ParentFull threadianpurton·ONNX doesn't support the same level of quantization as GGML.So basically GGML will run on hardware with less memory.View on HN