Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Appearance settings

Latest commit

 

History

History
History
28 lines (20 loc) · 1.6 KB

File metadata and controls

28 lines (20 loc) · 1.6 KB
Copy raw file
Download raw file
Outline
Edit and raw actions

Tensorflow2.1 GPU安装与Pytorch1.3 GPU安装

参考我之前写的博文:Centos7 安装Tensorflow2.1 GPU以及Pytorch1.3 GPU(CUDA10.1)

keras functional api训练的模型权重与subclassed训练的模型权重能否混用 [tensorflow2.0.0]

强烈不建议混用,即使两个模型的名称结构完全一致也不要混用,里面有坑,用什么方法训练的模型就载入相应的模型权重

使用subclassed模型时无法使用model.summary() [tensorflow2.0.0]

subclassed模型在实例化时没有自动进行build操作(只有在开始训练时,才会自动进行build),如果需要使用summary操作,需要提前手动build
model.build((batch_size, height, width, channel))

无法使用keras的plot_model(model, 'my_model.png')问题 [tensorflow2.0.0]

在linux下你需要安装一些包:

  • pip install pydot==1.2.3
  • sudo apt-get install graphviz

在windows中,同样需要安装一些包(windows比较麻烦):

为什么每计算一个batch,就需要调用一次optimizer.zero_grad() [Pytorch1.3]

如果不清除历史梯度,就会对计算的历史梯度进行累加(通过这个特性你能够变相实现一个很大batch数值的训练)
参考链接:https://www.zhihu.com/question/303070254

Pytorch1.3 ImportError: cannot import name 'PILLOW_VERSION' [Pytorch1.3]

pillow版本过高导致,安装版本号小于7.0.0即可

Morty Proxy This is a proxified and sanitized view of the page, visit original site.