Beginner’s Guide to Mambaforge Installation

1. Check if Python is already installed in your system
2. No existing Conda environment in your home directory
3. No ipython and jupyter installed in your System Python
4. Download Mambaforge
5. Install Mamba
6. Install Pytorch and Jupyter

We will be using the Linux shell to install Mambaforge. For Mac and Windows users new to using the terminal, refer to the following:
- For Mac Users, refer to Apple’s Terminal User Guide
- For Windows Users, refer to Microsoft’s Install Linux on Windows with WSL.

1. Check if Python is already installed in your system

Most of the time, you will find an existing version of Python already installed in your machine. This is the System Python, which is used by your system.

user2022@ubuntu:~$ python
Python 3.8.10 (default, Mar 15 2022, 12:22:08) 
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.

user2022@ubuntu:~$ which python
/usr/bin/python

This is not the Python we want to be using. We want to install a separate version of Python. Which Python you install is going to make a difference in making how easy everything is. Mambaforge is the recommended way to install Python for this course.
Next, make sure you are in your home directory by typing the command cd
Use the command pwd to print out the directory you are currently in. In the example below, the home directory is /home/user2022@ubuntu/
```
  user2022@ubuntu:~$ cd
  user2022@ubuntu:~$ pwd
  /home/user2022@ubuntu/
```

2. No existing Conda environment in your home directory

If, in your home directory, there is any conda environment such as Anaconda, Miniforge, Mambaforge. etc, delete them. If you are unsure about deleting them, move them somewhere else.

If you are confident, remove the directory permanently; if unsure, move the directory elsewhere instead.

In the example below, there is an existing miniconda in the home directory, If you are confident, you can run rm -rf to remove permanently the conda environment. Note that when you run rm -rf on a directory, all files and sub-folders will be removed permanently.
```
  user2022@ubuntu:~$ pwd
  /home/user2022@ubuntu/
  user2022@ubuntu:~$ ls
  miniconda
  user2022@ubuntu:~$ rm -rf miniconda   
```
If you are unsure about removing the conda environment, you may want to keep it. One of the ways, for example, is to use the command mv to rename it.
```
  user2022@ubuntu:~$ pwd
  /home/user2022@ubuntu/
  user2022@ubuntu:~$ ls
  miniconda
  user2022@ubuntu:~$ mv miniconda miniconda.old    
```

3. No ipython and jupyter installed in your System Python

Ideally, there should be no ipython and jupyter in your system Python. Running the command ipython and jupyter should give you error messages that both of them can’t be found.

  user2022@ubuntu:~$ ipython

  Command 'ipython' not found, did you mean:

    command 'ipython3' from deb ipython3 (7.13.0-1)
    command 'bpython' from deb bpython (0.18-3)

  Try: sudo apt install <deb name>

  user2022@ubuntu:~$ jupyter

  Command 'jupyter' not found, but can be installed with:

  sudo snap install jupyter       # version 1.0.0, or
  sudo apt  install jupyter-core  # version 4.6.3-3

  See 'snap info jupyter' for additional versions.

If they are already installed, you can remove them by running pip uninstall ipython and pip uninstall jupyter. For example,

  user2022@ubuntu:~$ sudo pip uninstall ipython
  Found existing installation: ipython 8.4.0
  Uninstalling ipython-8.4.0:
    Would remove:
      /home/user2022/.local/bin/ipython3
      /home/user2022/.local/lib/python3.8/site-packages/IPython/*
      /home/user2022/.local/lib/python3.8/site-packages/ipython-8.4.0.dist-info/*
      /home/user2022/.local/share/man/man1/ipython.1
  Proceed (y/n)? y
    Successfully uninstalled ipython-8.4.0

If you are unable to successfully uninstall them, you might want to remove them using rm -rf.

For example, if the ipython in your system Python stubbornly refused to be uninstalled, you can find out where are its files stored, and then remove them permanently.
```
  user2022@ubuntu:~$ which ipython
  /home/user2022/.local/bin/ipython
  user2022@ubuntu:~$ rm -rf /home/user2022/.local/bin/ipython
  user2022@ubuntu:~$ which ipython
  user2022@ubuntu:~$ 
```

Note that all files and sub-folders in the directory will be removed permanently when you run rm -rf.

4. Download Mambaforge

Create a new directory and name it anything that you like. Below we named it downloads. Change directory to downloads by typing cd downloads.

  user2022@ubuntu:~$ mkdir downloads
  user2022@ubuntu:~$ cd downloads
  user2022@ubuntu:~/downloads$ pwd
  /home/user2022@ubuntu/downloads 

Click on the following link to visit Conda-forge’s Github page: https://github.com/conda-forge/miniforge
Scroll down the page until you see Mambaforge.
Select the relevant Mambaforge installer for your OS and Architecture. Mouse over the download link to Copy the link address
Type the command wget, then paste the download link you have copied from the previous step. This will download the installer, which is a shell script, from the copied link. Once successfully downloaded, the shell script ending with .sh should be in the folder.
```
  user2022@ubuntu:~/downloads$ wget https://github.com/.../Mambaforge-Linux-x86_64.sh
  user2022@ubuntu:~/downloads$ls
  Mambaforge-Linux-x86_64.sh    
```
What's inside this Mambaforge shell script?
To look at what is inside this downloaded .sh file, run the following command:
less Mambaforge-Linux-x86_64.sh
Note that the filename may vary, depending on which file you'd downloaded. You will see the commands that shell will run, line by line, to install Mambaforge.
Press the letter q to quit reading.

5. Install Mamba

Let’s install Mambaforge by running the shell script using bash.

 user2022@ubuntu:~/downloads$ bash Mambaforge-Linux-x86_64.sh

This will start the installation process. You should see the following messages in your teminal.
```
  Welcome to Mambaforge 4.12.0-0

  In order to continue the installation process, please review the license
  agreement.
  Please, press ENTER to continue
  >>> 
```
Type yes to accept the license terms.
```
  Do you accept the license terms? [yes|no]
  [no] >>> yes
```
Press Enter to confirm the location of the installation. Note that by default Mambaforge will be installed in your home directory. In the example below, that location is home/user2022@ubuntu/mambaforge.
```
  Mambaforge will now be installed into this location:
  /home/user2022/mambaforge

    - Press ENTER to confirm the location
    - Press CTRL-C to abort the installation
    - Or specify a different location below

  [/home/user2022/mambaforge] >>> 
```
Type yes make the installer initialize Mambaforge. Next time when you open your terminal, the python environment will be set up automatically for you.
```
  Transaction finished
  installation finished.
  Do you wish the installer to initialize Mambaforge
  by running conda init? [yes|no]
  [no] >>> yes
```
Once conda init is done, it will ask you to close and restart your terminal. If the installation is successful, you will see the word (base) appearing in front of your shell prompt, upon restarting the terminal.
Notice that if you type the command which python, it will now show a Python that is different from the system Python from Step 1.
```
  (base) user2022@ubuntu:~$ 
  (base) user2022@ubuntu:~$ which python
  /home/user2022/mambaforge/bin/python
```
How do I get out of this Python environment?
Use the following two commands:
- mamba deactivate: to get out of the environment
- mamba activate: to get back into the environment

6. Install Pytorch and Jupyter

Try using 'mamba install' as much as possible, rather than 'conda install' or 'pip install'

Now that our Python environment has been set up, we can start installing softwares that we need for doing deep-learning, such as jupyterlab, ipython, pytorch …etc.

It’s especially important to have good Pytorch and Jupyter installation on your machine. The recommendation is to use mamba install when installing both of them, as it will give you a more optimized installation in a faster, fuss-free manner.
Installing Pytorch: on Pytorch’s Install Pytorch page, select your preference. The example below installs a cpu version of Pytorch, and assumes that there is no GPU on your laptop. If you have GPU on your machine, select the relevant CUDA version instead. Notice that it is however using conda install, which we will replace with mamba install.
```
 (base) user2022@ubuntu:~$ mamba install pytorch torchvision torchaudio cpuonly -c pytorch
```
Installing Jupyter Lab: on Jupyterlab’s Installation page, the installation command is mamba install -c conda-forge jupyterlab. Since the default channel for mamba is already conda-forge, we can simply ran
```
  (base) user2022@ubuntu:~$ mamba install jupyterlab
```

That’s all. Have fun!

  (base) user2022@ubuntu:~$ jupyter lab

P.S. After running rm -rf a few times and removing permanently files you shouldn’t, you’ll get used to it. I did.