JAX-NSL
PublicComprehensive JAX implementation of neural networks and scientific computing. Features distributed training, physics-informed networks, custom autodiff, and advanced optimization. Production-ready code with numerical stability, multi-device parallelism, and research-grade implementations.