summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* Bugfixes for multi-vector token handlingVolpeon2022-12-314-27/+53
|
* Simplified multi-vector embedding codeVolpeon2022-12-313-17/+14
|
* FixesVolpeon2022-12-312-3/+2
|
* Added multi-vector embeddingsVolpeon2022-12-318-81/+299
|
* Misc improvementsVolpeon2022-12-303-78/+56
|
* Training script improvementsVolpeon2022-12-305-25/+89
|
* UpdateVolpeon2022-12-292-11/+16
|
* Training improvementsVolpeon2022-12-293-29/+42
|
* Updated 1-cycle schedulerVolpeon2022-12-282-9/+15
|
* Integrated updates from diffusersVolpeon2022-12-285-39/+59
|
* Improved learning rate finderVolpeon2022-12-273-23/+30
|
* Added validation phase to learn rate finderVolpeon2022-12-272-15/+31
|
* Added learning rate finderVolpeon2022-12-274-162/+257
|
* Set default dimensions to 768; add config inheritanceVolpeon2022-12-267-26/+36
|
* Code simplifications, avoid autocastVolpeon2022-12-254-82/+92
|
* UpdateVolpeon2022-12-253-3/+3
|
* UpdateVolpeon2022-12-241-3/+3
|
* Fixed Textual InversionVolpeon2022-12-243-13/+21
|
* Training updateVolpeon2022-12-243-24/+8
|
* UpdateVolpeon2022-12-243-10/+26
|
* FixVolpeon2022-12-241-1/+1
|
* Fix againVolpeon2022-12-241-1/+1
|
* FixVolpeon2022-12-242-3/+6
|
* Better dataset prompt handlingVolpeon2022-12-233-11/+18
|
* num_class_images is now class images per train imageVolpeon2022-12-233-4/+4
|
* Improved class prompt handlingVolpeon2022-12-235-63/+31
|
* FixVolpeon2022-12-231-3/+1
|
* Simplified trainable embedding code againVolpeon2022-12-231-19/+6
|
* Fixed and simplified trainable embeddings codeVolpeon2022-12-231-12/+12
|
* Fixed Textual InversionVolpeon2022-12-223-14/+13
|
* Improved Textual Inversion: Completely exclude untrained embeddings from ↵Volpeon2022-12-222-18/+76
| | | | training
* Some LoRA fixes (still broken)Volpeon2022-12-212-138/+89
|
* Moved common training code into separate moduleVolpeon2022-12-214-231/+203
|
* Fix trainingVolpeon2022-12-205-15/+1073
|
* Dependency cleanup/upgradesVolpeon2022-12-202-27/+12
|
* Fix Textual Inversion dataset filteringVolpeon2022-12-201-1/+5
|
* Upstream patchesVolpeon2022-12-191-5/+17
|
* Improved dataset prompt handling, fixedVolpeon2022-12-194-23/+25
|
* Avoid increased VRAM usage on validationVolpeon2022-12-154-12/+29
|
* More generic datset filterVolpeon2022-12-145-13/+25
|
* Unified loading of TI embeddingsVolpeon2022-12-134-65/+54
|
* Optimized Textual Inversion training by filtering dataset by existence of ↵Volpeon2022-12-133-10/+47
| | | | added tokens
* Fixed sample/checkpoint frequencyVolpeon2022-12-132-14/+10
|
* Add support for resume in Textual InversionVolpeon2022-12-134-111/+101
|
* Dreambooth: Support loading Textual Inversion embeddingsVolpeon2022-12-124-31/+57
|
* Remove embedding checkpoints from Dreambooth trainingVolpeon2022-12-111-40/+16
|
* Package updatesVolpeon2022-12-111-3/+3
|
* Training improvements such as tag dropoutVolpeon2022-12-112-36/+68
|
* Support attention_mask of text encoderVolpeon2022-12-104-15/+29
|
* Various updated; shuffle prompt content during trainingVolpeon2022-12-103-18/+48
|