by: rao-v

This is a really neat use of the per-layer embedding trick. It's also worth noting that there viable TTS models that are ~20-30M param, so it might mean you can have a ESP32 with no network access read stuff out to you in near real time!

7/26/2026, 12:02:44 AM

by: chrishynes

Why can't this scale to run much larger models on CPU backed by flash with good access patterns?

7/26/2026, 12:04:58 AM

by: althea_tx

This is a really cool project. Thanks for sharing!

7/25/2026, 11:50:48 PM

by: cr125rider

9.7 tokens/sec actually seems like a lot! That’s fun!

7/25/2026, 11:39:16 PM

by:

7/25/2026, 6:59:50 PM