Closed Toshik-One closed 1 year ago
Hmm is it possible that this LoRA doesn't have a UNet trained? I can look into fixing it so it won't error but curious.
this LoRA https://civitai.com/models/48139/lowra
Hmm seems to work on my end. Maybe download it again to confirm it is all there? I will still look into the missing UNet because that's a thing that should be supported.
python lora-inspector.py -w /mnt/900/lora/LowRA.safetensors
/mnt/900/lora/LowRA.safetensors
model key count: 792
UNet attention weight average magnitude: 2.6101634470841035
UNet attention weight average strength: 0.010605590754123827
Text Encoder weight average magnitude: 2.3996765103651425
Text Encoder weight average strength: 0.008541954607252404
This happened to me too after a restart. But to save the meta data did not work.
Seems those LoRA's do not have any metadata. I fixed them so they will still go through the process but the result will limited.
{"file": "/mnt/900/lora/LowRA.safetensors", "filename": "LowRA.safetensors"}
Also fixed missing UNet which should pass through fine now.
Thanks for reporting! Let me know if you have any issues with these.
Please submit the following keys so we can get a parser made for it: dict_keys(['format']) model key count: 1145 Traceback (most recent call last): File "/content/lora-inspector/lora-inspector.py", line 397, in
results = process(args)
File "/content/lora-inspector/lora-inspector.py", line 372, in process
return process_safetensor_file(file, args)
File "/content/lora-inspector/lora-inspector.py", line 262, in process_safetensor_file
find_vectors_weights(f)
File "/content/lora-inspector/lora-inspector.py", line 175, in find_vectors_weights
avg_mag = sum_mag / num_results
ZeroDivisionError: division by zero