Commit message (Expand) | Author | Age | Files | Lines | ||
---|---|---|---|---|---|---|
... | ||||||
* | Improved inverted tokens | Volpeon | 2023-03-26 | 1 | -1/+15 | |
* | Update | Volpeon | 2023-03-25 | 1 | -4/+10 | |
* | Update | Volpeon | 2023-03-24 | 1 | -3/+3 | |
* | Bring back Perlin offset noise | Volpeon | 2023-03-23 | 1 | -0/+7 | |
* | Update | Volpeon | 2023-03-23 | 1 | -1/+1 | |
* | Fix | Volpeon | 2023-03-22 | 1 | -4/+0 | |
* | Log DAdam/DAdan d | Volpeon | 2023-03-21 | 1 | -2/+2 | |
* | Added dadaptation | Volpeon | 2023-03-21 | 1 | -0/+28 | |
* | Fixed SNR weighting, re-enabled xformers | Volpeon | 2023-03-21 | 1 | -2/+2 | |
* | Test: https://arxiv.org/pdf/2303.09556.pdf | Volpeon | 2023-03-17 | 1 | -12/+26 | |
* | Update | Volpeon | 2023-03-07 | 1 | -4/+4 | |
* | Pipeline: Perlin noise for init image | Volpeon | 2023-03-04 | 1 | -1/+1 | |
* | Removed offset noise from training, added init offset to pipeline | Volpeon | 2023-03-03 | 1 | -1/+0 | |
* | Implemented different noise offset | Volpeon | 2023-03-03 | 1 | -2/+2 | |
* | Update | Volpeon | 2023-03-01 | 1 | -1/+1 | |
* | Don't rely on Accelerate for gradient accumulation | Volpeon | 2023-02-21 | 1 | -1/+1 | |
* | Embedding normalization: Ignore tensors with grad = 0 | Volpeon | 2023-02-21 | 1 | -11/+10 | |
* | Update | Volpeon | 2023-02-18 | 1 | -2/+4 | |
* | Added Lion optimizer | Volpeon | 2023-02-17 | 1 | -11/+27 | |
* | Back to xformers | Volpeon | 2023-02-17 | 1 | -3/+2 | |
* | Remove xformers, switch to Pytorch Nightly | Volpeon | 2023-02-17 | 1 | -2/+4 | |
* | Integrated WIP UniPC scheduler | Volpeon | 2023-02-16 | 1 | -1/+2 | |
* | Update | Volpeon | 2023-02-15 | 1 | -0/+1 | |
* | Update | Volpeon | 2023-02-13 | 1 | -8/+11 | |
* | Integrate Self-Attention-Guided (SAG) Stable Diffusion in my custom pipeline | Volpeon | 2023-02-08 | 1 | -1/+1 | |
* | Fixed Lora training | Volpeon | 2023-02-08 | 1 | -6/+6 | |
* | Add Lora | Volpeon | 2023-02-07 | 1 | -4/+6 | |
* | Restored LR finder | Volpeon | 2023-01-20 | 1 | -2/+19 | |
* | Move Accelerator preparation into strategy | Volpeon | 2023-01-19 | 1 | -3/+3 | |
* | Smaller emb decay | Volpeon | 2023-01-17 | 1 | -1/+1 | |
* | Make embedding decay work like Adam decay | Volpeon | 2023-01-17 | 1 | -12/+4 | |
* | Update | Volpeon | 2023-01-17 | 1 | -49/+64 | |
* | Training update | Volpeon | 2023-01-16 | 1 | -3/+5 | |
* | If valid set size is 0, re-use one image from train set | Volpeon | 2023-01-16 | 1 | -5/+1 | |
* | Moved multi-TI code from Dreambooth to TI script | Volpeon | 2023-01-16 | 1 | -107/+114 | |
* | More training adjustments | Volpeon | 2023-01-16 | 1 | -5/+12 | |
* | Handle empty validation dataset | Volpeon | 2023-01-16 | 1 | -6/+3 | |
* | Implemented extended Dreambooth training | Volpeon | 2023-01-16 | 1 | -40/+22 | |
* | Added Dreambooth strategy | Volpeon | 2023-01-15 | 1 | -23/+23 | |
* | Restored functional trainer | Volpeon | 2023-01-15 | 1 | -61/+21 | |
* | Update | Volpeon | 2023-01-15 | 1 | -36/+38 | |
* | Removed unused code, put training callbacks in dataclass | Volpeon | 2023-01-15 | 1 | -48/+1 | |
* | Added functional TI strategy | Volpeon | 2023-01-15 | 1 | -78/+30 | |
* | Added functional trainer | Volpeon | 2023-01-15 | 1 | -26/+23 | |
* | Update | Volpeon | 2023-01-14 | 1 | -5/+5 | |
* | Update | Volpeon | 2023-01-14 | 1 | -3/+4 | |
* | WIP: Modularization ("free(): invalid pointer" my ass) | Volpeon | 2023-01-14 | 1 | -59/+15 | |
* | TI: Prepare UNet with Accelerate as well | Volpeon | 2023-01-14 | 1 | -12/+15 | |
* | Fix | Volpeon | 2023-01-14 | 1 | -2/+2 | |
* | Cleanup | Volpeon | 2023-01-14 | 1 | -21/+12 |