Expected `device_type` of type `str`, got: `<class 'torch.device'>`

[langbowang]

ComfyUI Error Report

Error Details

• Node Type: Joy_caption
• Exception Type: ValueError

• Exception Message: Expected device_type of type str, got: <class 'torch.device'>

  Stack Trace

  
  File "C:\Users\lbw\AIGC\ComfyUI_FluxFast\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 "C:\Users\lbw\AIGC\ComfyUI_FluxFast\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 "C:\Users\lbw\AIGC\ComfyUI_FluxFast\ComfyUI\execution.py", line 169, in _map_node_over_list
  process_inputs(input_dict, i)
  
  File "C:\Users\lbw\AIGC\ComfyUI_FluxFast\ComfyUI\execution.py", line 158, in process_inputs
  results.append(getattr(obj, func)(**inputs))
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  
  File "C:\Users\lbw\AIGC\ComfyUI_FluxFast\ComfyUI\custom_nodes\ComfyUI_HF_Servelress_Inference\nodes\Joy_Caption.py", line 171, in gen
  with torch.amp.autocast_mode.autocast(DEVICE, enabled=True):
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  
  File "C:\Users\lbw\AIGC\ComfyUI_FluxFast\python_embeded\Lib\site-packages\torch\amp\autocast_mode.py", line 225, in __init__
  raise ValueError(

[langbowang]

[mirek190]

Joy_caption
Expected `device_type` of type `str`, got: `<class 'torch.device'>`

The same error

If you switch off Joy_caption and leave Florence 2 and MiniCPM v2_6 then works. Seems error causing only Joy_caption node.

[qtnrtnd]

yep, same problem here

[train147369]

同样错误，双gpu

[qtnrtnd]

It worked for me when integrating the changes of #23 in ComfyUI\custom_nodes\ComfyUI_HF_Servelress_Inference\nodes\Joy_Caption.py

[train147369]

D:\AIProject\ComfyUI\custom_nodes\ComfyUI_HF_Servelress_Inference\nodes\ 找到这一行： pythonCopywith torch.amp.autocast_mode.autocast(DEVICE, enabled=True):

将其更改为： pythonCopywith torch.amp.autocast_mode.autocast(DEVICE.type if isinstance(DEVICE, torch.device) else DEVICE, enabled=True):

[mirek190]

Joy_Caption.py

Works thank you!

Just small update:

IF Joy_caption has error:
"Expected device_type of type str, got: <class 'torch.device'>"
Under 
ComfyUI\custom_nodes\ComfyUI_HF_Servelress_Inference\nodes\Joy_Caption.py
replace
"with torch.amp.autocast_mode.autocast(DEVICE, enabled=True):"
to
"with torch.amp.autocast_mode.autocast(DEVICE.type if isinstance(DEVICE, torch.device) else DEVICE, enabled=True):"