WebGPU back end in llama.cpp/ggml

3 points | by tosh 6 hours ago