ValueError: Checkpoint not supported
ValueError: Checkpoint not supported error code when attempting to load this William Eggleston style LoRA safetensor (wegg.safetensors) on top of the SD-XL-v1.0 base model. I'm following the recommendations from HF at: https://huggingface.co/TheLastBen/William_Eggleston_Style_SDXL
- Successfully loaded and inferenced the SD-XL-v1.0 base model:
base_model_id = "stabilityai/stable-diffusion-xl-base-1.0"pipeline = StableDiffusionXLPipeline.from_pretrained(base_model_id, variant="fp16", torch_dtype=torch.float16,).to("cuda")
- Error code arises when attempting to load LoRA weights:
pipeline.load_lora_weights("TheLastBen/William_Eggleston_Style_SDXL", weight_name="wegg.safetensors",)
Using diffusers version: 0.20.0.
Suggestions on how to successfully load these LoRA weights? This same pipeline is successful in loading the LoRA weights for: https://huggingface.co/nerijs/pixel-art-xl This result leads me to believe the issue might be in the .safetensors file and not the diffusers package. Thanks for your suggestions!
"Diffusers is working fast to support all LoRA loading with the native method pipe.load_lora_weights. Until then, bmaltais kohnya_ss lora method has worked..."
Work-around found here:
https://colab.research.google.com/drive/14aEJsKdEQ9_kyfsiV6JDok799kxPul0j
checkout the code used here https://huggingface.co/spaces/multimodalart/LoraTheExplorer/tree/main
Thank you!