On: Tiny Language Models Come of Age

[woadwarrior01]

Running (quantised) LLMs (encoder-decoder and decoder only transformers) on phones isn't hard. Running CNNs and encoder-only transformers, the kind used in vision on phones is even easier.