Java implementation of micrograd created by Andrej Karpathy from his YouTube lecture: "The spelled-out intro to neural networks and backpropagation: building micrograd"
- Origional Video
- The cooresponding GitHub Python Jupiter Notebooks
Note that there are commented out sections of code in the Value class "demo(*) functions" that follow along to his video to test the math. Graph that opens showing the loss over time when running the demo.
The code could use a little cleanup, but a good starting point to build simple neural networks in Java. Java needs overloaded operators! +1 ๐