未找到文件：Python Pytorch 上的 archive/constants.pkl

Question

我在准备序列化 torchscript 模型的运算符列表时遇到错误。这里有什么问题呢？加载的时候就来了吗？

Python代码

# Dump list of operators used by MobileNetV2:
import torch, yaml

root = '/content/drive/My Drive/Monitoring/'
model = torch.jit.load(root+'model.pt')
ops = torch.jit.export_opnames(model)
with open('MobileNetV2.yaml', 'w') as output:
    yaml.dump(ops, output)

堆栈跟踪

    ---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-39-8b61c35fb898> in <module>()
      3 
      4 root = '/content/drive/My Drive/Monitoring/'
----> 5 model = torch.jit.load(root+'model.pt')
      6 ops = torch.jit.export_opnames(model)
      7 with open('MobileNetV2.yaml', 'w') as output:

/usr/local/lib/python3.6/dist-packages/torch/jit/_serialization.py in load(f, map_location, _extra_files)
    159     cu = torch._C.CompilationUnit()
    160     if isinstance(f, str) or isinstance(f, pathlib.Path):
--> 161         cpp_module = torch._C.import_ir_module(cu, f, map_location, _extra_files)
    162     else:
    163         cpp_module = torch._C.import_ir_module_from_buffer(

RuntimeError: [enforce fail at inline_container.cc:222] . file not found: archive/constants.pkl

Answer 1

您是否尝试过直接加载模型而不使用jit：
torch.load(model_path, map_location=device)

如果您想通过 加载模型torch.jit.load，则必须通过 保存模型torch.jit.trace，例如：

model = ...
model.eval()
input_tensor = torch.rand(1, 3, 224, 224)
script_model = torch.jit.trace(model, input_tensor)
script_model.save("model.jit.pt")

您可能想检查此线程。