leonary

When performing VAE encoding on high-resolution images, there is a high probability that the original image will be changed and artifacts will be generated.

3

### Expected Behavior Since upscaling is an important feature, it is very important to remove these artifacts to preserve the original image content. These artifacts are difficult to remove at...

SD3 cannot achieve good fitting results on large data sets

30

I conducted several tests. DAdaptAdam achieved similar fitting results to SD15/SDXL on a dataset consisting of a single image. However, on datasets with hundreds of images, the fitting speed of...

[Bug]: Trying to overfit the SD3 Lora model on a dataset of one image

1

### What happened? I am used to overfitting the model on a dataset with only one image to observe the performance of the model. I got the expected overfitting results...

Commit "Update T5 attention mask handling in FLUX" results in an increase of tens of GB of video memory usage when using --apply_t5_attn_mask

8

Commit "Update T5 attention mask handling in FLUX" results in an increase of tens of GB of video memory usage when using --apply_t5_attn_mask.