Tiny language model runs at 60k tokens per second on a $250 FPGA

Original: Show HN: A tiny LLM running at 21,000 tok/s on a $250 FPGA (Live Demo)

Source: mikeayles.com ↗