lora icon indicating copy to clipboard operation
lora copied to clipboard

ValueError: No lora injected.

Open Otakufantasy opened this issue 9 months ago • 1 comments

i meet with ValueError: No lora injected when i am running run_lorpt.sh. and i have tried the methods of issues/273 and they don't work. here are the details: Before training: Unet First Layer lora up tensor([[0., 0., 0., 0.], [0., 0., 0., 0.], [0., 0., 0., 0.], ..., [0., 0., 0., 0.], [0., 0., 0., 0.], [0., 0., 0., 0.]]) Before training: Unet First Layer lora down tensor([[-0.1093, 0.0082, -0.4515, ..., 0.5442, -0.1840, 0.2209], [ 0.4050, -0.4351, 0.2604, ..., -0.1007, 0.2565, -0.0483], [ 0.1900, 0.5735, -0.1318, ..., 0.1821, 0.2752, 0.3000], [ 0.1973, 0.1657, -0.1456, ..., -0.2453, -0.1598, 0.0354]]) Traceback (most recent call last): File "/home/featurize/work/lora/training_scripts/train_lora_w_ti.py", line 1223, in main(args) File "/home/featurize/work/lora/training_scripts/train_lora_w_ti.py", line 807, in main for _up, _down in extract_lora_ups_down( ^^^^^^^^^^^^^^^^^^^^^^ File "/environment/miniconda3/lib/python3.11/site-packages/lora_diffusion-0.1.7-py3.11.egg/lora_diffusion/lora.py", line 395, in extract_lora_ups_down raise ValueError("No lora injected.") ValueError: No lora injected.

Otakufantasy avatar Feb 16 '25 04:02 Otakufantasy