Deep learning environment configuration jupyter notebook

today , Bo mainly systematically studies the relevant contents of in-depth learning , Record your learning process here , Make progress with everyone .

Content introduction

• Fundamentals of deep learning ： Linear neural networks . Multilayer perceptron
• Convolutional neural networks ：LeNet , AlexNet , VGG , Inception , ResNet
• Cyclic neural network ：RNN , GRU , LSTM , seq2seq
• Attention mechanism ： Attention , Transformer
• optimization algorithm ： SGD , Adam , Momentum
• High performance computing ： parallel , many GPU , Distributed
• Computer vision ： object detection , Semantic segmentation
• Computer language processing ： Word embedding ,BERT

Deep learning is a very “ Fantasy ” The process of , The reason why the model he generates is excellent or how it works is sometimes incomprehensible to us humans , Don't get too tangled here .
The first is the configuration of the environment , Bloggers have already installed PyCharm and Anaconda And configure the relevant environment , Therefore, only a common package is installed here d2l
For related installation tutorials, you can read my previous blog ：
Anaconda Installation and PyCharm To configure

This process may take a few minutes , Because there are many packages to install
After installation , Let's start learning

jupyter notebook Use

What we use here is jupyter notebook, After we installed Anaconda Generally, it will be installed for us by default .
Input... At the terminal jupyter notebook --generate-config enter , The location of the configuration document is shown below .

We can modify its default installation path , It's convenient for us to use

Look in Notepad （Ctrl+F）NotebookApp.notebook_dir The position of , Change the folder in the back to your favorite folder location .

Open it in the browser

Run a code and find no corresponding package , It turns out that the default is Anaconda Of base, In this environment, we have not installed the corresponding environment before , Then we need to switch our environment , Or we want to use it in a different environment jupyter notebook

Single environment creation

The specific way is as follows ：

conda create -n my-conda-env    #  Create a virtual environment 
conda activate my-conda-env     #  Activate our virtual environment 
conda install ipykernel      #  install python kernel 
ipython kernel install --user --name=my-conda-env-kernel  # python -m ipykernel install --user --name  Name of the environment  --display-name " stay jupyter The name of the environment shown in "
jupyter notebook      #  Start operation 

conda install ipykernel

ipython kernel install --user --name=my-conda-env-kernel

BUG explain ：

ImportError: cannot import name 'generator_to_async_generator' 

BUG solve ：

pip uninstall -y ipython prompt_toolkit
pip install ipython prompt_toolkit

Re install

Create for all environments

Of course, the above method can only be one conda The environment create , We use conda install nb_conda_kernels For all conda The environment create jupyter, Thus, there are many choices .

Install code prompt function

At this time, it cannot prompt code after installation , We just need to perform the following steps
1、 Start menu running Anaconda Prompt (Anaconda3), Enter commands one by one

pip install jupyter_contrib_nbextensions

jupyter contrib nbextension install

pip install jupyter_nbextensions_configurator

jupyter nbextensions_configurator enable

2、 Open again notebook, The successful implementation will be in http://localhost:8888/tree It appears that Nbextensions

Check Hinterland

Open one again python You can see the code prompt in the file

This completes our environment configuration