Training the model causes the issue.

[ishparsh]

This is the model architecture:

YOLO(
  (model): DetectionModel(
    (model): Sequential(
      (0): Conv(
        (conv): Conv2d(3, 16, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(16, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU(inplace=True)
      )
      (1): Conv(
        (conv): Conv2d(16, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(32, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
        (act): SiLU(inplace=True)
      )
      (2): C2f(
        (cv1): Conv(
          (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU(inplace=True)
        )
        (cv2): Conv(
          (conv): Conv2d(48, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)
          (act): SiLU(inplace=True)
        )
        (m): ModuleList(
...
        )
      )
    )
  )
)

After the pruning operation the architecture becomes:

YOLO(
  (model): DetectionModel(
    (model): Sequential(
      (0): Conv(
        (conv): Conv2d(3, 11, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(11, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (1): Conv(
        (conv): Conv2d(11, 22, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
        (bn): BatchNorm2d(22, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (act): SiLU()
      )
      (2): C2f(
        (cv1): Conv(
          (conv): Conv2d(22, 22, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(22, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (cv2): Conv(
          (conv): Conv2d(33, 22, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(22, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (m): ModuleList(
...
        )
      )
    )
    
    

But whenever I do the fine tuning, I guess it is there is transfer learning which downloads the model and changes to model architecture back to original architecture. To make it disable I uses pretrained = False, amp = False. But still the when I do the fine tuning the model doesn't train over the pruned model. It switch back to original yolov8 model.

[Goddaman]

i meet the same problem, did you solve it?