summaryrefslogtreecommitdiffstats
path: root/data
Commit message (Collapse)AuthorAgeFilesLines
* Avoid model recompilation due to varying prompt lengthsVolpeon2023-05-161-7/+24
|
* UpdateVolpeon2023-05-051-2/+2
|
* UpdateVolpeon2023-04-162-6/+21
|
* Randomize dataset across cyclesVolpeon2023-04-101-9/+5
|
* Add color jitterVolpeon2023-04-051-10/+19
|
* Fix memory leakVolpeon2023-04-032-5/+5
|
* Update dataset format: Separate prompt and keywordsVolpeon2023-04-022-42/+37
|
* Support num_train_steps arg againVolpeon2023-03-281-2/+6
|
* Improved inverted tokensVolpeon2023-03-261-23/+44
|
* Fix training with guidanceVolpeon2023-03-251-3/+5
|
* UpdateVolpeon2023-03-251-7/+14
|
* UpdateVolpeon2023-03-231-1/+1
|
* UpdateVolpeon2023-02-151-0/+1
|
* Improved batch paddingVolpeon2023-02-151-29/+26
|
* Better batch fillingVolpeon2023-02-151-3/+6
|
* Better batch filling behaviorVolpeon2023-02-151-3/+7
|
* Dataset: Repeat data to fill batch to batch_sizeVolpeon2023-02-151-0/+3
|
* UpdateVolpeon2023-02-131-4/+4
|
* UpdateVolpeon2023-01-171-1/+1
|
* If valid set size is 0, re-use one image from train setVolpeon2023-01-161-1/+1
|
* Moved multi-TI code from Dreambooth to TI scriptVolpeon2023-01-161-1/+1
|
* More training adjustmentsVolpeon2023-01-161-18/+21
|
* Pad dataset if len(items) < batch_sizeVolpeon2023-01-161-0/+3
|
* Handle empty validation datasetVolpeon2023-01-161-20/+27
|
* Extended Dreambooth: Train TI tokens separatelyVolpeon2023-01-161-0/+7
|
* Implemented extended Dreambooth trainingVolpeon2023-01-161-0/+19
|
* Restored functional trainerVolpeon2023-01-151-8/+0
|
* UpdateVolpeon2023-01-151-7/+4
|
* FixVolpeon2023-01-141-1/+1
|
* CleanupVolpeon2023-01-141-13/+8
|
* Removed PromptProcessor, modularized training loopVolpeon2023-01-131-16/+20
|
* Code deduplicationVolpeon2023-01-131-5/+42
|
* Added arg to disable tag shufflingVolpeon2023-01-101-1/+3
|
* Enable buckets for validation, fixed vaildation repeat argVolpeon2023-01-091-2/+3
|
* Add --valid_set_repeatVolpeon2023-01-091-1/+5
|
* No cache after allVolpeon2023-01-081-17/+7
|
* Cache token IDs in datasetVolpeon2023-01-081-8/+20
|
* FixVolpeon2023-01-081-5/+5
|
* Improved aspect ratio bucketingVolpeon2023-01-081-3/+19
|
* Fixed aspect ratio bucketingVolpeon2023-01-081-3/+5
|
* CleanupVolpeon2023-01-081-15/+25
|
* Fixed aspect ratio bucketing; allow passing token IDs to pipelineVolpeon2023-01-081-35/+43
|
* Improved aspect ratio bucketingVolpeon2023-01-081-119/+154
|
* Made aspect ratio bucketing configurableVolpeon2023-01-071-11/+22
|
* Added progressive aspect ratio bucketingVolpeon2023-01-071-56/+88
|
* FixVolpeon2023-01-061-2/+4
|
* UpdateVolpeon2023-01-052-21/+22
|
* UpdateVolpeon2023-01-041-1/+1
|
* Various updatesVolpeon2023-01-041-18/+27
|
* Fixed reproducibility, more consistant validationVolpeon2023-01-041-3/+7
|