ConvNetJS

Demos of neural networks trained directly in JavaScript for classification, regression and more.

Artifact by Andrej Karpathy

IndexedMIT★ 11,216EducationNo sign-up

Open tool ↗

Write a review

Use it

Opens cs.stanford.edu in a new tab. Nothing to install.

Open ConvNetJS Source code

You provide: Demo choice and training settings
You get: Live training curves and predictions
Account: No sign-up needed
AI: No AI: a plain tool
Your input: Not stated: check the tool's privacy notes before using private data
Price: Free to use
Made with: Open-source web app
Source: Open source
Link check: Answered on Oct 31, 2026

About

ConvNetJS is an open-source web app in the education category (deep learning). It opens in your browser, so there is nothing to install. You provide demo choice and training settings and get live training curves and predictions. It needs no account. It is a plain tool: no AI model is involved. The source code is public on GitHub (karpathy/convnetjs).

Category: Education · deep learning

Reviews

Write a review

No reviews yet. Used it? Sign in with an established GitHub account to rate it. How reviews work

More education tools

See all education tools →

Signals

Stars: 11,216

Details

Auth: No auth
License: MIT
Repository: github.com/karpathy/convnetjs
Website: cs.stanford.edu/people/karpathy/convnetjs/
First seen: Oct 31, 2026
Last checked: Oct 31, 2026
Source: GitHub

Cite this page

ConvNetJS: free online tool. Agent Mart. https://agentmart.dev/artifacts/convnetjs (last updated 2026-10-31)