Setup Fastai v1 on Paperspace

Machine

• Create a Fast.ai machine from public templates w/ P4000 and public IP
• If you are not approved for using the P4000, request it. It deserves the waiting time (powerful than the default one with the same costs).

Connect to the machine

• source deactivate fastai
• pip install virtualenv
• virtualenv fastai_v1
• source activate fastai_v1
• pip install torch_nightly -f https://download.pytorch.org/whl/nightly/cu92/torch_nightly.html
• pip install fastai

Install other dependencies prior to get stuck when using the Notebook

• pip install bcolz
• pip install opencv-python
• pip install seaborn
• pip install graphviz
• pip install sklearn_pandas
• pip install isoweek
• pip install pandas_summary
• pip install torchtext
• pip install feather-format
• pip install jupyter_contrib_nbextensions
• pip install plotnine
• pip install docrepr
• pip install awscli
• pip install kaggle-cli
• pip install pdpbox

Clone fastai_v1 from GitHub

• git clone https://github.com/fastai/fastai repo-fastai-v1/
• cd repo-fastai-v1/

Reconfigure jupyter configuration file

Open the jupyter config

• nano ~/.jupyter/jupyter_notebook_config.py

Edit lines

• c.NotebookApp.allow_origin = '' to c.NotebookApp.allow_origin = '*'
• c.NotebookApp.ip = 'localhost' to c.NotebookApp.ip = '0.0.0.0'
• [Probably you wouldn't find this config currently to modify, don't worry] c.NotebookApp.allow_remote_access = False to c.NotebookApp.allow_remote_access = True

Then you can run the Jupyter Notebook on the fastai v1 repo directory

• jupyter notebook

Access your Jupyter and run lesson1.ipynb on examples folder

• Try to execute the first cell, if it executes you're done

Success!