AttributeError: module 'torch_pruning' has no attribute 'prune_conv'

Question

AttributeError: module 'torch_pruning' has no attribute 'prune_conv'

Opened this issue 3 years ago · 12 comments

您好，我明明安装了pip install torch-pruning，但是还是再pruning_plan = DG.get_pruning_plan((model.backbone.backbone.dark2)[0].dconv.conv, tp.prune_conv, idxs=pruning_idxs)这个函数运行时候出现
AttributeError: module 'torch_pruning' has no attribute 'prune_conv'的错误，请问是什么原因呢

Answer 1 · 2022-07-19T03:03:17.000Z

你是不是没安装对呀？是pip install torch_pruning，不是torch-pruning

Answer 2 · 2022-07-19T03:06:26.000Z

你是没安装对呀？是pip install torch_pruning，不是torch-pruning

其实我两个都试了一遍还是不行，它好像安装到别的地方了，但是我把库文件复制过来了，按理说也不该出问题的啊

Answer 3 · 2022-07-19T03:09:13.000Z

那你就是没正确安装上呗，你试试在终端或者用conda激活你的环境然后安装，可以加上版本进行安装pip install torch_pruning==0.2.7

Answer 4 · 2022-07-21T01:46:26.000Z

剪枝后有转onnx的方法嘛，转onnx出现了size mismatch for backbone.backbone.stem.conv.conv.weight: copying a param with shape torch.Size([10, 12, 3, 3]) from checkpoint, the shape in current model is torch.Size([16, 12, 3, 3]).问题

Answer 5 · 2022-07-21T01:53:01.000Z

剪枝后我没有转onnx，你这个是出现了维度不匹配的问题。首先你在转onnx或者推理的时候不需要再用model=YOLOX()进行实例化，可以直接用model = torch.load('你剪枝后的权重')进行模型实例化，看看这里对不对，还有就是你传入的shape以及类别数量是否正确 | | 尹以鹏 | | ***@***.*** | On 7/21/2022 ***@***.***> wrote：剪枝后有转onnx的方法嘛，转onnx出现了size mismatch for backbone.backbone.stem.conv.conv.weight: copying a param with shape torch.Size([10, 12, 3, 3]) from checkpoint, the shape in current model is torch.Size([16, 12, 3, 3]).问题 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

Answer 6 · 2022-07-21T02:35:46.000Z

/

剪枝后我没有转onnx，你这个是出现了维度不匹配的问题。首先你在转onnx或者推理的时候不需要再用model=YOLOX()进行实例化，可以直接用model = torch.load('你剪枝后的权重')进行模型实例化，看看这里对不对，还有就是你传入的shape以及类别数量是否正确 | | 尹以鹏 | | @.*** | On 7/21/2022 @.> wrote：剪枝后有转onnx的方法嘛，转onnx出现了size mismatch for backbone.backbone.stem.conv.conv.weight: copying a param with shape torch.Size([10, 12, 3, 3]) from checkpoint, the shape in current model is torch.Size([16, 12, 3, 3]).问题 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.>

但是如何能够不加载YoloBody模型，直接用torch.load实例化呢

Answer 7 · 2022-07-21T02:46:41.000Z

首先你需要区分开仅保存权值和权值与网络结构一起保存。平常我们保存模型是采用torch.save(model.state_dict(),"mymodel.pth")这种方式是仅保存权值，因此我们在网络实例化的时候，首先需要用model = YOLOBody()先实例化模型【此刻的moel也有权重，但它是默认的权重，我们只是为了获取网络结构而已】，然后我们一般是再用model.load_state_dict(torch.load("mymodel.pth"))，将我们训练好的权值加载到网络中，这样是没问题的，只要我们的pth权重shape和model shape一样就行。但是！剪枝后的模型网络结构发生了变化！所以最后模型的保存是用torch.save(model,"pruned.pth")，这里的保存会将我们训练好的权重与我们的网络结构一起保存，即进行了捆绑。如果你还用model = YOLOBody()实例化模型后再用load会报shape不匹配问题，因为剪枝不会帮助你修改代码的。因此你仅仅用model = torch.load("pruned.pth")即可，这会将网络模型和权重都加载，后面的model.eval()推理等和之间一样就行。不知道这样你明白没有 | | 尹以鹏 | | ***@***.*** | On 7/21/2022 ***@***.***> wrote： / 剪枝后我没有转onnx，你这个是出现了维度不匹配的问题。首先你在转onnx或者推理的时候不需要再用model=YOLOX()进行实例化，可以直接用model = torch.load('你剪枝后的权重')进行模型实例化，看看这里对不对，还有就是你传入的shape以及类别数量是否正确 | | 尹以鹏 | | @.*** | On 7/21/2022 @.> wrote：剪枝后有转onnx的方法嘛，转onnx出现了size mismatch for backbone.backbone.stem.conv.conv.weight: copying a param with shape torch.Size([10, 12, 3, 3]) from checkpoint, the shape in current model is torch.Size([16, 12, 3, 3]).问题 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.> 但是如何能够不加载YoloBody模型，直接用torch.load实例化呢 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

Answer 8 · 2022-08-05T02:16:00.000Z

要是转onnx的话基本上就不能微调了是嘛，另外如果想进行通道剪枝该怎么做呢

Answer 9 · 2022-09-03T07:09:29.000Z

大佬，有没有BN层剪枝的代码啊

Answer 10 · 2022-09-03T08:51:09.000Z

将代码中的 ifisinstance(m, nn.Conv2d) andminincluded_layers: 改成ifisinstance(m, nn.nn.BatchNorm2d()) andminincluded_layers: | | 尹以鹏 | | ***@***.*** | ---- Replied Message ---- | From | ***@***.***> | | Date | 9/3/2022 15:09 | | To | ***@***.***> | | Cc | ***@***.***> , ***@***.***> | | Subject | Re: [YINYIPENG-EN/Pruning_for_YOLOX] AttributeError: module 'torch_pruning' has no attribute 'prune_conv' (Issue #1) | 大佬，有没有BN层剪枝的代码啊 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

Answer 11 · 2022-09-03T08:53:05.000Z

感觉还要稀疏化训练，我再看看，自己写了一部分

Answer 12 · 2022-10-11T07:39:32.000Z

onnx我没转，不过应该是可以的。这个代码就是通道剪枝啊 | | 尹以鹏 | | ***@***.*** | On 8/5/2022 ***@***.***> wrote：要是转onnx的话基本上就不能微调了是嘛，另外如果想进行通道剪枝该怎么做呢 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>