b8240

The llama.cpp project, a leading C++ implementation for running Meta's Llama models efficiently on consumer hardware, has pushed a significant update with commit b8240. The core of this release is the addition of support for the Exponential Linear Unit (ELU) activation function within its Vulkan compute backend. ELU is a non-linear function used in some neural network architectures to help models learn more complex patterns. By implementing this op, llama.cpp now allows models that utilize ELU layers to run natively and efficiently on GPUs that support the Vulkan API, a cross-platform graphics and compute standard.

This update is part of the project's ongoing effort to maximize hardware compatibility and performance. Vulkan support is crucial for running models on AMD GPUs and integrated graphics from Intel, providing an alternative to the dominant CUDA framework which is exclusive to NVIDIA hardware. The commit also includes routine code cleanup, fixing formatting and whitespace issues in the Vulkan implementation files. Alongside the Vulkan update, the release includes pre-built binaries for a vast array of platforms including macOS (Apple Silicon and Intel), Linux (with CPU, Vulkan, and ROCm backends), Windows (with CPU, CUDA, Vulkan, SYCL, and HIP), and openEuler, demonstrating the project's commitment to broad accessibility.