CoViz is a visual deep learning framework for the web that allows users to easily build and train neural networks. The framework supports basic neural network components such as dense layers of neurons, ReLU activations, Softmax, mean squared error loss, and cross-entropy error loss. With these building blocks, users can create and experiment with a wide range of neural network architectures.
CoViz is one of the first projects to use WebGPU 🔥 to implement a fully differentiable programming engine. The project is still in its early stages and constantly evolving.
You can try out CoViz by visiting the live demo at https://covizdemo.vercel.app/.
Note: CoViz requires WebGPU to be enabled in your browser!
CoViz's user interface features a node editor, powered by ReactFlow, that allows users to easily build a Neural Network computation graph. Users can select from a range of regression and classification tasks and datasets, and train their network to learn patterns in the data.
To run CoViz locally on your machine, follow these steps:
-
Make sure you have npm installed on your machine. If not, you can download it from the official website.
-
Clone the repository.
-
In your terminal, navigate to the cloned repository and run
npm installto install all the dependencies. -
Run
npm startto start the application. -
Open a browser (e.g., Chrome) and go to http://localhost:3000/. Any changes you make to the code will automatically be reflected in the browser window.
Note: CoViz requires WebGPU to be enabled in your browser. In Chrome, you can enable WebGPU by going to chrome://flags/ and searching for "WebGPU".