Xu

> @scarlett2018 yes, I will support this op in v2.7 if possible. have you fix it?

> @xuzhuang1996 This is a simple workaround, #4852, but as this pr description said, there may be insufficient speedup near `expand_as`. Any scenario that uses `expand_as` will help us to...

> hi @xuzhuang1996 Do you still have this problem? sadly not!

> Have you get the 82.5% mPA in Paris6K?

> Coud you help me how to create the images labels? > > And the labels what it represent > > thank you very much 你好，请问如何创建Labels？我用的Pittsbburgh250k数据集，每张图片所属的category 应该从哪里来。谢谢

If you want to use the functionality directly, you can use Microsoft's NNI. [show](https://github.com/Microsoft/nni/blob/6a082fe9a8c891f5367a4277d7335d39de9ca778/docs/en_US/Compression/Pruner.rst#id7)

我最终没有使用作者的代码仓来压缩剪纸模型。使用的是微软的nni，里面有集成作者的剪枝算法fpgm，你可以了解一下。 ---Original--- From: ***@***.***> Date: Sat, Feb 18, 2023 10:42 AM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject: Re: [he-y/filter-pruning-geometric-median] small model is not small(#71) If you want to use the functionality...

> Can you share a stack trace? Any other details you have? this is full stack trace: ``` Traceback (most recent call last): File "C:\Users\Administrator\PycharmProjects\pythonProject\main.py", line 129, in target_images =...