Commit message (Expand) | Author | Age | Files | Lines | ||
---|---|---|---|---|---|---|
... | ||||||
* | Update | Volpeon | 2023-04-02 | 1 | -3/+3 | |
* | Lora: Only register params with grad to optimizer | Volpeon | 2023-04-02 | 3 | -8/+7 | |
* | Revert | Volpeon | 2023-04-01 | 4 | -64/+132 | |
* | Fix | Volpeon | 2023-04-01 | 1 | -1/+3 | |
* | Combined TI with embedding and LoRA | Volpeon | 2023-04-01 | 3 | -87/+38 | |
* | Experimental: TI via LoRA | Volpeon | 2023-04-01 | 3 | -63/+44 | |
* | Update | Volpeon | 2023-04-01 | 5 | -10/+18 | |
* | Add support for Adafactor, add TI initializer noise | Volpeon | 2023-04-01 | 6 | -6/+67 | |
* | Update | Volpeon | 2023-03-31 | 3 | -3/+9 | |
* | Update | Volpeon | 2023-03-31 | 5 | -24/+32 | |
* | Fix | Volpeon | 2023-03-31 | 3 | -5/+6 | |
* | Fix | Volpeon | 2023-03-31 | 1 | -1/+1 | |
* | Support Dadaptation d0, adjust sample freq when steps instead of epochs are used | Volpeon | 2023-03-31 | 3 | -12/+33 | |
* | Fix | Volpeon | 2023-03-31 | 4 | -6/+9 | |
* | Fix | Volpeon | 2023-03-28 | 3 | -3/+3 | |
* | Support num_train_steps arg again | Volpeon | 2023-03-28 | 4 | -23/+49 | |
* | Fix TI | Volpeon | 2023-03-27 | 2 | -17/+33 | |
* | Fix TI | Volpeon | 2023-03-27 | 2 | -26/+19 | |
* | Fix TI | Volpeon | 2023-03-27 | 2 | -9/+11 | |
* | Revert to regular embeddings | Volpeon | 2023-03-27 | 1 | -19/+15 | |
* | Sparse TI embeddings without sparse tensors | Volpeon | 2023-03-27 | 2 | -27/+31 | |
* | Fix TI embeddings init | Volpeon | 2023-03-26 | 1 | -1/+1 | |
* | Improved TI embeddings | Volpeon | 2023-03-26 | 2 | -9/+24 | |
* | Improved inverted tokens | Volpeon | 2023-03-26 | 5 | -34/+70 | |
* | Fix training with guidance | Volpeon | 2023-03-25 | 2 | -6/+14 | |
* | Update | Volpeon | 2023-03-25 | 8 | -987/+99 | |
* | Update | Volpeon | 2023-03-24 | 4 | -9/+22 | |
* | Fixed Lora training perf issue | Volpeon | 2023-03-24 | 2 | -9/+9 | |
* | Lora fix: Save config JSON, too | Volpeon | 2023-03-24 | 1 | -0/+3 | |
* | Refactoring, fixed Lora training | Volpeon | 2023-03-24 | 5 | -66/+104 | |
* | Bring back Perlin offset noise | Volpeon | 2023-03-23 | 2 | -1/+21 | |
* | Update | Volpeon | 2023-03-23 | 10 | -30/+29 | |
* | Fix | Volpeon | 2023-03-22 | 3 | -6/+2 | |
* | Log DAdam/DAdan d | Volpeon | 2023-03-21 | 4 | -6/+20 | |
* | Added dadaptation | Volpeon | 2023-03-21 | 4 | -1/+85 | |
* | Fixed SNR weighting, re-enabled xformers | Volpeon | 2023-03-21 | 5 | -50/+97 | |
* | Restore min SNR | Volpeon | 2023-03-19 | 1 | -13/+12 | |
* | New loss weighting from arxiv.org:2204.00227 | Volpeon | 2023-03-18 | 1 | -2/+5 | |
* | Better SNR weighting | Volpeon | 2023-03-18 | 1 | -3/+3 | |
* | Fixed snr weight calculation | Volpeon | 2023-03-17 | 1 | -1/+4 | |
* | Fix loss=nan | Volpeon | 2023-03-17 | 1 | -2/+2 | |
* | Test: https://arxiv.org/pdf/2303.09556.pdf | Volpeon | 2023-03-17 | 3 | -39/+56 | |
* | Update | Volpeon | 2023-03-07 | 5 | -27/+37 | |
* | Update | Volpeon | 2023-03-06 | 3 | -25/+13 | |
* | More flexible pipeline wrt init noise | Volpeon | 2023-03-04 | 1 | -13/+44 | |
* | Added Perlin noise to training | Volpeon | 2023-03-04 | 3 | -8/+28 | |
* | Update | Volpeon | 2023-03-04 | 1 | -9/+2 | |
* | Pipeline: Perlin noise for init image | Volpeon | 2023-03-04 | 7 | -18/+70 | |
* | Pipeline: Improved initial image generation | Volpeon | 2023-03-04 | 1 | -23/+26 | |
* | Changed init noise algorithm | Volpeon | 2023-03-03 | 1 | -3/+11 |