How do you load flux loras in Diffusers?

[markrmiller]

It never works for me, it always complains about a ton of keys. Anyone have the secret?

/miniconda3/envs/kss/lib/python3.10/site-packages/peft/tuners/tuners_utils.py", line 435, in inject_adapter
    raise ValueError(
ValueError: Target modules {'base_model.model.single_transformer_blocks.1.attn.to_q', 'base_model.model.single_transformer_blocks.30.attn.to_v', 'base_model.model.single_transformer_blocks.5.proj_mlp', 'base_model.model.transformer_blocks.11.attn.to_out.0', 'base_model.model.transformer_blocks.11.attn.to_k', 'base_model.model.single_transformer_bl
ocks.17.proj_out', 'base_model.model.transformer_blocks.11.attn.to_v', 'base_model.model.single_transformer_blocks.35.proj_out', 'base_model.model.transformer_blocks.3.attn.add_v_proj', 'base_model.model.single_transformer_blocks.15.proj_out', 'base_model.model.single_transformer_blocks.31.attn.to_k', 'base_model.model.transformer_blocks.9.attn.to
_out.0', 'base_model.model.transformer_blocks.3.attn.add_k_proj', 'base_model.model.single_transformer_blocks.35.attn.to_k', 'base_model.model.transformer_blocks.15.attn.to_k', 'base_model.model.single_transformer_blocks.7.proj_out', 'base_model.model.transformer_blocks.7.attn.add_k_proj', 'base_model.model.transformer_blocks.6.attn.to_q', 'base_m
odel.model.single_transformer_blocks.25.attn.to_q', 'base_model.model.transformer_blocks.7.attn.to_q', 'base_model.model.single_transformer_blocks.25.attn.to_k', 'base_model.model.single_transformer_blocks.26.attn.to_v', 'base_model.model.transformer_blocks.5.attn.add_k_proj', 'base_model.model.single_transformer_blocks.4.attn.to_v', 'base_model.m
odel.single_transformer_blocks.12.proj_mlp', 'base_model.model.transformer_blocks.15.attn.add_q_proj', 'base_model.model.transformer_blocks.4.attn.to_q', 'base_model.model.single_transformer_blocks.20.proj_out', 'base_model.model.transformer_blocks.14.attn.to_k', 'base_model.model.transformer_blocks.7.attn.add_v_proj', 'base_model.model.transforme
r_blocks.4.attn.add_k_proj', 'base_model.model.single_transformer_blocks.21.attn.to_k', 'base_model.model.single_transformer_blocks.28.attn.to_v', 'base_model.model.single_transformer_blocks.2.attn.to_q', 'base_model.model.single_transformer_blocks.0.proj_mlp', 'base_model.model.single_transformer_blocks.23.proj_out', 'base_model.model.transformer
_blocks.5.attn.add_v_proj', 'base_model.model.transformer_blocks.2.attn.add_q_proj', 'base_model.model.single_transformer_blocks.34.attn.to_q', 'base_model.model.transformer_blocks.4.attn.add_q_proj', 'base_model.model.single_transformer_blocks.8.attn.to_v', 'base_model.model.transformer_blocks.12.attn.add_k_proj', 'base_model.model.transformer_bl
ocks.9.attn.to_q', 'base_model.model.single_transformer_blocks.0.attn.to_v', 'base_model.model.single_transformer_blocks.17.attn.to_q', 'base_model.model.single_transformer_blocks.1.attn.to_k', 'base_model.model.transformer_blocks.0.attn.to_q', 'base_model.model.single_transformer_blocks.26.proj_out', 'base_model.model.single_transformer_blocks.3.
attn.to_q', 'base_model.model.transformer_blocks.8.attn.to_q', 'base_model.model.transformer_blocks.12.attn.add_v_proj', 'base_model.model.single_transformer_blocks.8.attn.to_k', 'base_model.model.single_transformer_blocks.4.proj_mlp', 'base_model.model.single_transformer_blocks.6.attn.to_v', 'base_model.model.single_transformer_blocks.31.attn.to_
v', 'base_model.model.single_transformer_blocks.3.attn.to_k', 'base_model.model.transformer_blocks.1.attn.add_k_proj', 'base_model.model.transformer_blocks.18.attn.to_q', 'base_model.model.single_transformer_blocks.19.attn.to_v', 'base_model.model.single_transformer_blocks.28.proj_out', 'base_model.model.transformer_blocks.6.attn.add_k_proj', 'bas
e_model.model.single_transformer_blocks.32.attn.to_v', 'base_model.model.single_transformer_blocks.29.attn.to_k', 'base_model.model.single_transformer_blocks.7.attn.to_k', 'base_model.model.single_transformer_blocks.20.attn.to_v', 'base_model.model.single_transformer_blocks.13.attn.to_q', 'base_model.model.single_transformer_blocks.24.attn.to_k', 
'base_model.model.transformer_blocks.10.attn.to_k', 'base_model.model.transformer_blocks.3.attn.add_q_proj', 'base_model.model.single_transformer_blocks.36.attn.to_k', 'base_model.model.single_transformer_blocks.26.attn.to_q', 'base_model.model.transformer_blocks.18.attn.add_q_proj', 'base_model.model.transformer_blocks.3.attn.to_v', 'base_model.m
odel.single_transformer_blocks.2.attn.to_v', 'base_model.model.transformer_blocks.17.attn.add_k_proj', 'base_model.model.single_transformer_blocks.20.attn.to_q', 'base_model.model.transformer_blocks.16.attn.to_out.0', 'base_model.model.single_transformer_blocks.5.attn.to_v', 'base_model.model.single_transformer_blocks.33.proj_mlp', 'base_model.mod
el.single_transformer_blocks.14.attn.to_k', 'base_model.model.single_transformer_blocks.6.attn.to_k', 'base_model.model.single_transformer_blocks.35.attn.to_q', 'base_model.model.single_transformer_blocks.0.attn.to_q', 'base_model.model.single_transformer_blocks.30.proj_out', 'base_model.model.single_transformer_blocks.4.attn.to_k', 'base_model.mo
del.single_transformer_blocks.13.proj_out', 'base_model.model.single_transformer_blocks.9.attn.to_k', 'base_model.model.transformer_blocks.8.attn.to_k', 'base_model.model.single_transformer_blocks.17.proj_mlp', 'base_model.model.transformer_blocks.2.attn.to_v', 'base_model.model.transformer_blocks.15.attn.to_out.0', 'base_model.model.single_transf
ormer_blocks.10.attn.to_k', 'base_model.model.single_transformer_blocks.36.proj_mlp', 'base_model.model.transformer_blocks.4.attn.to_out.0', 'base_model.model.single_transformer_blocks.28.attn.to_k', 'base_model.model.transformer_blocks.14.attn.add_v_proj', 'base_model.model.transformer_blocks.17.attn.to_q', 'base_model.model.single_transformer_bl
ocks.36.proj_out', 'base_model.model.transformer_blocks.11.attn.to_q', 'base_model.model.transformer_blocks.16.attn.add_v_proj', 'base_model.model.transformer_blocks.0.attn.add_q_proj', 'base_model.model.single_transformer_blocks.7.proj_mlp', 'base_model.model.single_transformer_blocks.19.attn.to_k', 'base_model.model.transformer_blocks.14.attn.to
_q', 'base_model.model.transformer_blocks.17.attn.add_v_proj', 'base_model.model.transformer_blocks.11.attn.add_k_proj', 'base_model.model.single_transformer_blocks.5.proj_out', 'base_model.model.single_transformer_blocks.11.proj_mlp', 'base_model.model.transformer_blocks.0.attn.to_out.0', 'base_model.model.transformer_blocks.1.attn.add_q_proj', '
base_model.model.single_transformer_blocks.30.attn.to_q', 'base_model.model.transformer_blocks.8.attn.to_v', 'base_model.model.single_transformer_blocks.20.proj_mlp', 'base_model.model.single_transformer_blocks.8.attn.to_q', 'base_model.model.transformer_blocks.16.attn.add_k_proj', 'base_model.model.single_transformer_blocks.12.proj_out', 'base_mo
del.model.single_transformer_blocks.13.proj_mlp', 'base_model.model.transformer_blocks.1.attn.add_v_proj', 'base_model.model.transformer_blocks.9.attn.add_q_proj', 'base_model.model.transformer_blocks.2.attn.add_v_proj', 'base_model.model.single_transformer_blocks.22.proj_mlp', 'base_model.model.single_transformer_blocks.10.proj_out', 'base_model.
model.transformer_blocks.5.attn.add_q_proj', 'base_model.model.transformer_blocks.7.attn.to_out.0', 'base_model.model.single_transformer_blocks.9.attn.to_q', 'base_model.model.single_transformer_blocks.15.attn.to_k', 'base_model.model.single_transformer_blocks.9.attn.to_v', 'base_model.model.single_transformer_blocks.10.attn.to_q', 'base_model.mod
el.single_transformer_blocks.22.proj_out', 'base_model.model.single_transformer_blocks.24.proj_out', 'base_model.model.transformer_blocks.6.attn.to_out.0', 'base_model.model.transformer_blocks.12.attn.to_out.0', 'base_model.model.single_transformer_blocks.35.attn.to_v', 'base_model.model.single_transformer_blocks.34.proj_out', 'base_model.model.tr
ansformer_blocks.5.attn.to_k', 'base_model.model.transformer_blocks.9.attn.add_k_proj', 'base_model.model.transformer_blocks.1.attn.to_k', 'base_model.model.transformer_blocks.17.attn.to_v', 'base_model.model.single_transformer_blocks.2.proj_out', 'base_model.model.single_transformer_blocks.6.proj_out', 'base_model.model.single_transformer_blocks.
5.attn.to_q', 'base_model.model.transformer_blocks.12.attn.add_q_proj', 'base_model.model.single_transformer_blocks.19.attn.to_q', 'base_model.model.transformer_blocks.11.attn.add_v_proj', 'base_model.model.single_transformer_blocks.1.proj_out', 'base_model.model.transformer_blocks.4.attn.add_v_proj', 'base_model.model.single_transformer_blocks.17
.attn.to_v', 'base_model.model.single_transformer_blocks.21.attn.to_q', 'base_model.model.single_transformer_blocks.12.attn.to_v', 'base_model.model.transformer_blocks.2.attn.to_k', 'base_model.model.single_transformer_blocks.34.attn.to_k', 'base_model.model.single_transformer_blocks.16.attn.to_q', 'base_model.model.single_transformer_blocks.0.att
n.to_k', 'base_model.model.single_transformer_blocks.32.proj_out', 'base_model.model.transformer_blocks.16.attn.to_k', 'base_model.model.single_transformer_blocks.10.attn.to_v', 'base_model.model.single_transformer_blocks.23.attn.to_v', 'base_model.model.single_transformer_blocks.2.proj_mlp', 'base_model.model.transformer_blocks.16.attn.to_v', 'ba
se_model.model.transformer_blocks.17.attn.to_out.0', 'base_model.model.single_transformer_blocks.36.attn.to_q', 'base_model.model.transformer_blocks.15.attn.add_v_proj', 'base_model.model.single_transformer_blocks.18.attn.to_q', 'base_model.model.transformer_blocks.7.attn.to_v', 'base_model.model.proj_out', 'base_model.model.single_transformer_blo
cks.14.proj_mlp', 'base_model.model.single_transformer_blocks.12.attn.to_k', 'base_model.model.single_transformer_blocks.9.proj_mlp', 'base_model.model.single_transformer_blocks.30.attn.to_k', 'base_model.model.single_transformer_blocks.3.proj_out', 'base_model.model.transformer_blocks.14.attn.to_v', 'base_model.model.single_transformer_blocks.18.
attn.to_v', 'base_model.model.single_transformer_blocks.21.attn.to_v', 'base_model.model.transformer_blocks.15.attn.to_v', 'base_model.model.single_transformer_blocks.2.attn.to_k', 'base_model.model.single_transformer_blocks.11.attn.to_k', 'base_model.model.transformer_blocks.17.attn.to_k', 'base_model.model.single_transformer_blocks.29.attn.to_q'
, 'base_model.model.single_transformer_blocks.11.attn.to_q', 'base_model.model.transformer_blocks.16.attn.add_q_proj', 'base_model.model.single_transformer_blocks.16.attn.to_k', 'base_model.model.single_transformer_blocks.8.proj_mlp', 'base_model.model.single_transformer_blocks.24.proj_mlp', 'base_model.model.single_transformer_blocks.27.proj_out'
, 'base_model.model.transformer_blocks.13.attn.to_k', 'base_model.model.single_transformer_blocks.33.attn.to_k', 'base_model.model.single_transformer_blocks.32.attn.to_k', 'base_model.model.transformer_blocks.8.attn.add_q_proj', 'base_model.model.transformer_blocks.18.attn.to_out.0', 'base_model.model.single_transformer_blocks.15.attn.to_v', 'base
_model.model.single_transformer_blocks.14.attn.to_q', 'base_model.model.single_transformer_blocks.21.proj_out', 'base_model.model.transformer_blocks.1.attn.to_v', 'base_model.model.transformer_blocks.10.attn.to_v', 'base_model.model.single_transformer_blocks.29.proj_mlp', 'base_model.model.single_transformer_blocks.1.attn.to_v', 'base_model.model.
transformer_blocks.15.attn.to_q', 'base_model.model.single_transformer_blocks.25.proj_out', 'base_model.model.transformer_blocks.0.attn.add_v_proj', 'base_model.model.single_transformer_blocks.26.attn.to_k', 'base_model.model.transformer_blocks.18.attn.add_k_proj', 'base_model.model.single_transformer_blocks.37.attn.to_q', 'base_model.model.single
_transformer_blocks.27.proj_mlp', 'base_model.model.transformer_blocks.18.attn.to_v', 'base_model.model.transformer_blocks.6.attn.add_q_proj', 'base_model.model.transformer_blocks.14.attn.add_k_proj', 'base_model.model.single_transformer_blocks.27.attn.to_q', 'base_model.model.single_transformer_blocks.15.proj_mlp', 'base_model.model.transformer_b
locks.10.attn.to_out.0', 'base_model.model.single_transformer_blocks.29.attn.to_v', 'base_model.model.single_transformer_blocks.37.attn.to_v', 'base_model.model.transformer_blocks.5.attn.to_v', 'base_model.model.transformer_blocks.15.attn.add_k_proj', 'base_model.model.transformer_blocks.3.attn.to_k', 'base_model.model.transformer_blocks.18.attn.a
dd_v_proj', 'base_model.model.transformer_blocks.2.attn.to_q', 'base_model.model.single_transformer_blocks.37.proj_out', 'base_model.model.transformer_blocks.16.attn.to_q', 'base_model.model.transformer_blocks.13.attn.add_v_proj', 'base_model.model.single_transformer_blocks.16.attn.to_v', 'base_model.model.single_transformer_blocks.7.attn.to_q', '
base_model.model.transformer_blocks.1.attn.to_out.0', 'base_model.model.transformer_blocks.10.attn.add_k_proj', 'base_model.model.transformer_blocks.0.attn.add_k_proj', 'base_model.model.single_transformer_blocks.7.attn.to_v', 'base_model.model.single_transformer_blocks.11.attn.to_v', 'base_model.model.single_transformer_blocks.18.attn.to_k', 'bas
e_model.model.transformer_blocks.5.attn.to_q', 'base_model.model.single_transformer_blocks.24.attn.to_v', 'base_model.model.transformer_blocks.4.attn.to_v', 'base_model.model.single_transformer_blocks.17.attn.to_k', 'base_model.model.single_transformer_blocks.22.attn.to_q', 'base_model.model.single_transformer_blocks.16.proj_out', 'base_model.mode
l.single_transformer_blocks.22.attn.to_v', 'base_model.model.transformer_blocks.5.attn.to_out.0', 'base_model.model.single_transformer_blocks.28.proj_mlp', 'base_model.model.single_transformer_blocks.37.attn.to_k', 'base_model.model.single_transformer_blocks.3.proj_mlp', 'base_model.model.transformer_blocks.0.attn.to_k', 'base_model.model.single_t
ransformer_blocks.6.proj_mlp', 'base_model.model.single_transformer_blocks.9.proj_out', 'base_model.model.single_transformer_blocks.32.attn.to_q', 'base_model.model.single_transformer_blocks.8.proj_out', 'base_model.model.single_transformer_blocks.31.proj_mlp', 'base_model.model.transformer_blocks.6.attn.to_k', 'base_model.model.transformer_blocks
.14.attn.to_out.0', 'base_model.model.transformer_blocks.10.attn.add_v_proj', 'base_model.model.transformer_blocks.8.attn.add_v_proj', 'base_model.model.transformer_blocks.3.attn.to_out.0', 'base_model.model.single_transformer_blocks.31.proj_out', 'base_model.model.single_transformer_blocks.11.proj_out', 'base_model.model.single_transformer_blocks
.25.proj_mlp', 'base_model.model.transformer_blocks.12.attn.to_v', 'base_model.model.single_transformer_blocks.27.attn.to_v', 'base_model.model.single_transformer_blocks.21.proj_mlp', 'base_model.model.single_transformer_blocks.35.proj_mlp', 'base_model.model.single_transformer_blocks.36.attn.to_v', 'base_model.model.single_transformer_blocks.15.a
ttn.to_q', 'base_model.model.single_transformer_blocks.31.attn.to_q', 'base_model.model.transformer_blocks.13.attn.add_k_proj', 'base_model.model.single_transformer_blocks.27.attn.to_k', 'base_model.model.transformer_blocks.2.attn.add_k_proj', 'base_model.model.single_transformer_blocks.13.attn.to_k', 'base_model.model.single_transformer_blocks.16
.proj_mlp', 'base_model.model.single_transformer_blocks.26.proj_mlp', 'base_model.model.single_transformer_blocks.18.proj_mlp', 'base_model.model.transformer_blocks.13.attn.to_q', 'base_model.model.transformer_blocks.14.attn.add_q_proj', 'base_model.model.single_transformer_blocks.23.attn.to_q', 'base_model.model.single_transformer_blocks.33.proj_
out', 'base_model.model.single_transformer_blocks.34.proj_mlp', 'base_model.model.transformer_blocks.10.attn.to_q', 'base_model.model.transformer_blocks.8.attn.to_out.0', 'base_model.model.transformer_blocks.2.attn.to_out.0', 'base_model.model.transformer_blocks.9.attn.to_v', 'base_model.model.transformer_blocks.13.attn.to_out.0', 'base_model.mode
l.single_transformer_blocks.23.proj_mlp', 'base_model.model.transformer_blocks.7.attn.to_k', 'base_model.model.transformer_blocks.13.attn.to_v', 'base_model.model.single_transformer_blocks.14.attn.to_v', 'base_model.model.transformer_blocks.8.attn.add_k_proj', 'base_model.model.single_transformer_blocks.3.attn.to_v', 'base_model.model.transformer_
blocks.3.attn.to_q', 'base_model.model.single_transformer_blocks.34.attn.to_v', 'base_model.model.single_transformer_blocks.25.attn.to_v', 'base_model.model.single_transformer_blocks.32.proj_mlp', 'base_model.model.transformer_blocks.12.attn.to_q', 'base_model.model.transformer_blocks.9.attn.add_v_proj', 'base_model.model.single_transformer_blocks
.19.proj_out', 'base_model.model.transformer_blocks.10.attn.add_q_proj', 'base_model.model.single_transformer_blocks.33.attn.to_v', 'base_model.model.single_transformer_blocks.20.attn.to_k', 'base_model.model.single_transformer_blocks.4.proj_out', 'base_model.model.transformer_blocks.7.attn.add_q_proj', 'base_model.model.transformer_blocks.4.attn.
to_k', 'base_model.model.transformer_blocks.18.attn.to_k', 'base_model.model.transformer_blocks.6.attn.add_v_proj', 'base_model.model.transformer_blocks.9.attn.to_k', 'base_model.model.transformer_blocks.13.attn.add_q_proj', 'base_model.model.single_transformer_blocks.19.proj_mlp', 'base_model.model.single_transformer_blocks.10.proj_mlp', 'base_mo
del.model.single_transformer_blocks.18.proj_out', 'base_model.model.single_transformer_blocks.29.proj_out', 'base_model.model.single_transformer_blocks.13.attn.to_v', 'base_model.model.single_transformer_blocks.24.attn.to_q', 'base_model.model.transformer_blocks.6.attn.to_v', 'base_model.model.transformer_blocks.17.attn.add_q_proj', 'base_model.mo
del.transformer_blocks.12.attn.to_k', 'base_model.model.transformer_blocks.1.attn.to_q', 'base_model.model.single_transformer_blocks.37.proj_mlp', 'base_model.model.single_transformer_blocks.4.attn.to_q', 'base_model.model.single_transformer_blocks.23.attn.to_k', 'base_model.model.transformer_blocks.11.attn.add_q_proj', 'base_model.model.single_tr
ansformer_blocks.14.proj_out', 'base_model.model.single_transformer_blocks.30.proj_mlp', 'base_model.model.single_transformer_blocks.0.proj_out', 'base_model.model.single_transformer_blocks.5.attn.to_k', 'base_model.model.single_transformer_blocks.1.proj_mlp', 'base_model.model.transformer_blocks.0.attn.to_v', 'base_model.model.single_transformer_blocks.22.attn.to_k', 'base_model.model.single_transformer_blocks.6.attn.to_q', 'base_model.model.single_transformer_blocks.33.attn.to_q', 'base_model.model.single_transformer_blocks.12.attn.to_q', 'base_model.model.single_transformer_blocks.28.attn.to_q'} not found in the base model. Please check the target modules and try again.

[markrmiller]

Nothing loads in comfyui anymore either. Though it doesn't hard fail, it can't apply any of the weights.

I have a Lora from like the night before last that works perfect in comfy. Quantized. I know because every none quantized lora I've trained has been absolute garbage. Finally I did a quantized one and it worked. That was the last time. Now every Lora I train can't be loaded with anything.

[markrmiller]

I worked it out. I had to roll back to a commit from a couple days ago before something broke the quantized Lora’s.

[bghira]

i think i gathered what happened. i was trying to resolve the quanto load/save issue and applied a recommendation from the peft team to use get_peft_model but it seems to totally change how the keys work and requires me to then manually clean them up. not going to do that, but i did revert to the old way. if you want to venture to try again. sorry for the commotion, that resume bug really is a problem.

[bghira]

reports on discord that it works now