diffusers
diffusers copied to clipboard
TheLastBen
🤗 Diffusers: State-of-the-art diffusion models for image and audio generation in PyTorch
in SDXL-LoRA-RNPD.ipynb, when `External_Captions` is set to `TRUE` and captions exceed 77 tokens, it will cause a device mismatch and result in not training the TE for all the epochs...
Consistent with [train_dreambooth_rnpd_sdxl_lora.py](https://github.com/TheLastBen/diffusers/blob/b016ec6cf4ca3f850fdc145faffcb79174865d6f/examples/dreambooth/train_dreambooth_rnpd_sdxl_lora.py#L501C29-L501C29)
The model is not saved in the output directory. With this change, the model now is saved again.
Initially reported here https://github.com/TheLastBen/fast-stable-diffusion/issues/1174#issuecomment-1366085743 This keeps dashes, for things words like `x-men`, `t-shirt`, or wanting a trigger like `wa-vy` or `mdjrny-v4 style` (real high profile examples). It also preserves numbers...
I was using fast-stable-diffusion notebook of yours and I realized that your train_dreambooth.py was not using xformers so I checked out your code and the original diffusers. Based on this...
Metadata
Owner
Metadata
🤗 Diffusers: State-of-the-art diffusion models for image and audio generation in PyTorch