A char-level transformer trained from scratch on an $8 ESP32-S3

An SLM trained on $8 ESP32-S3

Carloscodix open-sourced qapla, a char-level transformer trained entirely on an $8 ESP32-S3 chip. The chip runs the full training loop, not just inference, with backprop hand-written in C. The repo doesn't disclose parameter count, dataset size, or training time, but the code is public with 6 commits. Treat this as an extreme engineering demo rather than a practical tool—fitting training onto a microcontroller is the wild part.

Why it matters: Running a full training loop on an $8 microcontroller is a hardcore engineering demo — H and K both hit. But the repo has only 6 commits, no param count, dataset size, or training time disclosed, so practical value is limited and R misses. Scored 72 at the featured threshold, ...