Lora 加载报错

[tolozine]

ComfyUI Error Report

Error Details

• Node Type: QuickMfluxNode
• Exception Type: RuntimeError
• Exception Message: [load_safetensors] Failed to open file /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/models/loras/Flux/Hyper-FLUX.1-dev-8steps-lora.safetensors

Stack Trace

  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 323, 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 "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 198, 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 "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 169, in _map_node_over_list
    process_inputs(input_dict, i)

  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 158, in process_inputs
    results.append(getattr(obj, func)(**inputs))
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/Mflux-ComfyUI/Mflux_Air.py", line 194, in generate
    flux = load_or_create_flux(model, quantize, Local_model if Local_model else None, lora_paths, lora_scales, ControlNet is not None)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/Mflux-ComfyUI/Mflux_Air.py", line 24, in load_or_create_flux
    flux_cache[key] = FluxClass(
                      ^^^^^^^^^^

  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/flux/flux.py", line 52, in __init__
    weights = WeightHandler(
              ^^^^^^^^^^^^^^

  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/weight_handler.py", line 28, in __init__
    LoraUtil.apply_loras(self.transformer, lora_paths, lora_scales)

  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/lora_util.py", line 18, in apply_loras
    LoraUtil._apply_lora(transformer, lora_file, lora_scale)

  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/lora_util.py", line 36, in _apply_lora
    lora_transformer, _ = WeightHandler.load_transformer(lora_path=lora_file)
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/weight_handler.py", line 64, in load_transformer
    weights, quantization_level = WeightHandler._get_weights("transformer", root_path, lora_path)
                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/weight_handler.py", line 123, in _get_weights
    weight = list(mx.load(lora_path).items())
                  ^^^^^^^^^^^^^^^^^^

System Information

• ComfyUI Version: unknown
• Arguments: main.py --fp16-vae --lowvram
• OS: posix
• Python Version: 3.12.1 | packaged by Anaconda, Inc. | (main, Jan 19 2024, 09:45:58) [Clang 14.0.6 ]
• Embedded Python: false
• PyTorch Version: 2.3.1

Devices

• Name: mps
  • Type: mps
  • VRAM Total: 34359738368
  • VRAM Free: 17109286912
  • Torch VRAM Total: 34359738368
  • Torch VRAM Free: 17109286912

Logs

2024-11-02 09:59:26,655 - root - INFO - Total VRAM 32768 MB, total RAM 32768 MB
2024-11-02 09:59:26,655 - root - INFO - pytorch version: 2.3.1
2024-11-02 09:59:26,655 - root - INFO - Set vram state to: SHARED
2024-11-02 09:59:26,655 - root - INFO - Device: mps
2024-11-02 09:59:27,243 - root - INFO - Using sub quadratic optimization for cross attention, if you have memory or speed issues try using: --use-split-cross-attention
2024-11-02 09:59:28,031 - root - INFO - [Prompt Server] web root: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/web
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path checkpoints /Volumes/WD_Element/comfyui/models/checkpoints/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path configs /Volumes/WD_Element/comfyui/models/configs/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path clip /Volumes/WD_Element/comfyui/models/clip/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path clip_vision /Volumes/WD_Element/comfyui/models/clip_vision/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path vae /Volumes/WD_Element/comfyui/models/vae/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path Mflux /Volumes/WD_Element/comfyui/models/Mflux/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path loras /Volumes/WD_Element/comfyui/models/Loras/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path loras /Volumes/WD_Element/comfyui/models/LyCORIS/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path ESRGAN /Volumes/WD_Element/comfyui/models/ESRGAN/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path RealESRGAN /Volumes/WD_Element/comfyui/models/RealESRGAN/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path SwinIR /Volumes/WD_Element/comfyui/models/SwinIR/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path upscale_models /Volumes/WD_Element/comfyui/models/upscale_models/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path embeddings /Volumes/WD_Element/comfyui/models/embeddings/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path hypernetworks /Volumes/WD_Element/comfyui/models/hypernetworks/
2024-11-02 09:59:28,032 - root - INFO - Adding extra search path controlnet /Volumes/WD_Element/comfyui/models/ControlNet/
2024-11-02 09:59:28,033 - root - INFO - Adding extra search path controlnet /Volumes/WD_Element/comfyui/models/xlabs/controlnets/
2024-11-02 09:59:28,033 - root - INFO - Adding extra search path photomaker /Volumes/WD_Element/comfyui/models/photomaker/
2024-11-02 09:59:28,033 - root - INFO - Adding extra search path inpaint /Volumes/WD_Element/comfyui/models/inpaint/
2024-11-02 09:59:28,033 - root - INFO - Adding extra search path ipadapter /Volumes/WD_Element/comfyui/models/ipadapter/
2024-11-02 09:59:28,033 - root - INFO - Adding extra search path unet /Volumes/WD_Element/comfyui/models/unet/
2024-11-02 09:59:29,439 - root - INFO - Total VRAM 32768 MB, total RAM 32768 MB
2024-11-02 09:59:29,439 - root - INFO - pytorch version: 2.3.1
2024-11-02 09:59:29,439 - root - INFO - Set vram state to: SHARED
2024-11-02 09:59:29,439 - root - INFO - Device: mps
2024-11-02 09:59:31,542 - root - INFO - 
Import times for custom nodes:
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/websocket_image_save.py
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/Cup-ClipBoard
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/cg-use-everywhere
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI_IPAdapter_plus
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI-CUP
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/comfyui-inpaint-nodes
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI_UltimateSDUpscale
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/comfyui-tooling-nodes
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI-KJNodes
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI_bitsandbytes_NF4
2024-11-02 09:59:31,542 - root - INFO -    0.0 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI-Manager
2024-11-02 09:59:31,542 - root - INFO -    0.1 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/Mflux-ComfyUI
2024-11-02 09:59:31,542 - root - INFO -    0.2 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/comfyui_segment_anything_plus
2024-11-02 09:59:31,542 - root - INFO -    0.2 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/comfyui_controlnet_aux
2024-11-02 09:59:31,542 - root - INFO -    0.2 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/virtuoso-nodes
2024-11-02 09:59:31,542 - root - INFO -    0.3 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI-Image-Filters
2024-11-02 09:59:31,542 - root - INFO -    0.6 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUI-Easy-Use
2024-11-02 09:59:31,542 - root - INFO -    0.7 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/ComfyUi-Ollama-YN
2024-11-02 09:59:31,542 - root - INFO -    0.8 seconds: /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/was-node-suite-comfyui
2024-11-02 09:59:31,542 - root - INFO - 
2024-11-02 09:59:31,546 - root - INFO - Starting server

2024-11-02 09:59:31,546 - root - INFO - To see the GUI go to: http://127.0.0.1:8188
2024-11-02 09:59:42,910 - root - INFO - got prompt
2024-11-02 09:59:43,144 - root - ERROR - !!! Exception during processing !!! [load_safetensors] Failed to open file /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/models/loras/Flux/Hyper-FLUX.1-dev-8steps-lora.safetensors
2024-11-02 09:59:43,146 - root - ERROR - Traceback (most recent call last):
  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 323, 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 "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 198, 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 "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 169, in _map_node_over_list
    process_inputs(input_dict, i)
  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/execution.py", line 158, in process_inputs
    results.append(getattr(obj, func)(**inputs))
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/Mflux-ComfyUI/Mflux_Air.py", line 194, in generate
    flux = load_or_create_flux(model, quantize, Local_model if Local_model else None, lora_paths, lora_scales, ControlNet is not None)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/custom_nodes/Mflux-ComfyUI/Mflux_Air.py", line 24, in load_or_create_flux
    flux_cache[key] = FluxClass(
                      ^^^^^^^^^^
  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/flux/flux.py", line 52, in __init__
    weights = WeightHandler(
              ^^^^^^^^^^^^^^
  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/weight_handler.py", line 28, in __init__
    LoraUtil.apply_loras(self.transformer, lora_paths, lora_scales)
  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/lora_util.py", line 18, in apply_loras
    LoraUtil._apply_lora(transformer, lora_file, lora_scale)
  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/lora_util.py", line 36, in _apply_lora
    lora_transformer, _ = WeightHandler.load_transformer(lora_path=lora_file)
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/weight_handler.py", line 64, in load_transformer
    weights, quantization_level = WeightHandler._get_weights("transformer", root_path, lora_path)
                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/wengyinghui/miniconda3/lib/python3.12/site-packages/mflux/weights/weight_handler.py", line 123, in _get_weights
    weight = list(mx.load(lora_path).items())
                  ^^^^^^^^^^^^^^^^^^
RuntimeError: [load_safetensors] Failed to open file /Users/wengyinghui/ComfyUI_windows_portable/ComfyUI/models/loras/Flux/Hyper-FLUX.1-dev-8steps-lora.safetensors

2024-11-02 09:59:43,146 - root - INFO - Prompt executed in 0.23 seconds
2024-11-02 09:59:43,291 - root - WARNING - Failed to get ComfyUI version: Command '['git', 'describe', '--tags']' returned non-zero exit status 128.

Attached Workflow

Please make sure that workflow does not contain any sensitive information such as API keys or passwords.

{"last_node_id":21,"last_link_id":15,"nodes":[{"id":3,"type":"PreviewImage","pos":{"0":1506,"1":595},"size":{"0":210,"1":246},"flags":{},"order":3,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":1}],"outputs":[],"properties":{"Node name for S&R":"PreviewImage"},"widgets_values":[]},{"id":2,"type":"QuickMfluxNode","pos":{"0":1047,"1":574},"size":{"0":400,"1":352},"flags":{},"order":2,"mode":0,"inputs":[{"name":"Local_model","type":"PATH","link":14,"shape":7},{"name":"image","type":"MfluxImagePipeline","link":null,"shape":7},{"name":"Loras","type":"MfluxLorasPipeline","link":15,"shape":7},{"name":"ControlNet","type":"MfluxControlNetPipeline","link":null,"shape":7}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[1],"slot_index":0}],"properties":{"Node name for S&R":"QuickMfluxNode"},"widgets_values":["a girl play guitar on sofa,","dev","8",1120017715420786,"randomize",1024,1024,8,3.5,true]},{"id":9,"type":"MfluxModelsLoader","pos":{"0":562,"1":566},"size":{"0":315,"1":58},"flags":{},"order":0,"mode":0,"inputs":[],"outputs":[{"name":"Local_model","type":"PATH","links":[14],"slot_index":0}],"properties":{"Node name for S&R":"MfluxModelsLoader"},"widgets_values":["MFLUX.1-dev-8-bit"]},{"id":17,"type":"MfluxLorasLoader","pos":{"0":558,"1":706},"size":{"0":393.82196044921875,"1":178},"flags":{},"order":1,"mode":0,"inputs":[{"name":"Loras","type":"MfluxLorasPipeline","link":null,"shape":7}],"outputs":[{"name":"Loras","type":"MfluxLorasPipeline","links":[15],"slot_index":0}],"properties":{"Node name for S&R":"MfluxLorasLoader"},"widgets_values":["Flux/Hyper-FLUX.1-dev-8steps-lora.safetensors",1,"None",1,"None",1]}],"links":[[1,2,0,3,0,"IMAGE"],[14,9,0,2,0,"PATH"],[15,17,0,2,2,"MfluxLorasPipeline"]],"groups":[],"config":{},"extra":{"ds":{"scale":1.4641000000000062,"offset":[-258.8469816069085,-310.0283203110333]}},"version":0.4}

Additional Context

(Please add any additional context or steps to reproduce the error here)

[tolozine]

另外，设置的extra_model_paths下载模型不会放置到这个目录

[raysers]

@tolozine 啊？我没有设置过extra_model_paths啊，lora一直是comfyui自带的路径，就是model/loras.只要lora放进去就可以识别，但不是所有的lora都能兼容mflux。据mflux官网的说明，应该支持比较好的是C站的大部分lora，但注意官网只说大部分，不是全部，有些不兼容的lora在运行过程中才能得知。

路径这些我在旧的说明文件中有作过说明，但这次更新图生图被我删除了，因为我看这个插件发布至今总共才百来个用户，老用户已经都熟知路径了。这是我的失误，插件比较冷清，我也有点颓丧，以为不会有多少新用户了，因此在readme里我删除了路径说明。如果这个插件仍然有新用户，我会将路径说明添加回去。

[tolozine]

@tolozine 啊？我没有设置过extra_model_paths啊，lora一直是comfyui自带的路径，就是model/loras.只要lora放进去就可以识别，但不是所有的lora都能兼容mflux。据mflux官网的说明，应该支持比较好的是C站的大部分lora，但注意官网只说大部分，不是全部，有些不兼容的lora在运行过程中才能得知。

路径这些我在旧的说明文件中有作过说明，但这次更新图生图被我删除了，因为我看这个插件发布至今总共才百来个用户，老用户已经都熟知路径了。这是我的失误，插件比较冷清，我也有点颓丧，以为不会有多少新用户了，因此在readme里我删除了路径说明。如果这个插件仍然有新用户，我会将路径说明添加回去。

知道了。感谢~你可以放几个你用过的推荐lora这样会更友好

[raysers]

@tolozine 可惜我不是lora重度使用者，不然我确实可以做一些推荐。其实官方就开过这种类似的兼容测试帖子，具体filipstrand/mflux#47

要是让我个人推荐，当然4步dev的lora是值得推荐的，有些lora还有现在仅支持的的controlnet canny也是只能在dev上生效的，能4步dev出图的话确实能大大节省时间，快速抽卡。我这几天在研究怎么实现实时预览，遇到了瓶颈，在实现实时预览之前，这个4步lora确实是方便的权宜选择