[Bug] UnboundLocalError: local variable 'output_shape' referenced before assignment

[acorderob]

I get sometimes this error, trying to generate an image (even with no lycoris loras).

Commit: 8e0ebd7

Traceback (most recent call last):
File "D:\Other Applications\stable-diffusion-webui\modules\call_queue.py", line 56, in f
res = list(func(*args, **kwargs))
File "D:\Other Applications\stable-diffusion-webui\modules\call_queue.py", line 37, in f
res = func(*args, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\modules\txt2img.py", line 56, in txt2img
processed = process_images(p)
File "D:\Other Applications\stable-diffusion-webui\modules\processing.py", line 503, in process_images
res = process_images_inner(p)
File "D:\Other Applications\stable-diffusion-webui\modules\processing.py", line 642, in process_images_inner
uc = get_conds_with_caching(prompt_parser.get_learned_conditioning, negative_prompts, p.steps, cached_uc)
File "D:\Other Applications\stable-diffusion-webui\modules\processing.py", line 587, in get_conds_with_caching
cache[1] = function(shared.sd_model, required_prompts, steps)
File "D:\Other Applications\stable-diffusion-webui\modules\prompt_parser.py", line 140, in get_learned_conditioning
conds = model.get_learned_conditioning(texts)
File "D:\Other Applications\stable-diffusion-webui\repositories\stable-diffusion-stability-ai\ldm\models\diffusion\ddpm.py", line 669, in get_learned_conditioning
c = self.cond_stage_model(c)
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\modules\sd_hijack_clip.py", line 229, in forward
z = self.process_tokens(tokens, multipliers)
File "D:\Other Applications\stable-diffusion-webui\modules\sd_hijack_clip.py", line 254, in process_tokens
z = self.encode_with_transformers(tokens)
File "D:\Other Applications\stable-diffusion-webui\modules\sd_hijack_clip.py", line 302, in encode_with_transformers
outputs = self.wrapped.transformer(input_ids=tokens, output_hidden_states=-opts.CLIP_stop_at_last_layers)
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 811, in forward
return self.text_model(
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 721, in forward
encoder_outputs = self.encoder(
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 650, in forward
layer_outputs = encoder_layer(
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 379, in forward
hidden_states, attn_weights = self.self_attn(
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 268, in forward
query_states = self.q_proj(hidden_states) * self.scale
File "D:\Other Applications\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "D:\Other Applications\stable-diffusion-webui\extensions\a1111-sd-webui-locon\scripts......\extensions-builtin/Lora\lora.py", line 305, in lora_Linear_forward
lora_apply_weights(self)
File "D:\Other Applications\stable-diffusion-webui\extensions\a1111-sd-webui-locon\scripts......\extensions-builtin/Lora\lora.py", line 273, in lora_apply_weights
self.weight += lora_calc_updown(lora, module, self.weight)
File "D:\Other Applications\stable-diffusion-webui\extensions\a1111-sd-webui-locon\scripts\main.py", line 576, in lora_calc_updown
updown = rebuild_weight(module, target)
File "D:\Other Applications\stable-diffusion-webui\extensions\a1111-sd-webui-locon\scripts\main.py", line 565, in rebuild_weight
if len(output_shape) == 4:
UnboundLocalError: local variable 'output_shape' referenced before assignment

[acorderob]

Trying to debug the problem. The check for the module class in that function is failing, so the output_shape is never assigned.

if isinstance(module, LoraUpDownModule): fails, but the variable has an object of that class!

I don't know enough python to know what could be happening.

[KohakuBlueleaf]

@acorderob I have checked and this can only happend when you have other lora extension that will modify the loaded lora or modify the load lora function.

I have added a check for this, but the only solution is you may need to disable other lora function.

[acorderob]

The only other extension that seems to do something to loras would be "stable-diffusion-webui-composable-lora", but i have it disabled. I'll try removing it, just in case.

[KohakuBlueleaf]

@acorderob It looks like it is some bug in webui (but I don't know how it happend)
I have change the way to check the type
It should work now