wladimiravila··on HN: I ran a 56M parameter LLM across three ESP32-S3 boards via ESP-NOWhttps://github.com/wladimiravila/esp32s3-distributed-ai