Comment by tatef

Hypura reads tensor weights from the GGUF file on NVMe into RAM/GPU memory pools, then compute happens entirely in RAM/GPU.

There is no writing to SSDs on inference with this architecture.