djl

Why we included this project

For Java teams tired of shuttling between Python and JVM code, DJL is a practical way to keep machine learning inside the Java world. It is a high-level framework that behaves like any other Java library, so you can train and run inference from the IDE and build tooling you already use. The engine-agnostic design means your code targets one consistent API while the framework routes work to PyTorch, TensorFlow, MXNet, or ONNX Runtime underneath; switching engines becomes a configuration change rather than a rewrite. You also get a model zoo of ready-to-use pre-trained architectures and automatic CPU/GPU selection. For developers who know Java but not deep learning, this is a rare on-ramp in a field dominated by Python.