Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Remove convnext | Volpeon | 2023-06-22 | 7 | -482/+0 |
| | |||||
* | Update | Volpeon | 2023-06-21 | 5 | -36/+11 |
| | |||||
* | Update | Volpeon | 2023-06-21 | 24 | -989/+1873 |
| | |||||
* | Fix LoRA training with DAdan | Volpeon | 2023-05-16 | 5 | -43/+42 |
| | |||||
* | Avoid model recompilation due to varying prompt lengths | Volpeon | 2023-05-16 | 5 | -65/+43 |
| | |||||
* | Patch xformers to cast dtypes | Volpeon | 2023-05-16 | 2 | -7/+17 |
| | |||||
* | Update deps | Volpeon | 2023-05-16 | 1 | -3/+3 |
| | |||||
* | LoRA: Apply to out layers as well | Volpeon | 2023-05-16 | 1 | -2/+4 |
| | |||||
* | Support LoRA training for token embeddings | Volpeon | 2023-05-16 | 5 | -25/+36 |
| | |||||
* | Fix for latest PEFT | Volpeon | 2023-05-12 | 3 | -156/+50 |
| | |||||
* | Update | Volpeon | 2023-05-11 | 4 | -13/+45 |
| | |||||
* | Update | Volpeon | 2023-05-06 | 5 | -13/+205 |
| | |||||
* | Update | Volpeon | 2023-05-05 | 5 | -32/+71 |
| | |||||
* | torch.compile won't work yet, keep code prepared | Volpeon | 2023-04-29 | 4 | -4/+52 |
| | |||||
* | Optional xformers | Volpeon | 2023-04-29 | 1 | -3/+10 |
| | |||||
* | Fixed model compilation | Volpeon | 2023-04-29 | 2 | -3/+9 |
| | |||||
* | Support torch.compile | Volpeon | 2023-04-28 | 3 | -9/+27 |
| | |||||
* | Fixed loss/acc logging | Volpeon | 2023-04-28 | 2 | -17/+22 |
| | |||||
* | Fix | Volpeon | 2023-04-27 | 1 | -8/+19 |
| | |||||
* | Update | Volpeon | 2023-04-27 | 7 | -41/+119 |
| | |||||
* | Fix cycle loop | Volpeon | 2023-04-23 | 2 | -6/+10 |
| | |||||
* | Update | Volpeon | 2023-04-21 | 4 | -8/+12 |
| | |||||
* | Fix PTI | Volpeon | 2023-04-20 | 1 | -2/+4 |
| | |||||
* | Update | Volpeon | 2023-04-20 | 3 | -34/+126 |
| | |||||
* | Fix | Volpeon | 2023-04-17 | 3 | -7/+6 |
| | |||||
* | Fix | Volpeon | 2023-04-16 | 1 | -1/+1 |
| | |||||
* | Improved automation caps | Volpeon | 2023-04-16 | 6 | -54/+100 |
| | |||||
* | Added option to use constant LR on cycles > 1 | Volpeon | 2023-04-16 | 2 | -2/+20 |
| | |||||
* | Fix | Volpeon | 2023-04-16 | 1 | -1/+1 |
| | |||||
* | Fix | Volpeon | 2023-04-16 | 6 | -90/+177 |
| | |||||
* | Fix | Volpeon | 2023-04-16 | 1 | -16/+18 |
| | |||||
* | Fix | Volpeon | 2023-04-16 | 2 | -23/+21 |
| | |||||
* | Update | Volpeon | 2023-04-16 | 6 | -32/+69 |
| | |||||
* | Fix | Volpeon | 2023-04-15 | 1 | -1/+1 |
| | |||||
* | Fix | Volpeon | 2023-04-15 | 3 | -24/+5 |
| | |||||
* | TI via LoRA | Volpeon | 2023-04-15 | 9 | -140/+212 |
| | |||||
* | Added cycle LR decay | Volpeon | 2023-04-13 | 3 | -9/+16 |
| | |||||
* | Update | Volpeon | 2023-04-13 | 8 | -67/+80 |
| | |||||
* | Experimental convnext discriminator support | Volpeon | 2023-04-11 | 4 | -25/+109 |
| | |||||
* | Store sample images in Tensorboard as well | Volpeon | 2023-04-11 | 1 | -13/+12 |
| | |||||
* | Update | Volpeon | 2023-04-11 | 4 | -27/+43 |
| | |||||
* | Randomize dataset across cycles | Volpeon | 2023-04-10 | 3 | -12/+12 |
| | |||||
* | Update | Volpeon | 2023-04-10 | 4 | -27/+30 |
| | |||||
* | Fix sample gen: models sometimes weren't in eval mode | Volpeon | 2023-04-10 | 4 | -28/+29 |
| | |||||
* | Update | Volpeon | 2023-04-10 | 6 | -55/+89 |
| | |||||
* | Update | Volpeon | 2023-04-09 | 4 | -12/+33 |
| | |||||
* | Fix | Volpeon | 2023-04-09 | 1 | -1/+1 |
| | |||||
* | Update | Volpeon | 2023-04-09 | 1 | -6/+1 |
| | |||||
* | Update | Volpeon | 2023-04-09 | 7 | -160/+72 |
| | |||||
* | Fix | Volpeon | 2023-04-09 | 2 | -5/+7 |
| |