In my opinion, some people might not be able to contribute to CuPy because of not having an NVIDIA GPU. But they might not know that we can build a development env on google colab(As I did here).

import os
from google.colab import drive
drive.mount('/content/drive')
os.chdir("/content/drive/My Drive/")

!git clone ht

The docs generated for #431 at https://mratsim.github.io/Arraymancer/pca.html have broken formatting:

• Upstream bug on the type:
  
• Random bolding:
  and
  ![image](https:/

在我的应用里，发现带BN+scale结构推理速度更快；但是，我发现scale.cpp不支持bias的参数预加载，即在网络初始化时，无法提前规划权重。我觉得可以采用caffe源码中的实现，将57-59行
bias_param_id_ = this->blobs_.size();
this->blobs_.resize(bias_param_id_ + 1);
this->blobs_[bias_param_id_] = bias_layer_->blobs()[0];
替换为：
if (this->blobs_.size() + bottom.size() < 3) {
// case: blobs.size == 1 && bottom.size == 1
// or blobs.size == 0 && bottom.size == 2