I built a native Vulkan training backend for 143 modern Transformer architectures — no CUDA or PyTorch required

A developer has created a native Rust and Vulkan training backend, initially for their Hierarchos architecture, which now supports training and inference for 143 canonical Transformer architectures. This project, which does not require CUDA or PyTorch, also includes additional Hugging Face model_type aliases, expanding the advertised registry to 224 model-type spellings. The developer is seeking technical criticism, architecture compatibility reports, benchmarks on various GPUs, and contributors to stress-test the backend.