dxqb
dxqb
any UI solution would have to reliably stop tensorboard on exit, but also still automatically kill tensorboard directly after a CLI or cloud training, because otherwise the process never ends....
any ideas how a good UI can look like? First you'd have to detach the UI train config from the currently running training task, before the user can set the...
can be closed. not a bug in OT, but different behaviour of Prodigy based on different settings. It doesn't surprise me that Prodigy fails if Bias Correction is disabled. I'm...
closing as discussed, bug no longer occurs
> But, since there is also the need of modifying the ADA model structure itself (remove norm layers and add 1/4 pixel norm layers), this involves considerable effort. Isn't this...
interesting, but not for finetuning
I think this is outdated. Timestep weighting is implemented now.
If no samples .json exists when you press "Start Training", a "samples.json" is created, even if you have not created any samples in the UI. This is probably an obsolete...
ok, sorry. I thought we don't understand the issue because no reply. reopened.