-
## 一言でいうと
チャンネルの枝刈りをする手法の提案。学習済みのモデルを枝刈りしていくスタイルではなく、任意の構造に対し適切な重みを生成できるネットワーク(PruningNet)を学習し、生成した重みをセットした様々な構造から適切なものを選ぶ方式を取っている。これにより作成後のFine Tuningを不要にしている。
![image](https://user-images.gith…
-
Hi, I've look through the docs and could not find how to set in the pruning config(suppose we use l1-norm-pruner) so that remained convs have channel number that is divisible by a certain number, say …
-
I try to prune resnet-50
first, Merge batchnorm and conv , the precise is 92%
second, I set channel ration = 0.9 , I only prune before res2a-brach2c , but the precise drop to 65%
I don't know w…
-
Hello,
I followed the example and come up with this:
```
model = resnet18(pretrained=True).eval()
example_inputs = torch.randn(1, 3, 224, 224)
DG = tp.DependencyGraph().build_dependency(mod…
-
# Description
HistFactory includes a threshold setting for the size of statistical uncertainties related to predicted distributions, for example
```
```
taken from [here](https://github.com/ale…
-
hi @wenwei202 , Thank you for your sharing. Based on your paper, I implemented [ZJCV/SSL](https://github.com/ZJCV/SSL) using pytorch, including `train-prune-finetuing` for VGGNet and ResNet
* For V…
-
I trained a model and it contains 'sigmoid' op, it cannot be compressed with chnnel_prunning.
-
/kind bug
I am not sure if this is a bug or just an issue with my system. Any help or suggestion would be greatly appreciated.
**1. What `kops` version are you running? The command `kops version…
-
Hi. I am trying to prune MViTv2 with random pruning with the following code
```python
import torch
import timm
from timm.models.mvitv2 import MultiScaleAttention, MultiScaleVit as MViT
import tor…
-
修改为s模型使用prune_train/prune.py训练后报错:
```
/usr/local/lib/python3.9/dist-packages/torch/functional.py:445: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing a…