这是我的代码，我试图将 .npy 文件加载到火炬：
# Load the state dictionary from the numpy file
state_dict_np = np.load(model_path, allow_pickle=True).item()
# Convert numpy arrays within the state dictionary to PyTorch tensors
state_dict_torch = {k: torch.tensor(v, dtype=torch.float32).cpu() for k, v in state_dict_np.items()}
# Load the converted state dictionary into the model
self.model.load_state_dict(state_dict_torch)
这是我遇到的错误：
Error while subprocess initialization: Traceback (most recent call last):
File "/app/core/joblib/SubprocessorBase.py", line 62, in _subprocess_run
self.on_initialize(client_dict)
File "/app/mainscripts/Extractor.py", line 73, in on_initialize
self.rects_extractor = facelib.S3FDExtractor(place_model_on_cpu=place_model_on_cpu)
File "/app/facelib/S3FDExtractor.py", line 156, in __init__
self.model.load_state_dict(state_dict_torch)
File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for S3FD:
Missing key(s) in state_dict: "conv1_1.weight", "conv1_1.bias", "conv1_2.weight", "conv1_2.bias", "conv2_1.weight", "conv2_1.bias", "conv2_2.weight", "conv2_2.bias", "conv3_1.weight", "conv3_1.bias", "conv3_2.weight", "conv3_2.bias", "conv3_3.weight", "conv3_3.bias", "conv4_1.weight", "conv4_1.bias", "conv4_2.weight", "conv4_2.bias", "conv4_3.weight", "conv4_3.bias", "conv5_1.weight", "conv5_1.bias", "conv5_2.weight", "conv5_2.bias", "conv5_3.weight", "conv5_3.bias", "fc6.weight", "fc6.bias", "fc7.weight", "fc7.bias", "conv6_1.weight", "conv6_1.bias", "conv6_2.weight", "conv6_2.bias", "conv7_1.weight", "conv7_1.bias", "conv7_2.weight", "conv7_2.bias", "conv3_3_norm.weight", "conv4_3_norm.weight", "conv5_3_norm.weight", "conv3_3_norm_mbox_conf.weight", "conv3_3_norm_mbox_conf.bias", "conv3_3_norm_mbox_loc.weight", "conv3_3_norm_mbox_loc.bias", "conv4_3_norm_mbox_conf.weight", "conv4_3_norm_mbox_conf.bias", "conv4_3_norm_mbox_loc.weight", "conv4_3_norm_mbox_loc.bias", "conv5_3_norm_mbox_conf.weight", "conv5_3_norm_mbox_conf.bias", "conv5_3_norm_mbox_loc.weight", "conv5_3_norm_mbox_loc.bias", "fc7_mbox_conf.weight", "fc7_mbox_conf.bias", "fc7_mbox_loc.weight", "fc7_mbox_loc.bias", "conv6_2_mbox_conf.weight", "conv6_2_mbox_conf.bias", "conv6_2_mbox_loc.weight", "conv6_2_mbox_loc.bias", "conv7_2_mbox_conf.weight", "conv7_2_mbox_conf.bias", "conv7_2_mbox_loc.weight", "conv7_2_mbox_loc.bias".
Unexpected key(s) in state_dict: "conv1_1/weight:0", "conv1_1/bias:0", "conv1_2/weight:0", "conv1_2/bias:0", "conv2_1/weight:0", "conv2_1/bias:0", "conv2_2/weight:0", "conv2_2/bias:0", "conv3_1/weight:0", "conv3_1/bias:0", "conv3_2/weight:0", "conv3_2/bias:0", "conv3_3/weight:0", "conv3_3/bias:0", "conv4_1/weight:0", "conv4_1/bias:0", "conv4_2/weight:0", "conv4_2/bias:0", "conv4_3/weight:0", "conv4_3/bias:0", "conv5_1/weight:0", "conv5_1/bias:0", "conv5_2/weight:0", "conv5_2/bias:0", "conv5_3/weight:0", "conv5_3/bias:0", "fc6/weight:0", "fc6/bias:0", "fc7/weight:0", "fc7/bias:0", "conv6_1/weight:0", "conv6_1/bias:0", "conv6_2/weight:0", "conv6_2/bias:0", "conv7_1/weight:0", "conv7_1/bias:0", "conv7_2/weight:0", "conv7_2/bias:0", "conv3_3_norm/weight:0", "conv4_3_norm/weight:0", "conv5_3_norm/weight:0", "conv3_3_norm_mbox_conf/weight:0", "conv3_3_norm_mbox_conf/bias:0", "conv3_3_norm_mbox_loc/weight:0", "conv3_3_norm_mbox_loc/bias:0", "conv4_3_norm_mbox_conf/weight:0", "conv4_3_norm_mbox_conf/bias:0", "conv4_3_norm_mbox_loc/weight:0", "conv4_3_norm_mbox_loc/bias:0", "conv5_3_norm_mbox_conf/weight:0", "conv5_3_norm_mbox_conf/bias:0", "conv5_3_norm_mbox_loc/weight:0", "conv5_3_norm_mbox_loc/bias:0", "fc7_mbox_conf/weight:0", "fc7_mbox_conf/bias:0", "fc7_mbox_loc/weight:0", "fc7_mbox_loc/bias:0", "conv6_2_mbox_conf/weight:0", "conv6_2_mbox_conf/bias:0", "conv6_2_mbox_loc/weight:0", "conv6_2_mbox_loc/bias:0", "conv7_2_mbox_conf/weight:0", "conv7_2_mbox_conf/bias:0", "conv7_2_mbox_loc/weight:0", "conv7_2_mbox_loc/bias:0".
之前我尝试像这样重命名键：
renamed_state_dict = {}
for key, value in state_dict.items():
new_key = key.split("/")[0]
if "weight" in key:
new_key += ".weight"
elif "bias" in key:
new_key += ".bias"
renamed_state_dict[new_key] = value
self.model.load_state_dict(renamed_state_dict)
但随后出现此错误：
Error while subprocess initialization: Traceback (most recent call last):
File "/app/core/joblib/SubprocessorBase.py", line 62, in _subprocess_run
self.on_initialize(client_dict)
File "/app/mainscripts/Extractor.py", line 73, in on_initialize
self.rects_extractor = facelib.S3FDExtractor(place_model_on_cpu=place_model_on_cpu)
File "/app/facelib/S3FDExtractor.py", line 165, in __init__
self.model.load_state_dict(renamed_state_dict)
File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for S3FD:
While copying the parameter named "conv1_1.weight", expected torch.Tensor or Tensor-like object from checkpoint but received <class 'numpy.ndarray'>
While copying the parameter named "conv1_1.bias", expected torch.Tensor or Tensor-like object from checkpoint but received <class 'numpy.ndarray'>
While copying the parameter named "conv1_2.weight", expected torch.Tensor or Tensor-like object from checkpoint but received <class 'numpy.ndarray'>
While copying the parameter named "conv1_2.bias", expected torch.Tensor or Tensor-like object from checkpoint but received <class 'numpy.ndarray'>
...
While copying the parameter named "conv3_1.weight", expected torch.Tensor or Tensor-like object from checkpoint but received <class 'numpy.ndarray'>
我哪里出错了？
更新
根据用户建议，我尝试了这个：
def rename_key(key):
new_key = key.split(":")[0] # discard :0 and similar
key_elements = new_key.split("/")
new_key = ".".join(key_elements) # replace every / with .
return new_key
# Load the state dictionary from the numpy file
state_dict_np = np.load(model_path, allow_pickle=True).item()
# Convert numpy arrays within the state dictionary to PyTorch tensors
state_dict_torch = {rename_key(k): torch.tensor(v, dtype=torch.float32).cpu() for k, v in state_dict_np.items()}
# Load the converted state dictionary into the model
self.model.load_state_dict(state_dict_torch)
并收到此错误：
Error while subprocess initialization: Traceback (most recent call last):
File "/app/core/joblib/SubprocessorBase.py", line 62, in _subprocess_run
self.on_initialize(client_dict)
File "/app/mainscripts/Extractor.py", line 73, in on_initialize
self.rects_extractor = facelib.S3FDExtractor(place_model_on_cpu=place_model_on_cpu)
File "/app/facelib/S3FDExtractor.py", line 162, in __init__
self.model.load_state_dict(state_dict_torch)
File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for S3FD:
size mismatch for conv1_1.weight: copying a param with shape torch.Size([3, 3, 3, 64]) from checkpoint, the shape in current model is torch.Size([64, 3, 3, 3]).
size mismatch for conv1_1.bias: copying a param with shape torch.Size([1, 1, 1, 64]) from checkpoint, the shape in current model is torch.Size([64]).
size mismatch for conv1_2.weight: copying a param with shape torch.Size([3, 3, 64, 64]) from checkpoint, the shape in current model is torch.Size([64, 64, 3, 3]).
size mismatch for conv1_2.bias: copying a param with shape torch.Size([1, 1, 1, 64]) from checkpoint, the shape in current model is torch.Size([64]).
size mismatch for conv2_1.weight: copying a param with shape torch.Size([3, 3, 64, 128]) from checkpoint, the shape in current model is torch.Size([128, 64, 3, 3]).
size mismatch for conv2_1.bias: copying a param with shape torch.Size([1, 1, 1, 128]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for conv2_2.weight: copying a param with shape torch.Size([3, 3, 128, 128]) from checkpoint, the shape in current model is torch.Size([128, 128, 3, 3]).
size mismatch for conv2_2.bias: copying a param with shape torch.Size([1, 1, 1, 128]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for conv3_1.weight: copying a param with shape torch.Size([3, 3, 128, 256]) from checkpoint, the shape in current model is torch.Size([256, 128, 3, 3]).
size mismatch for conv3_1.bias: copying a param with shape torch.Size([1, 1, 1, 256]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for conv3_2.weight: copying a param with shape torch.Size([3, 3, 256, 256]) from checkpoint, the shape in current model is torch.Size([256, 256, 3, 3]).
size mismatch for conv3_2.bias: copying a param with shape torch.Size([1, 1, 1, 256]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for conv3_3.weight: copying a param with shape torch.Size([3, 3, 256, 256]) from checkpoint, the shape in current model is torch.Size([256, 256, 3, 3]).
size mismatch for conv3_3.bias: copying a param with shape torch.Size([1, 1, 1, 256]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for conv4_1.weight: copying a param with shape torch.Size([3, 3, 256, 512]) from checkpoint, the shape in current model is torch.Size([512, 256, 3, 3]).
size mismatch for conv4_1.bias: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for conv4_2.weight: copying a param with shape torch.Size([3, 3, 512, 512]) from checkpoint, the shape in current model is torch.Size([512, 512, 3, 3]).
size mismatch for conv4_2.bias: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for conv4_3.weight: copying a param with shape torch.Size([3, 3, 512, 512]) from checkpoint, the shape in current model is torch.Size([512, 512, 3, 3]).
size mismatch for conv4_3.bias: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for conv5_1.weight: copying a param with shape torch.Size([3, 3, 512, 512]) from checkpoint, the shape in current model is torch.Size([512, 512, 3, 3]).
size mismatch for conv5_1.bias: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for conv5_2.weight: copying a param with shape torch.Size([3, 3, 512, 512]) from checkpoint, the shape in current model is torch.Size([512, 512, 3, 3]).
size mismatch for conv5_2.bias: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for conv5_3.weight: copying a param with shape torch.Size([3, 3, 512, 512]) from checkpoint, the shape in current model is torch.Size([512, 512, 3, 3]).
size mismatch for conv5_3.bias: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for fc6.weight: copying a param with shape torch.Size([3, 3, 512, 1024]) from checkpoint, the shape in current model is torch.Size([1024, 512, 3, 3]).
size mismatch for fc6.bias: copying a param with shape torch.Size([1, 1, 1, 1024]) from checkpoint, the shape in current model is torch.Size([1024]).
size mismatch for fc7.weight: copying a param with shape torch.Size([1, 1, 1024, 1024]) from checkpoint, the shape in current model is torch.Size([1024, 1024, 1, 1]).
size mismatch for fc7.bias: copying a param with shape torch.Size([1, 1, 1, 1024]) from checkpoint, the shape in current model is torch.Size([1024]).
size mismatch for conv6_1.weight: copying a param with shape torch.Size([1, 1, 1024, 256]) from checkpoint, the shape in current model is torch.Size([256, 1024, 1, 1]).
size mismatch for conv6_1.bias: copying a param with shape torch.Size([1, 1, 1, 256]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for conv6_2.weight: copying a param with shape torch.Size([3, 3, 256, 512]) from checkpoint, the shape in current model is torch.Size([512, 256, 3, 3]).
size mismatch for conv6_2.bias: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for conv7_1.weight: copying a param with shape torch.Size([1, 1, 512, 128]) from checkpoint, the shape in current model is torch.Size([128, 512, 1, 1]).
size mismatch for conv7_1.bias: copying a param with shape torch.Size([1, 1, 1, 128]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for conv7_2.weight: copying a param with shape torch.Size([3, 3, 128, 256]) from checkpoint, the shape in current model is torch.Size([256, 128, 3, 3]).
size mismatch for conv7_2.bias: copying a param with shape torch.Size([1, 1, 1, 256]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for conv3_3_norm.weight: copying a param with shape torch.Size([1, 1, 1, 256]) from checkpoint, the shape in current model is torch.Size([1, 256, 1, 1]).
size mismatch for conv4_3_norm.weight: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([1, 512, 1, 1]).
size mismatch for conv5_3_norm.weight: copying a param with shape torch.Size([1, 1, 1, 512]) from checkpoint, the shape in current model is torch.Size([1, 512, 1, 1]).
size mismatch for conv3_3_norm_mbox_conf.weight: copying a param with shape torch.Size([3, 3, 256, 4]) from checkpoint, the shape in current model is torch.Size([4, 256, 3, 3]).
size mismatch for conv3_3_norm_mbox_conf.bias: copying a param with shape torch.Size([1, 1, 1, 4]) from checkpoint, the shape in current model is torch.Size([4]).
size mismatch for conv3_3_norm_mbox_loc.weight: copying a param with shape torch.Size([3, 3, 256, 4]) from checkpoint, the shape in current model is torch.Size([4, 256, 3, 3]).
size mismatch for conv3_3_norm_mbox_loc.bias: copying a param with shape torch.Size([1, 1, 1, 4]) from checkpoint, the shape in current model is torch.Size([4]).
size mismatch for conv4_3_norm_mbox_conf.weight: copying a param with shape torch.Size([3, 3, 512, 2]) from checkpoint, the shape in current model is torch.Size([2, 512, 3, 3]).
size mismatch for conv4_3_norm_mbox_conf.bias: copying a param with shape torch.Size([1, 1, 1, 2]) from checkpoint, the shape in current model is torch.Size([2]).
size mismatch for conv4_3_norm_mbox_loc.weight: copying a param with shape torch.Size([3, 3, 512, 4]) from checkpoint, the shape in current model is torch.Size([4, 512, 3, 3]).
size mismatch for conv4_3_norm_mbox_loc.bias: copying a param with shape torch.Size([1, 1, 1, 4]) from checkpoint, the shape in current model is torch.Size([4]).
size mismatch for conv5_3_norm_mbox_conf.weight: copying a param with shape torch.Size([3, 3, 512, 2]) from checkpoint, the shape in current model is torch.Size([2, 512, 3, 3]).
size mismatch for conv5_3_norm_mbox_conf.bias: copying a param with shape torch.Size([1, 1, 1, 2]) from checkpoint, the shape in current model is torch.Size([2]).
size mismatch for conv5_3_norm_mbox_loc.weight: copying a param with shape torch.Size([3, 3, 512, 4]) from checkpoint, the shape in current model is torch.Size([4, 512, 3, 3]).
size mismatch for conv5_3_norm_mbox_loc.bias: copying a param with shape torch.Size([1, 1, 1, 4]) from checkpoint, the shape in current model is torch.Size([4]).
size mismatch for fc7_mbox_conf.weight: copying a param with shape torch.Size([3, 3, 1024, 2]) from checkpoint, the shape in current model is torch.Size([2, 1024, 3, 3]).
size mismatch for fc7_mbox_conf.bias: copying a param with shape torch.Size([1, 1, 1, 2]) from checkpoint, the shape in current model is torch.Size([2]).
size mismatch for fc7_mbox_loc.weight: copying a param with shape torch.Size([3, 3, 1024, 4]) from checkpoint, the shape in current model is torch.Size([4, 1024, 3, 3]).
size mismatch for fc7_mbox_loc.bias: copying a param with shape torch.Size([1, 1, 1, 4]) from checkpoint, the shape in current model is torch.Size([4]).
size mismatch for conv6_2_mbox_conf.weight: copying a param with shape torch.Size([3, 3, 512, 2]) from checkpoint, the shape in current model is torch.Size([2, 512, 3, 3]).
size mismatch for conv6_2_mbox_conf.bias: copying a param with shape torch.Size([1, 1, 1, 2]) from checkpoint, the shape in current model is torch.Size([2]).
size mismatch for conv6_2_mbox_loc.weight: copying a param with shape torch.Size([3, 3, 512, 4]) from checkpoint, the shape in current model is torch.Size([4, 512, 3, 3]).
size mismatch for conv6_2_mbox_loc.bias: copying a param with shape torch.Size([1, 1, 1, 4]) from checkpoint, the shape in current model is torch.Size([4]).
size mismatch for conv7_2_mbox_conf.weight: copying a param with shape torch.Size([3, 3, 256, 2]) from checkpoint, the shape in current model is torch.Size([2, 256, 3, 3]).
size mismatch for conv7_2_mbox_conf.bias: copying a param with shape torch.Size([1, 1, 1, 2]) from checkpoint, the shape in current model is torch.Size([2]).
size mismatch for conv7_2_mbox_loc.weight: copying a param with shape torch.Size([3, 3, 256, 4]) from checkpoint, the shape in current model is torch.Size([4, 256, 3, 3]).
size mismatch for conv7_2_mbox_loc.bias: copying a param with shape torch.Size([1, 1, 1, 4]) from checkpoint, the shape in current model is torch.Size([4]).
解决方案是将您的两个想法结合起来，因为它们解决了两个不同的问题：一个是重命名键，另一个是将 numpy 数组转换为 torch 张量。这可能会起作用：
def rename_key(key):
new_key = key.split(":")[0] # discard :0 and similar
key_elements = new_key.split("/")
new_key = ".".join(key_elements) # replace every / with .
return new_key
# Load the state dictionary from the numpy file
state_dict_np = np.load(model_path, allow_pickle=True).item()
# Convert numpy arrays within the state dictionary to PyTorch tensors
state_dict_torch = {rename_key(k): torch.tensor(v, dtype=torch.float32).cpu() for k, v in state_dict_np.items()}
# Load the converted state dictionary into the model
self.model.load_state_dict(state_dict_torch)