experiments would be really really helpful to me, so if there's anything I can do to improve the release time of this feature, I'd be interested in helping.

Mar 14 '21 18:03 samuelstevens