So i have a stable diffusion pipeline that already has a base model . I now want to to merge this model with a pretrained lora safetensor model .
when i used a hugging face repo id . i could easily merge it . But when i used a safetensor file it gave me errors .
pipe.unet.load_attn_procs(lora_model_path) was the command used–>gives the same error as below
from lora.lora_diffusion import patch_pipe , tune_lora_scale
import diffusers
from safetensors.torch import load_file
pipe = DiffusionPipeline.from_pretrained(“runwayml/stable-diffusion-v1-5”,use_safetensors=True)
pipe = pipe.to(“mps”)
prompt = “a photo of an astronaut with a pokemon on mars”
lora_model = load_file(“lora/example_loras/arcane_offset.safetensors”, device=“mps”)
patch_pipe(pipe,“lora/example_loras/arcane_offset.safetensors”,
patch_text=True,
patch_ti=True,
patch_unet=True)
error
in parse_safeloras
info = metadata.get(name)
AttributeError: ‘NoneType’ object has no attribute ‘get’
so i tried both patch_pipe and load_attn_procs .gave me the same error