extract LoRA from checkpoints in colab

[hako-mikan]

Discuss this issue here because it is reported a lot.
The diffusers version and memory are the two most likely causes so far; setting the diffusers version to 0.10.2 or later may solve the problem. The memory problem may be solved by reducing the model size used.

[gilroff]

what im getting is memory error im sure. i gets red in memory graph and then crashes.

[hako-mikan]

You might try the "unload model" button. This button allows you to unload the model being loaded for image generation to increase the amount of memory available.

[gilroff]

Ok I will try that

[Omenizer]

I have paid Colab and no memory issues but still no luck. As of recently SM doesn't load in Colab at all - Python issue?

[hako-mikan]

If this script is not on web-ui, there are something errors in command prompt.

[Omenizer]

Extension is loading again in colab now

[zhuliyi0]

I am getting "duplicate lora name" error in multiple stable diffusion environments, in windows and mac and colab, all of them newly installed. I tried to lower diffuser version to 0.10.2 but not helping. Same no matter what model I use, even the same model for A and B. Error meesage below. I saw similar reported issue with that "320 modules" for U-Net.

make LoRA start
loading u-net:
loading vae:
loading text encoder:
create LoRA network. base dim (rank): 128, alpha: 128
create LoRA for Text Encoder: 72 modules.
create LoRA for U-Net: 320 modules.
Traceback (most recent call last):
File "D:\StableDefusion\2_Src\stable-diffusion-webui\venv\lib\site-packages\gradio\routes.py", line 394, in run_predict
output = await app.get_blocks().process_api(
File "D:\StableDefusion\2_Src\stable-diffusion-webui\venv\lib\site-packages\gradio\blocks.py", line 1075, in process_api
result = await self.call_function(
File "D:\StableDefusion\2_Src\stable-diffusion-webui\venv\lib\site-packages\gradio\blocks.py", line 884, in call_function
prediction = await anyio.to_thread.run_sync(
File "D:\StableDefusion\2_Src\stable-diffusion-webui\venv\lib\site-packages\anyio\to_thread.py", line 31, in run_sync
return await get_asynclib().run_sync_in_worker_thread(
File "D:\StableDefusion\2_Src\stable-diffusion-webui\venv\lib\site-packages\anyio_backends_asyncio.py", line 937, in run_sync_in_worker_thread
return await future
File "D:\StableDefusion\2_Src\stable-diffusion-webui\venv\lib\site-packages\anyio_backends_asyncio.py", line 867, in run
result = context.run(func, *args)
File "D:\StableDefusion\2_Src\stable-diffusion-webui\extensions\sd-webui-supermerger\scripts\mergers\pluslora.py", line 150, in makelora
svd(fullpathfromname(model_a),fullpathfromname(model_b),False,dim,precision,saveto,alpha,beta)
File "D:\StableDefusion\2_Src\stable-diffusion-webui\extensions\sd-webui-supermerger\scripts\mergers\pluslora.py", line 438, in svd
lora_network_o = create_network(1.0, dim, dim, None, text_encoder_o, unet_o)
File "D:\StableDefusion\2_Src\stable-diffusion-webui\extensions\sd-webui-supermerger\scripts\mergers\pluslora.py", line 986, in create_network
network = LoRANetwork(
File "D:\StableDefusion\2_Src\stable-diffusion-webui\extensions\sd-webui-supermerger\scripts\mergers\pluslora.py", line 1089, in init
assert lora.lora_name not in names, f"duplicated lora name: {lora.lora_name}"
AssertionError: duplicated lora name: lora_unet_down_blocks_0_attentions_0_transformer_blocks_0_attn1_to_q

[zhuliyi0]

Found this issue solved on KohyaGUI repo by simply downgrade diffuser. I must have missed some other packages in the environment? Will try a clean install later.

[siluoke]

for error:

Traceback (most recent call last):
File "/mnt/workspace/v3/lib/python3.10/site-packages/gradio/routes.py", line 337, in run_predict
output = await app.get_blocks().process_api(
File "/mnt/workspace/v3/lib/python3.10/site-packages/gradio/blocks.py", line 1015, in process_api
result = await self.call_function(
File "/mnt/workspace/v3/lib/python3.10/site-packages/gradio/blocks.py", line 833, in call_function
prediction = await anyio.to_thread.run_sync(
File "/mnt/workspace/v3/lib/python3.10/site-packages/anyio/to_thread.py", line 31, in run_sync
return await get_asynclib().run_sync_in_worker_thread(
File "/mnt/workspace/v3/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 937, in run_sync_in_worker_thread
return await future
File "/mnt/workspace/v3/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 867, in run
result = context.run(func, *args)
File "/mnt/workspace/stable-diffusion-webui/extensions/sd-webui-supermerger/scripts/mergers/pluslora.py", line 150, in makelora
svd(fullpathfromname(model_a),fullpathfromname(model_b),False,dim,precision,saveto,alpha,beta)
File "/mnt/workspace/stable-diffusion-webui/extensions/sd-webui-supermerger/scripts/mergers/pluslora.py", line 438, in svd
lora_network_o = create_network(1.0, dim, dim, None, text_encoder_o, unet_o)
File "/mnt/workspace/stable-diffusion-webui/extensions/sd-webui-supermerger/scripts/mergers/pluslora.py", line 986, in create_network
network = LoRANetwork(
File "/mnt/workspace/stable-diffusion-webui/extensions/sd-webui-supermerger/scripts/mergers/pluslora.py", line 1089, in init
assert lora.lora_name not in names, f"duplicated lora name: {lora.lora_name}"
AssertionError: duplicated lora name: lora_unet_down_blocks_0_attentions_0_transformer_blocks_0_attn1_to_q

try : pip install diffusers==0.14.0