Extracting Lora/Lyco fail

[killerciao]

loading original SDXL model
building U-Net
no_half = False
loading U-Net...
U-Net: None
building text encoders
loading text encoders...
text encoder 1:
text encoder 2:
create LoRA network. base dim (rank): 64, alpha: 64
neuron dropout: p=None, rank dropout: p=None, module dropout: p=None
create LoRA for Text Encoder 1:
create LoRA for Text Encoder 2:
create LoRA for Text Encoder: 264 modules.
create LoRA for U-Net: 722 modules.
create LoRA network. base dim (rank): 64, alpha: 64
neuron dropout: p=None, rank dropout: p=None, module dropout: p=None
create LoRA for Text Encoder 1:
create LoRA for Text Encoder 2:
create LoRA for Text Encoder: 264 modules.
create LoRA for U-Net: 722 modules.
Calculate svd: 0%| | 0/986 [00:00<?, ?it/s]Text encoder is different. 0.0024471282958984375 > 0.0001ttn_k_proj
264/264 100%: lora_te2_text_model_encoder_layers_31_mlp_fc2
0/722 0%: lora_unet_down_blocks_1_attentions_0_proj_inin
Traceback (most recent call last):
File "H:\IA\Packages\Stable Diffusion WebUI Forge\venv\lib\site-packages\gradio\routes.py", line 488, in run_predict
output = await app.get_blocks().process_api(
File "H:\IA\Packages\Stable Diffusion WebUI Forge\venv\lib\site-packages\gradio\blocks.py", line 1431, in process_api
result = await self.call_function(
File "H:\IA\Packages\Stable Diffusion WebUI Forge\venv\lib\site-packages\gradio\blocks.py", line 1103, in call_function
prediction = await anyio.to_thread.run_sync(
File "H:\IA\Packages\Stable Diffusion WebUI Forge\venv\lib\site-packages\anyio\to_thread.py", line 33, in run_sync
return await get_asynclib().run_sync_in_worker_thread(
File "H:\IA\Packages\Stable Diffusion WebUI Forge\venv\lib\site-packages\anyio_backends_asyncio.py", line 877, in run_sync_in_worker_thread
return await future
File "H:\IA\Packages\Stable Diffusion WebUI Forge\venv\lib\site-packages\anyio_backends_asyncio.py", line 807, in run
result = context.run(func, *args)
File "H:\IA\Packages\Stable Diffusion WebUI Forge\venv\lib\site-packages\gradio\utils.py", line 707, in wrapper
response = f(*args, **kwargs)
File "H:\IA\Packages\Stable Diffusion WebUI Forge\extensions\sd-webui-model-mixer\scripts\model_mixer.py", line 4730, in extract_lora_from_current_model
extracted_lora = svd(dict(state_dict_base), dict(state_dict_trained), None, lora_dim, min_diff=min_diff, clamp_quantile=clamp_quantile, device=calc_device,
File "H:\IA\Packages\Stable Diffusion WebUI Forge\extensions\sd-webui-model-mixer\scripts\kohya\extract_lora_from_models.py", line 186, in svd
if torch.allclose(module_t.weight, module_o.weight):
RuntimeError: BFloat16 did not match Half

I tried every combination of settings but alwasys stop at 27%