An open API service indexing awesome lists of open source software.

https://github.com/jalberty2018/run-pytorch-cuda-develop

Compile environment for Pytorch with CUDA
https://github.com/jalberty2018/run-pytorch-cuda-develop

cloud code-server compiler cuda cuda-toolkit docker-image flash-attn jupyterlab python python3 pytorch sage-attention

Last synced: 23 days ago
JSON representation

Compile environment for Pytorch with CUDA

Awesome Lists containing this project

README

          

[![Docker Image Version](https://img.shields.io/docker/v/ls250824/run-pytorch-cuda-develop)](https://hub.docker.com/r/ls250824/run-pytorch-cuda-develop)

# run-pytorch-cuda-develop

## Synopsis

A streamlined setup for developing on high-performance hardware.

- Authentication credentials set via secrets for:
- **Code server** authentication
- **Jupyterlab** authentication
- **Hugging Face** and **CivitAI** tokens for model access (mandatory).

Ensure that the required environment variables and secrets are correctly set before running the pod.
See below for options.

## Hardware provisioning

- [Runpod.io](https://runpod.io/)
- GPU

## Software Repositories

### Core

- [Code Server](https://github.com/coder/code-server)
- [Jupyterlab](https://jupyter.org)
- [Jupyter Server](https://jupyter-server.readthedocs.io/en/latest/index.html)
- [HuggingFace cli](https://huggingface.co/docs/huggingface_hub/guides/cli)
- [Nvidia CUDA](https://hub.docker.com/r/nvidia/cuda/tags?name=12)
- [Pytorch.org](https://pytorch.org)
- [Triton](https://triton-lang.org/main/index.html)

## Documentation

- [SM architecture](https://developer.nvidia.com/cuda-gpus)
- [Wikipedia](https://en.wikipedia.org/wiki/CUDA)

## Available Images

### Pytorch 2.8 CUDA 12.9

```bash
docker pull ls250824/run-pytorch-cuda-develop:09102025
```

### Pytorch 2.9 CUDA 13.0

```bash
docker pull ls250824/run-pytorch-cuda-develop:05112025
```

### Pytorch 2.9 CUDA 12.8

```bash
docker pull ls250824/run-pytorch-cuda-develop:08112025
```

### Pytorch 2.10 CUDA 12.8.1

```bash
docker pull ls250824/run-pytorch-cuda-develop:10032026
```

## Settings

### Services

| Service | Port |
|-----------------|---------------|
| **Code Server** | `9000` (HTTP) |
| **Jupyterlab** | `8888` (HTTP) |
| **SSH/SCP** | `22` (TCP) |
| **Gradio** | `7860` (HTTP) |

### Authentication Tokens

| Token | Environment Variable |
|--------------|----------------------|
| Civitai | `CIVITAI_TOKEN` |
| Huggingface | `HF_TOKEN` |
| Code Server | `PASSWORD` |
| Jupyterlab | `JUPYTERLAB_PASS` |

### Sources

| Variable | Description |
|------------------|----------------------------------|
| `SOURCE[1-50]` | source download links (compressed or plain) |

## Building the Docker Image

### `build-docker.py` script options

| Option | Description | Default |
|----------------|-----------------------------------------------------|------------------------|
| `--username` | Docker Hub username | Current user |
| `--tag` | Tag to use for the image | Today's date |
| `--latest` | If specified, also tags and pushes as `latest` | Not enabled by default |

### Build & push Command

Run the following command to clone the repository and build the image:

```bash
git clone https://github.com/jalberty2018/run-pytorch-cuda-develop.git
cp run-pytorch-cuda-develop/build-docker.py ..

export DOCKER_BUILDKIT=1
export COMPOSE_DOCKER_CLI_BUILD=1

python build-docker.py \
--username= \
--tag= \
run-pytorch-cuda-develop
```

Note: If you want to push the image with the latest tag, add the --latest flag at the end.