Eclipse Deeplearning4j, or DL4J, is an open-source distributed deep-learning framework for the JVM, used to build, train, and deploy neural networks from Java and Scala applications. It is not a single pretrained model or a hosted generation service; it is a composable ecosystem covering tensor computation, automatic differentiation, data pipelines, model import, neural-network training, and inference.
The ecosystem is intended for developers and engineering teams bringing machine learning into JVM software, with dependencies added through Maven or Gradle and compute backends configured for CPUs or CUDA-capable GPUs. Use requires an existing Java or Scala development environment and compatible dependencies; imported models must follow supported Keras, TensorFlow, or ONNX paths, while hardware acceleration depends on the corresponding native runtime configuration.