[bug]: Lora not working (RuntimeError: output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320])

[pivot69]

Is there an existing issue for this?

• [X] I have searched the existing issues

OS

macOS

GPU

mps

VRAM

64gb

What version did you experience this issue on?

2.3.5

What happened?

Loras placed in lora directory is recognised fine, but when I try to include it in a prompt it will only result in an error:

` | Loading lora 21NEIGHBORS.safetensors with weight 0.75 | Found lora 21NEIGHBORS at /Users/pivot/invokeai/loras/21NEIGHBORS.safetensors ** Could not generate image. output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320]

Traceback (most recent call last): File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/invokeai/backend/invoke_ai_web_server.py", line 1344, in generate_images self.generate.prompt2image( File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/ldm/generate.py", line 525, in prompt2image uc, c, extra_conditioning_info = get_uc_and_c_and_ec( File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/ldm/invoke/conditioning.py", line 79, in get_uc_and_c_and_ec with InvokeAIDiffuserComponent.custom_attention_context(model.unet, File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/contextlib.py", line 135, in enter return next(self.gen) File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/ldm/models/diffusion/shared_invokeai_diffusion.py", line 103, in custom_attention_context condition() # target model is stored in condition state for some reason File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/ldm/modules/lora_manager.py", line 43, in call self.kohya_manager.apply_lora_model(self.name,self.weight) File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/ldm/modules/kohya_lora_manager.py", line 508, in apply_lora_model lora = self.load_lora_module(name, path_file, mult) File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/ldm/modules/kohya_lora_manager.py", line 489, in load_lora_module lora.load_from_dict(checkpoint) File "/Users/pivot/invokeai/.venv/lib/python3.10/site-packages/ldm/modules/kohya_lora_manager.py", line 366, in load_from_dict layer_down.weight.copy_(value_down) RuntimeError: output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320]

`

Screenshots

No response

Additional context

No response

Contact Details

No response