Before starting, please complete the system-level GPU setup
See: SYSTEM_GPU_SETUP.md
A reproducible Docker-based pipeline for running Machine/Deep learning exps, with GPU support. This repository provides pre-configured Docker images, environment files, and scripts for:
-
- Setting up GPU-enabled containers
-
- Running training and inference
-
- Managing environments reproducible
A complete, reproducible setup for deep learning experiments using Docker, Conda, and JupyterLab with GPU support.
git clone https://github.com/<your-username>/ModelApp.git
cd ModelApp
2️. Build the Docker Image
sudo docker build -t modelapp-container .
3️. Run the Container (GPU + Jupyter)
docker run --gpus all -it --rm \
-p 8888:8888 \
-v $(pwd):/workspace \
modelapp-container
-------------------------------------------------------------------------------------------------
📎 Flags:
--gpus all → enables GPU access inside container
-p 8888:8888 → maps JupyterLab port
-v $(pwd):/workspace → mounts project folder
Once running, open the browser link that appears (e.g. http://127.0.0.1:8888/lab?...).
---------------------------------------------------------------------------------------------------
4️. Verify GPU Access
Inside the container:
nvidia-smi
python -c "import torch; print(torch.cuda.is_available())"
**You should see your GPU and True.**
Development Inside Container
**To open a terminal inside container:**
docker exec -it <container_id> /bin/bash
##To run training, or any other programs, you can start implementing your training/testing, etc.
i.e:
python train.py
------------------------------------------------------------------------------------------------------------------
Why This Setup?
Reproducible ML experiments
Full GPU acceleration
No “it works on my machine” issues
Portable across OS/machines
Seamless JupyterLab workflowcd /workspace/src
---------------------------------------------------------------------------------------------------------------------
Contributing
Fork this repo
Create a new branch (feature/my-feature)
Commit your changes
Submit a pull request