Backprop Documentation

Backprop is a cloud ML experiment platform. Submit training scripts via CLI or web dashboard, and Backprop provisions GPUs, runs the job, streams logs, and collects results.

Ready to get started?

Install the Backprop CLI
Terminal window
```
npm install -g @backprophq/cli
```
Not using npm? You can also use pnpm add -g @backprophq/cli or yarn global add @backprophq/cli.
Log in to your account
Terminal window
```
bp auth login
```
Run your first experiment
Terminal window
```
bp run --gpu l4 train.py
```

Explore the docs

Quickstart

Train a language model in 3 minutes. Walk through a complete experiment from start to finish.

Get started →

Getting Started

Install the CLI and configure your environment to start running experiments.

Read the guide →

Metrics

Log training metrics and view real-time loss curves in the Backprop console.

Learn more →

Reference

Complete reference documentation for the Backprop CLI and API.

View reference →