Error when running ApplyFluxIPadapter node

#16
by SlavaOn - opened

linear-error.JPG

I searched for solution online by using key words, but got no hits. Am I the only one getting this error?

Error occurred when executing ApplyFluxIPAdapter:

'Linear' object has no attribute 'out_features'

File "E:\Projects\ComfyUI\ComfyUI\execution.py", line 317, in execute
output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Projects\ComfyUI\ComfyUI\execution.py", line 192, in get_output_data
return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Projects\ComfyUI\ComfyUI\execution.py", line 169, in _map_node_over_list
process_inputs(input_dict, i)
File "E:\Projects\ComfyUI\ComfyUI\execution.py", line 158, in process_inputs
results.append(getattr(obj, func)(**inputs))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Projects\ComfyUI\ComfyUI\custom_nodes\x-flux-comfyui\nodes.py", line 520, in applymodel
patches=FluxUpdateModules(tyanochky, pbar)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Projects\ComfyUI\ComfyUI\custom_nodes\x-flux-comfyui\utils.py", line 81, in FluxUpdateModules
patches[f"double_blocks.{i}"]=CopyDSB(flux_model.diffusion_model.double_blocks[i])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Projects\ComfyUI\ComfyUI\custom_nodes\x-flux-comfyui\utils.py", line 19, in CopyDSB
mlp_hidden_dim = tyan.img_mlp[0].out_features
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Projects\miniconda3\envs\comfyenv\Lib\site-packages\torch\nn\modules\module.py", line 1729, in getattr
raise AttributeError(f"'{type(self).name}' object has no attribute '{name}'")

BTW: I used the workflow offered on the instructions page: https://github.com/XLabs-AI/x-flux-comfyui

XLabs AI org

Did you use gguf model?

Yes, I was.

Today, I ran the same prompt but got a different error. I have not done anything different - only the computer was powered off and on...

Error occurred when executing UnetLoaderGGUF:
newbyteorder was removed from the ndarray class in NumPy 2.0. Use arr.view(arr.dtype.newbyteorder(order)) instead.

Actually, I applied updates available in ComfyUI Manager and the new problem with loader unet surfaced. It affected even the workflows that were running before.
See the error

error.JPG

I'm getting the same error when using an BNB-NF4 checkpoint + LoRa

But using the basic flux-dev-fp8 Diffusion Model worked correctly

Error occurred when executing LoadFluxIPAdapter:

'NoneType' object has no attribute 'model'

File "K:\StabilityMatrix-win-x64\Data\Packages\ComfyUI2\execution.py", line 317, in execute
output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
File "K:\StabilityMatrix-win-x64\Data\Packages\ComfyUI2\execution.py", line 192, in get_output_data
return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
File "K:\StabilityMatrix-win-x64\Data\Packages\ComfyUI2\execution.py", line 169, in _map_node_over_list
process_inputs(input_dict, i)
File "K:\StabilityMatrix-win-x64\Data\Packages\ComfyUI2\execution.py", line 158, in process_inputs
results.append(getattr(obj, func)(**inputs))
File "K:\StabilityMatrix-win-x64\Data\Packages\ComfyUI2\custom_nodes\x-flux-comfyui\nodes.py", line 433, in loadmodel
clip = load_clip_vision(path_clip).model

XLabs AI org

GGUF models do not supported yet

Thank you for pointing that out. Once I switched to the fp8 version everything worked fine.

XLabs AI org

hi, we added GGUF support, please update the https://github.com/XLabs-AI/x-flux-comfyui and try

Sign up or log in to comment