Sensitivity Analysis error

[DAVID-Hown]

Layer 52: Removed 32/128 filters based on criterion 0 and pruned 18432/73728 (25.0%) of the parameters in this convolutional layer. Pruning this layer affects the following layers:
Layer 55: Removed 32/128 input channels and pruned 768/3072 (25.0%) of the parameters in this convolutional layer.

Pruned 19296/6020400 parameters in total. Global sparsity: 0.32%

Model Summary: 263 layers, 6001104 parameters, 57888 gradients
Fusing layers...
IDetect.fuse
Traceback (most recent call last):
File "test.py", line 410, in
r = test(opt.data,
File "test.py", line 68, in test
model = load_pruned_model(weights, pruning_params, criterion, map_location=device)
File "/yolov7-pruning/prune.py", line 210, in load_pruned_model
return pruned_model.float().fuse().eval()
File "/yolov7-pruning/models/yolo.py", line 707, in fuse
m.fuse()
File "/yolov7-pruning/models/yolo.py", line 184, in fuse
self.m[i].bias += torch.matmul(self.m[i].weight.reshape(c1,c2),self.ia[i].implicit.reshape(c2_,c1_)).squeeze(1)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1283, in setattr
self.register_parameter(name, value)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 423, in register_parameter
raise ValueError(
ValueError: Cannot assign non-leaf Tensor to parameter 'bias'. Model parameters must be created explicitly. To express 'bias' as a function of another Tensor, compute the value in the forward() method.