summaryrefslogtreecommitdiffstats
path: root/data
Commit message (Collapse)AuthorAgeFilesLines
* Better dataset prompt handlingVolpeon2022-12-231-9/+16
|
* num_class_images is now class images per train imageVolpeon2022-12-231-2/+2
|
* Improved class prompt handlingVolpeon2022-12-231-17/+12
|
* Dependency cleanup/upgradesVolpeon2022-12-201-3/+2
|
* Improved dataset prompt handling, fixedVolpeon2022-12-191-18/+23
|
* Avoid increased VRAM usage on validationVolpeon2022-12-151-3/+10
|
* More generic datset filterVolpeon2022-12-141-5/+5
|
* Optimized Textual Inversion training by filtering dataset by existence of ↵Volpeon2022-12-131-7/+31
| | | | added tokens
* Training improvements such as tag dropoutVolpeon2022-12-111-3/+10
|
* Various updated; shuffle prompt content during trainingVolpeon2022-12-101-9/+20
|
* UpdateVolpeon2022-11-141-1/+1
|
* UpdateVolpeon2022-11-011-10/+1
|
* Added CLI arg to set dataloader worker num; improved text encoder handling ↵Volpeon2022-10-271-3/+7
| | | | with Dreambooth
* Advanced datasetsVolpeon2022-10-241-25/+39
|
* UpdateVolpeon2022-10-241-17/+16
|
* Revert lat; fix skip attribute in datasetVolpeon2022-10-221-1/+1
|
* Improved prompt handlingVolpeon2022-10-181-46/+37
|
* Implemented extended prompt limitVolpeon2022-10-171-4/+10
|
* Added support for Aesthetic GradientsVolpeon2022-10-131-1/+1
|
* Various updatesVolpeon2022-10-121-2/+1
|
* UpdateVolpeon2022-10-081-77/+85
|
* Training: Create multiple class images per training imageVolpeon2022-10-071-14/+40
|
* UpdateVolpeon2022-10-062-150/+0
|
* Add Textual Inversion with class dataset (a la Dreambooth)Volpeon2022-10-042-25/+4
|
* BugfixVolpeon2022-10-041-2/+2
|
* Dreambooth: Generate specialized class images from input promptsVolpeon2022-10-033-62/+57
|
* Small perf improvementsVolpeon2022-10-032-4/+5
|
* Added negative prompt support for training scriptsVolpeon2022-10-032-13/+19
|
* Fixed euler_a generator argumentVolpeon2022-10-032-5/+9
|
* Assign unused images in validation dataset to train datasetVolpeon2022-10-032-5/+15
|
* Use euler_a for samples in learning scripts; backported improvement from ↵Volpeon2022-10-032-53/+46
| | | | Dreambooth to Textual Inversion
* Batches of size 1 cause error: Expected query.is_contiguous() to be true, ↵Volpeon2022-09-281-9/+9
| | | | but got false
* Improved sample output and progress barsVolpeon2022-09-281-5/+5
|
* Various improvements, added inference scriptVolpeon2022-09-271-0/+1
|
* Supply dataset CSV file instead of dir with hardcoded CSV filenameVolpeon2022-09-271-10/+14
|
* Added Dreambooth training scriptVolpeon2022-09-272-0/+193
|
* Undo textual inversion dataset improvementsVolpeon2022-09-271-10/+21
|
* More cleanupVolpeon2022-09-271-0/+134