Llama3.java: featuring .GGUF file format support, Q8_0 and Q4_0 quantizations, fast matrix/vector multiplication routines using Java's Vector API; served by a simple CLI with a --chat mode to interact with the Llama 3 models.