the problems of torch-pruning on yolov12n:MemoryError

[muzili-pengyang]

pruning on yolov12,have this problem:

Traceback (most recent call last):
File "E:\code\Python\big thesis\yolov12-main\ultralytics\yolov12_pruning.py", line 418, in
prune(args)
File "E:\code\Python\big thesis\yolov12-main\ultralytics\yolov12_pruning.py", line 346, in prune
pruner = tp.pruner.BNScalePruner(
^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\anaconda3\envs\yolov12\Lib\site-packages\torch_pruning\pruner\algorithms\batchnorm_scale_pruner.py", line 85, in init
super(BNScalePruner, self).init(
File "D:\anaconda3\envs\yolov12\Lib\site-packages\torch_pruning\pruner\algorithms\base_pruner.py", line 253, in init
for group in self.DG.get_all_groups(ignored_layers=self.ignored_layers, root_module_types=self.root_module_types):
File "D:\anaconda3\envs\yolov12\Lib\site-packages\torch_pruning\dependency\graph.py", line 314, in get_all_groups
group = self.get_pruning_group(
^^^^^^^^^^^^^^^^^^^^^^^
File "D:\anaconda3\envs\yolov12\Lib\site-packages\torch_pruning\dependency\graph.py", line 251, in get_pruning_group
_fix_dependency_graph_non_recursive(*group[0])
File "D:\anaconda3\envs\yolov12\Lib\site-packages\torch_pruning\dependency\graph.py", line 236, in _fix_dependency_graph_non_recursive
new_indices = mapping(new_indices)
^^^^^^^^^^^^^^^^^^^^
File "D:\anaconda3\envs\yolov12\Lib\site-packages\torch_pruning\dependency\index_mapping.py", line 37, in call
new_idxs.append( _helpers._HybridIndex( idx = (i.idx // self._stride), root_idx=i.root_idx ) )
MemoryError

I have tried to change code,like this:
for i in range(args.iterative_steps):

    model.model.train()
    for name, param in model.model.named_parameters():
        param.requires_grad = True

    ignored_layers = []
    unwrapped_parameters = []
    # for m in model.model.modules():
    #     if isinstance(m, (Detect,)):
    #         ignored_layers.append(m)
    for m in model.model.modules():
        if isinstance(m, (Detect, C2f, SPPF)):
            ignored_layers.append(m)

    example_inputs = example_inputs.to(model.device)
    # pruner = tp.pruner.GroupNormPruner(
    #     model.model,
    #     example_inputs,
    #     importance=tp.importance.GroupMagnitudeImportance(),  # L2 norm pruning,
    #     iterative_steps=1,
    #     pruning_ratio=pruning_ratio,
    #     ignored_layers=ignored_layers,
    #     unwrapped_parameters=unwrapped_parameters
    # )

    # 1. 建议明确指定 root_module_types，避免剪枝器尝试分析不必要的层
    # 2. importance 建议使用 MagnitudeImportance，这是基于 BN 层 gamma 值剪枝的标准做法
    pruner = tp.pruner.BNScalePruner(
        model.model,  # 【修改点1】建议保持使用 model.model (即 nn.Module 对象)
        example_inputs,
        ignored_layers=ignored_layers,
        importance=tp.importance.MagnitudeImportance(p=2),  # 【修改点2】替换 GroupMagnitudeImportance
        pruning_ratio=0.5,
        root_module_types=[nn.Conv2d]  # 【建议新增】只针对卷积层进行修剪
    )

but it is still in trouble,can you help me?