d-adaptation notes

.gitattributes

@@ -32,3 +32,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.jpg filter=lfs diff=lfs merge=lfs -text
+*.jpeg filter=lfs diff=lfs merge=lfs -text
+*.png filter=lfs diff=lfs merge=lfs -text

d-adaptation/notes.md

@@ -0,0 +1,21 @@
+# D-Adaptation Experiment Notes
+
+## Learning rates
+Unet 1, text 0.5 as seen in thread: https://twitter.com/kohya_tech/status/1627194651034943490?cxt=HHwWhIDUtb66-pQtAAAA
+
+## Alpha
+Alpha=Dim was recommended in the github thread https://github.com/kohya-ss/sd-scripts/issues/181
+I have tried dim 8 alpha 1 with success as well as failure. Both Amber and Castoria are alpha=1 and seem to work fine. 
+UMP ends up with image generations that look like a single brown square, still testing if alpha has a relationship to this issue.
+As noted in the same github issue, alpha/rank scaling modifies the gradient update to become smaller and thus d-adaptation to boost the learning rate. This could be the reason why it goes bad.
+
+## Dim
+128 dim shows some local noisy patterns. Reranking the model to a lower dim from 128 doesn't get rid of it. Converting the weights of the last up block in the unet does but also causes a noticable change in the generated character. Obviously you could reduce the last up block by a smaller amount.
+Lower dims show good performance. Need much larger test to check for accuracy between them. 
+
+## Resolution
+To be tested
+
+## 2.X models
+To be tested. 
+Candidate base models: wd1.5, replicant, subtly