求助：safety_checker文件缺失及Torch未编译CUDA的技术问题

运行基于FastAPI的Stable Diffusion代码时遇到两个错误：

• safety_checker\model.safetensors not found
• AssertionError: Torch not compiled with CUDA enabled

运行代码 #

from auth_token import auth_token
from fastapi import FastAPI, Response
from fastapi.middleware.cors import CORSMiddleware
import torch
from torch import autocast
from diffusers import StableDiffusionPipeline
from io import BytesIO
import base64 


app = FastAPI()

app.add_middleware(
    CORSMiddleware, 
    allow_credentials=True, 
    allow_origins=["*"], 
    allow_methods=["*"], 
    allow_headers=["*"]
)

device = "cuda"
model_id = "CompVis/stable-diffusion-v1-4"
pipe = StableDiffusionPipeline.from_pretrained(model_id, revision="fp16", torch_dtype=torch.float16, use_auth_token=auth_token)
pipe.to(device)

@app.get("/")
def generate(prompt: str): 
    with autocast(device): 
        image = pipe(prompt, guidance_scale=8.5).images[0]

    image.save("testimage.png")
    buffer = BytesIO()
    image.save(buffer, format="PNG")
    imgstr = base64.b64encode(buffer.getvalue())

    return Response(content=imgstr, media_type="image/png")

完整报错信息 #

safety_checker\model.safetensors not found
Loading pipeline components...:  71%|█████████████████████████████████████████████████████████████████████████████▊                               | 5/7 [00:01<00:00,  3.89it/s]C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\transformers\models\clip\feature_extraction_clip.py:28: FutureWarning: The class CLIPFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please use CLIPImageProcessor instead.
  warnings.warn(
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["id2label"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["bos_token_id"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["eos_token_id"]` will be overriden.
Loading pipeline components...: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:01<00:00,  3.63it/s]
Traceback (most recent call last):
  File "g:\Projects\Code-That-ReactStableDiffusion-main\api\api.py", line 23, in <module>
    pipe.to(device)
  File "C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\diffusers\pipelines\pipeline_utils.py", line 727, in to
    module.to(torch_device, torch_dtype)
  File "C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\torch\nn\modules\module.py", line 1145, in to
    return self._apply(convert)
           ^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\torch\nn\modules\module.py", line 797, in _apply
    module._apply(fn)
  File "C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\torch\nn\modules\module.py", line 797, in _apply
    module._apply(fn)
  File "C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\torch\nn\modules\module.py", line 820, in _apply
    param_applied = fn(param)
                    ^^^^^^^^^
  File "C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\torch\nn\modules\module.py", line 1143, in convert
    return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\mouba\AppData\Roaming\Python\Python311\site-packages\torch\cuda\__init__.py", line 239, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

1. 解决safety_checker\model.safetensors not found错误 #

这个错误是安全检查器权重文件缺失，有两种处理方式：

• 禁用安全检查器：初始化StableDiffusionPipeline时添加safety_checker=None参数，跳过该组件加载，修改代码如下：

  pipe = StableDiffusionPipeline.from_pretrained(
      model_id, 
      revision="fp16", 
      torch_dtype=torch.float16, 
      use_auth_token=auth_token,
      safety_checker=None  # 新增该行
  )
  

• 重新下载完整模型：删除本地模型缓存目录（通常在C:\Users\你的用户名\.cache\huggingface\hub下的CompVis/stable-diffusion-v1-4文件夹），重新运行代码让diffusers自动下载完整权重。

2. 解决AssertionError: Torch not compiled with CUDA enabled错误 #

该错误说明当前PyTorch版本不支持CUDA，按以下步骤处理：

• 卸载现有PyTorch：

  pip uninstall torch torchvision torchaudio
  

• 安装带CUDA支持的PyTorch：根据你的NVIDIA显卡CUDA版本（通过nvidia-smi查看），安装对应版本。例如CUDA 11.8版本运行：

  pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
  

• 验证CUDA是否启用：运行python -c "import torch; print(torch.cuda.is_available())"，输出True则说明成功。

• 无NVIDIA显卡的处理：将代码中device = "cuda"改为device = "cpu"，同时移除revision="fp16"和torch_dtype=torch.float16参数（CPU不支持半精度运算），修改后代码：

  device = "cpu"
  model_id = "CompVis/stable-diffusion-v1-4"
  pipe = StableDiffusionPipeline.from_pretrained(model_id, use_auth_token=auth_token, safety_checker=None)
  pipe.to(device)
  

内容的提问来源于stack exchange，提问作者131 Mohammad Ubaida