mirror of
https://github.com/ostris/ai-toolkit.git
synced 2026-04-28 10:11:14 +00:00
added prompt dropout to happen indempendently on each TE
This commit is contained in:
@@ -194,6 +194,9 @@ class TrainConfig:
|
||||
self.noise_multiplier = kwargs.get('noise_multiplier', 1.0)
|
||||
self.img_multiplier = kwargs.get('img_multiplier', 1.0)
|
||||
|
||||
# dropout that happens before encoding. It functions independently per text encoder
|
||||
self.prompt_dropout_prob = kwargs.get('prompt_dropout_prob', 0.0)
|
||||
|
||||
# match the norm of the noise before computing loss. This will help the model maintain its
|
||||
# current understandin of the brightness of images.
|
||||
|
||||
|
||||
Reference in New Issue
Block a user