ConvNetJS
A JavaScript library for training neural networks, including convolutional ones, in a web browser.
- Stars
- 11.2k
- Forks
- 2.1k
- Licence
- MIT
- Language
- JavaScript
- Last update
- 3.8 years ago
- Type
- Library
No commits in over a year — check for forks or alternatives before adopting.
What it is
ConvNetJS is a JavaScript library for building and training deep learning models, including convolutional neural networks, directly in a web browser. It is described as a way to train ordinary and convolutional neural networks without installing heavy machine learning software. People use it mainly for learning and experimenting with how these networks work.
- Train a small convolutional network on image-like data inside a web page
- Explore how layers, filters, and training settings change results in a browser
- Run a neural network demo without a server or local Python setup
Great for
- Developers who already work in JavaScript and want to try neural networks in the browser
- Students learning how convolutional neural networks are trained
- Teachers or presenters who want a visual, in-browser demo of training
- Small experiments where installing a Python ML stack is not worth the effort
Not ideal for
- Large-scale training jobs, since the README does not claim performance suitable for big datasets or GPUs
- People who need a actively maintained production ML toolkit, because the last push was in January 2023 and the latest release is from 2014
- Users who want a Python-based ecosystem with pretrained model hubs, which this project does not describe
How to get it
- 1
Clone and build from source
You'll need: A web browser · Git, to clone the repository
Clone the repository
git clone https://github.com/karpathy/convnetjs - 2
Run it
- Clone the repository with git
- Open the included HTML demo files in a web browser
- Edit a demo file or script to change the network or training data
- Reload the page to see the training results
- 3
Get the most out of it
- Read the included demo pages to see working examples before writing your own
- Start with small networks and datasets so training finishes quickly in the browser
- Use the browser developer console to inspect values during training
Downloads
Strengths
- Runs in a browser with no server-side setup, according to its description
- Released under the MIT license, which allows broad reuse
- Widely starred and known as an educational deep learning project
Trade-offs
- The latest release is from 2014 and the repository's last push was in 2023, so it may not reflect current deep learning practice
- The README provided is truncated, so the full feature list and API could not be verified here
- Browser-based training is limited in speed and scale compared with dedicated ML frameworks
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| convnetjs (this) | 11.2k | Library | intermediate | 3.8 years ago | Developers who already work in JavaScript and want to try neural networks in the browser |
| brain.js | 14.9k | Library | beginner | 2.1 years ago | JavaScript developers who want to try neural networks without learning a new language |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.