This is a replica of the lab environment used in the Duke Machine Learning Virtual Summer School. The environmnet was adapted to use the latest PyTorch image + installs JupyterLab on it to allow easy work.
The workspace directory used by JupyterLab is mapped to ./workspace in this repo, so you have easy access to all your files. Note that some of the exercises were slightly adaptad to allow datasets download and support to torchtext.legacy
Clone the repository: git clone https://github.com/OneLastTry/pytorch-duke.git and then execute from the main directory one of the commands below to perform the desired action:
- build the image:
docker-compose build - start container:
docker-compose up -d - stop container:
docker-compose stop - reset container:
docker-compose down
With the container running you can access JupyterLab using the link below: http://localhost:5000/lab The password is admin, you can change that on the Dockerfile.
├── Dockerfile
├── LICENSE
├── README.md
├── docker-compose.yml
├── pytorch-duke.code-workspace
└── workspace
├── MLvSS_2021_Day2_Coding
│ ├── Figures
│ │ ├── chapel.jpg
│ │ ├── conv_filters.png
│ │ ├── maxpool.png
│ │ ├── mnist_cnn_ex.png
│ │ ├── mnist_digital.png
│ │ └── mnist_matmul.png
│ ├── Student_MLvSS2021_ImageAnalysis-clear.ipynb
│ ├── Student_MLvSS2021_ImageAnalysis-complete.ipynb
│ └── Student_MLvSS2021_ImageAnalysis.ipynb
├── Student_Pytorch_NLP_tutorial-clear.ipynb
├── Student_Pytorch_NLP_tutorial-complete.ipynb
├── Student_Pytorch_NLP_tutorial.ipynb
└── intro_pytorch
├── figures
│ ├── mnist_examples.png
│ └── mnist_matmul.png
├── intro_pytorch-clean.ipynb
├── intro_pytorch.ipynb
└── intro_pytorch_complete.ipynb
5 directories, 22 files