Model upload issue

[GMSL1706]

While training, I used SaveModelCallback() to save the model with the best valid loss value as suggested here.

Now, I am trying to upload my model for inference but it doesn't work. I used to the same data preprocessing.
Here is my code:

model_type = models.mmdet.faster_rcnn
backbone = model_type.backbones.resnext101_32x4d_fpn_2x
extra_args = {}

model = model_type.model(backbone=backbone(pretrained=True), num_classes=len(parser.class_map), **extra_args) 
state_dict = torch.load("./faster_rcnn__resnext101_32x4d_fpn_2x.pth")
model.load_state_dict(state_dict)

Output:

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
[<ipython-input-12-decb596209c8>](https://localhost:8080/#) in <module>()
----> 1 model.load_state_dict(state_dict)

[/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py](https://localhost:8080/#) in load_state_dict(self, state_dict, strict)
   1481         if len(error_msgs) > 0:
   1482             raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
-> 1483                                self.__class__.__name__, "\n\t".join(error_msgs)))
   1484         return _IncompatibleKeys(missing_keys, unexpected_keys)
   1485 

RuntimeError: Error(s) in loading state_dict for FasterRCNN:
	Missing key(s) in state_dict: "backbone.conv1.weight", "backbone.bn1.weight", "backbone.bn1.bias", "backbone.bn1.running_mean", "backbone.bn1.running_var", "backbone.layer1.0.conv1.weight", "backbone.layer1.0.bn1.weight", "backbone.layer1.0.bn1.bias", "backbone.layer1.0.bn1.running_mean", "backbone.layer1.0.bn1.running_var", "backbone.layer1.0.conv2.weight", "backbone.layer1.0.bn2.weight", "backbone.layer1.0.bn2.bias", "backbone.layer1.0.bn2.running_mean", "backbone.layer1.0.bn2.running_var", "backbone.layer1.0.conv3.weight", "backbone.layer1.0.bn3.weight", "backbone.layer1.0.bn3.bias", "backbone.layer1.0.bn3.running_mean", "backbone.layer1.0.bn3.running_var", "backbone.layer1.0.downsample.0.weight", "backbone.layer1.0.downsample.1.weight", "backbone.layer1.0.downsample.1.bias", "backbone.layer1.0.downsample.1.running_mean", "backbone.layer1.0.downsample.1.running_var", "backbone.layer1.1.conv1.weight", "backbone.layer1.1.bn1.weight", "backbone.layer1.1.bn1.bias", "backbone.layer1.1.bn1.running_mean", "backbone.layer1.1.bn1.running_var", "backbone.layer1.1.conv2.weight", "backbone.layer1.1.bn2.weight", "backbone.layer1.1.bn2.bias", "backbone.layer1.1.bn2.running_mean", "backbone.layer1.1.bn2.running_var", "backbone.layer1.1.conv3.weight", "backbone.layer1.1.bn3.weight", "backbone.layer1.1.bn3.bias", "backbone.layer1.1.bn3.running_mean", "backbone.layer1.1.bn3.running_var", "backbone.layer1.2.conv1.weight", "backbone.layer1.2.bn1.weight", "backbone.layer1.2.bn1.bias", "bac...
	Unexpected key(s) in state_dict: "model.backbone.conv1.weight", "model.backbone.bn1.weight", "model.backbone.bn1.bias", "model.backbone.bn1.running_mean", "model.backbone.bn1.running_var", "model.backbone.bn1.num_batches_tracked", "model.backbone.layer1.0.conv1.weight", "model.backbone.layer1.0.bn1.weight", "model.backbone.layer1.0.bn1.bias", "model.backbone.layer1.0.bn1.running_mean", "model.backbone.layer1.0.bn1.running_var", "model.backbone.layer1.0.bn1.num_batches_tracked", "model.backbone.layer1.0.conv2.weight", "model.backbone.layer1.0.bn2.weight", "model.backbone.layer1.0.bn2.bias", "model.backbone.layer1.0.bn2.running_mean", "model.backbone.layer1.0.bn2.running_var", "model.backbone.layer1.0.bn2.num_batches_tracked", "model.backbone.layer1.0.conv3.weight", "model.backbone.layer1.0.bn3.weight", "model.backbone.layer1.0.bn3.bias", "model.backbone.layer1.0.bn3.running_mean", "model.backbone.layer1.0.bn3.running_var", "model.backbone.layer1.0.bn3.num_batches_tracked", "model.backbone.layer1.0.downsample.0.weight", "model.backbone.layer1.0.downsample.1.weight", "model.backbone.layer1.0.downsample.1.bias", "model.backbone.layer1.0.downsample.1.running_mean", "model.backbone.layer1.0.downsample.1.running_var", "model.backbone.layer1.0.downsample.1.num_batches_tracked", "model.backbone.layer1.1.conv1.weight", "model.backbone.layer1.1.bn1.weight", "model.backbone.layer1.1.bn1.bias", "model.backbone.layer1.1.bn1.running_mean", "model.backbone.layer1.1.bn1.running_var", "mod...

Can someone help me ?

[FraPochetti]

As you can see torch is looking for backbone.conv1.weight but it is finding model.backbone.conv1.weight.
Basically all the keys in the pth dictionary have model. appended because of how the weights were saved.

Just open the dictionary and manually fix the keys.
Something like that

state_dict = torch.load("./faster_rcnn__resnext101_32x4d_fpn_2x.pth")
for k in state_dict.keys():
    state_dict[k.replace("model.", "")] = state_dict.pop(k)

[GMSL1706]

Thank you very much!

Here is the code that I used and it works according to this issue :

state_dict = torch.load("./faster_rcnn__resnext101_32x4d_fpn_2x.pth")
my_dic_keys = list(state_dict.keys())
for key in my_dic_keys:
    state_dict[key.replace("model.", "")] = state_dict[key]
    del state_dict[key]

model.load_state_dict(state_dict)