rpdaiml

Nice work, that throughput is wild.

How are you handling the on device speech pipeline, especially around model size, latency, and accuracy tradeoffs on consumer hardware?

This is a nice idea. A tiny implementation can be way more useful for learning than yet another wrapper around a big model, especially if it keeps the training loop and inference path small enough to read end to end.

curious and wondering its porting and usage on edge devices?